Skip to main content

Full text of "Dimensions of conflicting expectations and the leader behavior of principals"

See other formats


For  Reference 


NOT  TO  BE  TAKEN  FROM  THIS  ROOM 


©x  mm 


University  of  Alberta 
Printing  Department 


] 


Digitized  by  the  Internet  Archive 
in  2018  with  funding  from 
University  of  Alberta  Libraries 


https://archive.org/details/dimensionsofconfOOmikl 


X'Wfc 
i$,  V2>-  ^ 


THE  UNIVERSITY  OP  ALBERTA 


DIMENSIONS  OP  CONFLICTING  EXPECTATIONS  AND  THE 
LEADER  BEHAVIOR  OP  PRINCIPALS 

by 

ERWIN  MIKLOS 


A  THESIS 

SUBMITTED  TO  THE  FACULTY  OF  GRADUATE  STUDIES 
IN  PARTIAL  FULFILMENT  OF  THE  REQUIREMENTS  FOB  THE  DEGREE 
OF  DOCTOR  OF  PHILOSOPHY 


DEPARTMENT  OF  EDUCATIONAL  ADMINISTRATION 


EDMONTON,  ALBERTA 
MAY,  1965 


ABSTRACT 


The  study  reported  in  this  thesis  explored  the  utility  of 
Guttman  scale  analysis  procedures  for  quantifying  and  for  analyzing  the 
structure  of  the  expectations  which  teachers  hold  for  the  "behavior  of 
principals.  The  results  of  the  scale  analysis  were  used  for  testing 
hypothesized  relationships  among  the  degree  of  consensus  on  expec¬ 
tations  within  a  school,  the  degree  of  teacher- principal  agreement  on 
expectations  for  the  role  of  the  principal,  and  the  teachers'  de¬ 
scriptions  of  the  leader  "behavior  of  principals.  Leader  ambivalence 
was  defined  as  the  intensity  with  which  principals  held  expectations 
for  their  role,  and  the  relationship  of  this  variable  to  leader  behavior 
descriptions  was  also  explored.  Data  for  the  study  were  obtained  from 
fifty-six  principals  and  765  teachers  in  fifty-six  non-urban  centralized 
schools . 

Items  were  constructed  for  each  of  authority,  personal,  status, 
and  means-ends  categories  and  the  responses  were  tested  for  unidi¬ 
mensionality  using  scale  analysis  techniques.  None  of  these  sets  of 
items  satisfied  the  criteria  for  the  existence  of  a  scale  or  even  a  quasi 
scale.  However,  approximately  scalable  subareas  were  identified  and  were 
used  for  quantifying  the  expectations  held  by  teachers  and  principals  for 
the  role  of  principals.  From  these  were  derived  indices  of  Agreement 
and  Consensus  while  scale  analysis  of  questionnaire  responses  was  used 
for  deriving  an  index  of  principal  Ambivalence. 

No  significant  linear  or  curvilinear  relationships  were  found  to 


iii 


.  ' 


« 


. 


■ 

«  v  '  ’  '  '•  ; 

' 

- 


exist  between  Ambivalence  and  teachers’  descriptions  of  the  behavior  of 
principals  on  the  Leader  Behavior  Description  Questionnaire.  However, 
when  schools  were  classed  as  being  either  high  or  low  in  Consensus,  it 
was  found  that  Ambivalence  correlated  positively  with  Structure  in  low 
Consensus  Schools  while  the  same  correlation  was  negative  in  high  Con¬ 
sensus  schools.  The  difference  between  the  two  correlation  coefficients 
was  statistically  significant  and  was  interpreted  as  suggesting  that 
Ambivalence  is  related  to  what  might  be  considered  desirable  leader 
behavior  only  if  it  is  appropriate  to  the  situation. 

The  intensity  with  which  principals  held  their  self-expectations 
was  not  related  to  either  Consensus  or  Agreement,  but  these  variables 
were  related  to  style  of  leader  behavior  as  defined  by  Initiating 
Structure  and  Consideration  scores.  Leader  behavior  characterized  by 
scores  high  in  both  Structure  and  Consideration  was  found  to  be 
associated  with  significantly  higher  Agreement  scores  than  was  any  other 
style  of  leader  behavior.  The  same  trend  was  evident  when  leader  behavior 
was  related  to  Consensus  but  the  results  were  not  statistically  signifi¬ 
cant.  For  principals  who  were  high  in  Structure,  both  Agreement  and 
Consensus  correlated  positively  with  the  average  length  of  time  that 
principals  had  been  in  the  present  school  with  the  present  teachers 
while  the  same  relationship  was  negative  for  principals  classed  as  low 
in  Structure.  In  the  case  of  Agreement  the  difference  between  the  two 
correlations  was  highly  significant. 


iv 


. 


• 

• 

- 

. 

- 

' 

.  ' 

. 

. 


ACKNOWLEDGEMENTS 


The  writer  wishes  to  express  his  thanks  to  the  supervisor  of  the 
thesis,  Dr.  J.H.M.  Andrews,  for  his  valuable  guidance  at  various  stages 
of  the  study.  Thanks  are  extended  also  to  the  other  committee  members, 
Dr.  L.W.  Downey  and  Dr.  R.L.  James,  for  their  assistance  during  the 
investigation  and  during  the  writer's  program  of  studies  in  general. 

Appreciation  is  expressed  to  the  many  teachers  and  principals 
whose  cooperation  made  possible  the  collection  of  the  necessary  data. 

The  courtesy  with  which  the  writer  was  received  by  the  principals  at  a 
busy  time  of  year  is  evidence  of  their  concern  for  research  in  education 
the  interest  of  teachers  in  such  studies  w as  indicated  by  the  excellent 
response  in  the  form  of  completed  questionnaires. 

Mention  must  also  be  made  of  the  fellow  students,  other  friends, 
and  family  without  whose  assistance,  encouragement,  and  consideration 
study  and  research  would  have  been  much  more  arduous. 

Finally,  financial  assistance  in  the  form  of  an  Alberta  Teachers* 
Association  Fellowship  in  Education  and  a  Province  of  Alberta  Graduate 
Scholarship  which  enabled  the  writer  to  pursue  graduate  study  is  grate¬ 
fully  acknowledged. 


v 


.  »  .  - 


. 

■> 


' 


TABLE  OF  CONTENTS 


CHAPTER  PAGE 

I.  INTRODUCTION  .  1 

Purposes  of  the  Study  ...  . 

Need  for  the  Study .  4 

Limitations  of  the  Study .  8 

Organization  of  the  Thesis .  10 

II.  THEORETICAL  BASES  OF  THE  STUDY .  13 

A  Theoretical  Model  .  13 

Institutional  and  Individual  Conflict .  17 

Role  Conflict .  17 

Dimensions  of  Conflicting  Expectations  ....  19 

Value  Contradictions .  20 

Relation  to  Role  Conflict .  21 

Leader  .Ambivalence .  23 

Attitude  Intensity  .  24 

Generalized  Intensity .  26 

Ambivalence .  27 

Leader  Behavior  .  28 

Research  Hypotheses  .  30 

Definitions  of  Terms  .  30 

Scale  Analysis .  31 

Hypotheses .  32 

III.  RELATED  RESEARCH  .  36 

Research  on  Role  Conflict .  36 

vi 


l'  . .  ■ 


-  ;  /  ;• 


. 


CHAPTER 


PAGE 


Non -Education  Role  Conflict  Studies  .  36 

Research  in  Education .  42 

Intensity  Function  and  .Ambivalence  .  . .  45 

Attitude  Intensity  .  45 

Ambivalence . 46 

Some  Studies  of  Leader  Behavior .  48 

IV.  SCALE  ANALYSIS:  THEORY  AND  METHODOLOGY  .  55 

The  Theory  of  Scale  Analysis .  55 

The  Universe  of  Attributes .  55 

Example  of  a  Dichotomous  Scale .  58 

Definition  of  a  Scale .  59 

Measurement  of  Error .  60 

Quasi  Scales .  62 

The  Methodology  of  Scale  Analysis .  63 

The  Scalogram  Board .  63 

Initial  Arrangement  .  64 

Ranking  of  Respondents .  65 

Combining  Categories  .  66 

Final  Arrangement .  67 

Using  one  Board .  67 

Testing  for  Scalability  .  67 

Scoring  the  Questionnaires  .  68 

Discussion .  68 

The  Utility  of  Scale  Analysis .  69 

vii 


;•  •• 


(  ■  * 


f  i  .  •. to  i  li 

•  -  *  .  •- . 

.  .  .  . 

..... 

•  t  • 

’  *  •  • 

♦  ••••*  o 

...  v.'  .  .  I 


CHAPTER 


PAGE 


Defining  a  Problem .  69 

Implications  of  Scores  .  70 

Prediction .  70 

Some  Problems  in  Scale  Analysis .  71 

Sampling  of  People  . .  71 

Sampling  of  Items .  72 

Test-Retest  Reliability  .  74 

Summary .  74 

V.  SCALE  ANALYSIS:  CRITICISMS,  DEVELOPMENTS ,  AND 

APPLICATIONS  .  76 

Some  Criticisms  of  Scale  Analysis  .  . .  76 

Universe  of  Attributes .  78 

Item  Construction  and  Arrangement  .  79 

Interviewer  Effect  .  82 

Coefficient  of  Reproducibility  .  82 

Some  Developments .  84 

Applications  of  Scale  Analysis  .  86 

Checking  Existing  Scales  .  87 

Data,  from  Interviews .  88 

Church  Orthodoxy  .  89 

Urban  Structures .  89 

Student  Attitudes .  90 

Voting  Behavior .  91 

Delinquent  Behavior  . 

Concept  Development  .  93 


viii 


: 


■ 


CHAPTER 


PAGE 


Aggressive  Fantasy  .  93 

Object  Scales  .  94 

Scale  Analysis  in  the  Study  of  Role  Conflict  ...  94 

Conclusion .  97 

VI.  COLLECTION  OF  DATA .  101 

Teachers'  and  Principals’  Questionnaires  .  101 

Selection  of  Items  .....  .  101 

Pretests,  Validity,  and  Revisions .  102 

LBDQ, .  105 

Principal  Ambivalence  .  105 

Selecting  the  Sample  and  Obtaining  Data .  105 

The  Sample .  105 

Visiting  Schools  .  106 

Response .  107 

Characteristics  of  Respondents  .  109 

VII.  SCALOGRAM  ANALYSIS  OF  DATA .  114 

General  Procedures  .  114 

Approximately  Scalable  Areas  .  116 

Status  Dimension .  117 

Authority  Dimension  .  119 

Personal  Dimension  .  123 

Means -Ends  Dimension  .  126 

Scoring .  127 

Scale  Analysis  of  Principals’  Responses  .  129 


ix 


..  ' J' 


! 


CHAPTER  PAGE 

Ambivalence  Scores  .  131 

VIII.  TREATMENT  OP  DATA .  134 

Leader  Behavior  Scores  .  134 

Ambivalence,  Consensus,  and  Agreement  .  137 

Ambivalence  Scores  .  137 

Consensus . 138 

Teacher -Principal  Agreement  .  139 

Statistical  Techniques . .  .  142 

Differences  Between  Means  .  143 

Measures  of  Relationship .  145 

Correlation  Ratio .  145 

IX.  TESTING  THE  HYPOTHESES .  148 

Ambivalence  and  Leader  Behavior .  148 

Testing  Hypothesis  One .  148 

Discussion .  151 

Relationships  of  Structure  and  Consideration  to 
Agreement  and  Consensus  .  155 

Testing  Hypothesis  Two  .  153 

Discussion .  157 

Testing  Hypothesis  Three  .  157 

Discussion .  161 

Relationship  of  Ambivalence  to  Consensus  and 

Agreement .  161 

Testing  Hypotheses  Pour  and  Five .  161 

x 


.  V-.'H 


•  M 


. .  •-•.  i 

........on....  i...  .  .  v 

...........  •■'£'■•  /. 

............ 

. . .  •  i 


fjOl 


CHAPTER 


PAGE 


Discussion .  164 

Relationship  of  Style  of  Leader  Behavior  to 

Consensus . I65 

Hypotheses . .  .  166 

Discussion  and  Further  Analysis .  168 

Relationship  of  Style  of  Leader  Behavior  to 

Agreement .  172 

Hypotheses .  172 

Discussion .  17  6 

Style  of  Leader  Behavior,  Tenure  and  Consensus.  176 

Testing  Hypothesis  Six  .  177 

Discussion . 179 

Style  of  Leader  Behavior,  Tenure  and  Agreement.  181 

Testing  Hypothesis  Seven  .  181 

Discussion .  184 

Summary . « .  I84 

Ambivalence,  Structure,  Consideration  .  .  .  185 

Ambivalence,  Agreement,  Consensus  .  185 

Structure,  Consideration,  Agreement  ....  186 

Structure,  Consideration,  Consensus  ....  186 

Discussion .  186 

X.  SUMMARY  AND  CONCLUSIONS .  188 

Summary  of  the  Study .  188 

Conclusions  and  Implications  .  194 

Conclusions .  195 


xi 


l 


O  J' 


r  s. 


-L  . 

•»  -»  O  V. 


1<  •. .  •• 


1  1 


i  r  v .  • 


o  : 


Q 


CHAPTER 


PAGE 


Implications  for  Research .  19 6 

Implications  for  Administrators .  199 


BIBLIOGRAPHY 
APPENDIX  A 
APPENDIX  B 
APPENDIX  C 
APPENDIX  D 
APPENDIX  E 


EXPECTATIONS  FOR  PRINCIPAL  LEADER  BEHAVIOR 
LEADER  BEHAVIOR  DESCRIPTION  QUESTIONNAIRE 

PRINCIPAL'S  QUESTIONNAIRE  . 

QUESTIONNAIRE  RETURNS  BY  SCHOOLS  .  .  . 
NORMALIZED  DATA  BY  SCHOOLS . 


203 

212 

222 

225 

231 

233 


xii 


. .  .  .  •  . 

. .■ 

•  r».« 

. 

......  ' 

...... 


LIST  OF  TABLES 


TABLE  PAGE 

I.  Number  of  Questionnaires  Distributed  and  Completed 

by  Size  of  School  .  108 

II.  Selected  Characteristics  of  Respondents  .  Ill 

III.  Distribution  of  Principals  by  Experience  and  Teachers 

by  Tenure . 112 

IV.  Frequency  Distribution  of  Teachers*  Scale  Scores  ..  129 

V.  Coefficients  of  Reproducibility  for  Teachers'  and 

Principals'  Responses  to  Seven  Sets  of  Items  ....  130 

VI.  Frequency  Distribution  of  Principals'  Scale 

Scores  (n=60)  .  131 

VII.  Coefficients  of  Reproducibility  for  Principals' 

Content  and  Ambivalence  Scales  .  132 

VIII.  Frequency  Distribution  of  Principals'  Ambivalence 

Scores  (N=60)  .  133 

IX.  Frequency  Distribution  of  Raw  and  T -Scores  for 
Consideration  and  Initiating  Structure  LBDQ 

Scores  (N=56)  .  135 

X.  Frequency  Distribution  of  Raw  and  T-Scores  for 

Principals'  Ambivalence  Scores  .  138 

XI.  Frequency  Distribution  of  Raw  and  Normalized 

Consensus  Scores  .  140 

XII.  Frequency  Distribution  of  Raw  and  Normalized 

Teacher- Principal  Agreement  Scores  .  141 


xiii 


TABLE 


PAGE 


XIII.  Mean  Structure  and  Consideration  Scores,  Variances, 
and  Significance  of  Differences  for  Principals 

of  High  and  Low  .Ambivalence  .  149 

XIV.  Linear  and  Nonlinear  Correlation  Coefficients  for 
the  Relationship  Between  Ambivalence  and  Two 

Dimensions  of  Leader  Behavior  (N=56)  .  151 

XV.  P  Ratios  for  Significance  of  Linear  and  Curvi¬ 
linear  Relationships  Between  Ambivalence  and  Two 

Dimensions  of  Leader  Behavior  . . .  152 

XVI.  Mean  Agreement  Scores,  Variances,  and  Significance 
of  Differences  for  Principals  of  High  and  Low 

Structure  and  Consideration  . .  154 

XVII.  Linear  and  Nonlinear  Correlation  Coefficients  for 
the  Relationship  Between  Agreement  and  Two 

Dimensions  of  Leader  Behavior  (N=56)  .  155 

XVIII.  F  Ratios  For  Significance  of  Linear  and  Curvilinear 
Relationships  Between  Two  Dimensions  of  Leader 

Behavior  and  Agreement  . 

XIX.  Mean  Consensus  Scores,  Variances,  and  Significance 
of  Differences  for  Principals  of  High  and  Low 

Structure  and  Consideration  .  158 

XX.  Linear  and  Nonlinear  Correlation  Coefficients  for 
the  Relationship  Between  Consensus  and  Two 

Dimensions  of  Leader  Behavior  (N=56)  .  180 


XXV 


0 


-O'  s  . :  t  ■ 


v  ) 


TABLE 


PAGE 


XXI.  F-Ratios  For  Significance  of  Linear  and  Curvi¬ 
linear  Relationships  Between  Two  Dimensions  of 

Leader  Behavior  and  Consensus  . .  160 

XXII.  Mean  Consensus  and  Agreement  Scores,  Variances, 
and  Significance  of  Differences  for  Principals 

of  High  and  Low  Ambivalence . . .  162 

XXIII.  Linear  and  Nonlinear  Correlation  Coefficients 
for  the  Relationship  Between  Ambivalence  and 

Teacher  Consensus  and  Agreement  .  163 

XXIV.  F-Ratios  for  Significance  of  Linear  and  Curvi¬ 
linear  Relationships  Between  Ambivalence  and 

Teacher  Consensus  and  Agreement  .  I64 

XXV.  Mean  Consensus  Scores  by  Leadership  Style  of 

Principal  (N=56)  .  I67 

XXVI.  F-Ratios  Obtained  Through  Analysis  of  Variance 
of  Consensus  Scores  Grouped  by  Principals' 

Style  of  Leader  Behavior .  168 

XXVII.  Intercorrelations  of  Three  Variables  and 

Significance  of  Differences  For  Schools  of  High 

and  Low  Structure  Principals  .  170 

XXVIII.  Intercorrelations  of  Three  Variables  and 

Significance  of  Differences  for  High  and  Low 

Consensus  Schools  .  171 


xv 


' 


TABLE 


PAGE 


XXIX.  Mean  Agreement  Scores  by  Leadership  Style  of 

Principal  (N=56)  .  175 

XXX.  F  Ratios  Obtained  through  Analysis  of  Variance  of 
Agreement  Scores  Grouped  by  Principals' 

Style  of  Leader  Behavior  .  174 

XXXI.  Analysis  of  Variance  of  Agreement  Scores  for 

Structure  and  Consideration  .  174 

XXXII.  Significance  of  Differences  Between  Mean  Agreement 
Scores  for  S-+C  +  Group  and  Agreement  Means  of 

Three  Other  Groups  .  175 

XXXIII.  Mean  Consensus  Scores  by  Style  of  Leader 

Behavior  and  Tenure  .  178 

XXXIV.  F  Ratios  Obtained  Through  Analysis  of  Variances 
of  Consensus  Scores  Grouped  by  Principals' 

Style  of  Leader  Behavior  and  Tenure  .  179 

XXXV.  Linear  Correlation  of  Tenure  with  Consensus  for 
both  High  and  Low  Styles  of  Leader  Behavior 

With  Significance  of  Differences  .  180 

XXXVI.  Mean  Agreement  Scores  by  Style  of  Leader  Behavior 

and  Tenure  .  182 

XXXVII.  F  Ratios  Obtained  Through  Analysis  of  Variance  of 
Agreement  Scores  Grouped  by  Principals'  Style 
of  Leader  Behavior  and  Tenure  . 185 


xvi 


•  r 


■  •  ■  o 


.  . 


I 


TABLE 


PAGE 


XXXVIII. 


XXXIX. 


Linear  Correlation  of  Tenure  and  Agreement  for 
Both  High  and  Low  Styles  of  Leader  Behavior 


with  Significance  of  Differences  .  183 

Correlation  Matrix  for  Five  Variables  . . .  185 


xvii 


. 


LIST  OF  FIGURES 


FIGURE  PAGE 

1  The  Social  Systems  Model  .  14 

2  Response  Patterns  for  Three  Dichotomous  Items  ....  57 

3  Scalogram  for  Three  Dichotomous  Items  .  58 

4  Scalograms  for  Status  Items  . 120 

5  Scalograms  for  Authority  Items  . 122 

6  Scalograms  for  Personal  Items  .  125 

7  Scalogram  for  Mean-Ends  Items  .  128 

8  Distribution  of  Principals  in  Four  Quadrants 

Derived  from  LBDQ  Dimensions  (U=56)  .  138 


xviii 


r  '■ 


CHAPTER  I 


INTRODUCTION 


Daring  the  past  decade  there  has  "been  a  significant  change  in 
the  methodology  and  the  content  of  investigations  carried  out  hy 
students  of  educational  admini strati on.  Whenever  there  is  such  a 
change  there  probably  is  a.  danger  that  new  concepts  and  methods  will 
be  introduced  rapidly,  will  be  dealt  with  superficially  in  research, 
and  will  then  be  passed  over  in  favor  of  still  newer  approaches.  An 
additional  danger  in  the  study  of  administration  is  that  topics  which 
are  dealt  with  at  a  theoretical  level  may  never  be  developed  to  the 
point  where  their  full  implications  for  the  practice  of  administration 
become  known.  It  was  concerns  such  as  these  which  led  to  the  develop¬ 
ment  of  this  study.  Neither  the  concepts  nor  the  methods  are  particu¬ 
larly  novel;  they  have  been  used  before.  The  unique  contribution  of 
this  study  lies  in  the  attempt  to  combine  methods  and  concepts  in 
such  a  way  as  to  make  more  explicit  the  implications  of  administrative 
theory  for  administrative  practice  in  a  small  but  significant  area. 

The  methodology  employed  in  the  study  may  also  suggest  fruitful 
approaches  for  extending  our  knowledge  in  a  problem  area  where  there 
is  need  for  further  developments. 

The  sections  in  this  chapter  outline  the  purposes  of  the  study, 
indicate  the  need  for  extending  knowledge  in  this  aspect  of  administ¬ 
ration,  and  also  give  recognition  to  some  of  the  limitations  of  the  study. 
The  concluding  section  contains  a  guide  to  the  organisation  of  the  thesis. 


:  -  c7  7  J  c  7  •  ■ 

-  7'  .  r  :  7  ■  •  -jJ  '  7->-  01  ' 

--  •  ■ :7  ■'  7  i  77  :  '  7  ■ 

0  0  .  •  .  C  i  ■  :  7  • .7,  ■  ■  _  ■  •  •  .  ■  ; 

-  --  -  xoxi  <  77  .  '  ..  •  oi  : 

.  ■'  777  ■ 

j  i  j'  -  _r«.  ‘ 

.  V.  .  •  '  •  c 77  ,  7  ■  j  •  7.';  .  J-.,'  -  •,  • 

'  :  7  :  ■  7  •  77  :  7  77  7  7  •  '  -  ,, 


J.lXJ  -  x  ' 


^  ‘-I  '  -W  U 


i 


7  ■  • 


;  . 


>  ■  •  o.  . :  -.to  " 

7  . 


,r  ■. 


O'  i 


7  7  •  7  t 


-  o 

-  .  .  ;  7  ; 

7  r/7c  7J  ci  ■ 

7 

•  •  •  • 

---  •  o  7  7  77 0  ' 

•  i  8  7  7  • 

i  .  X  ;  ’  . ;  ;  t  ; 


2 


I.  RJRFOSES  OF  THE  STUDY 


Much  of  the  recent  research  in  educational  administration  has 
focused  on  the  social  nature  of  the  tasks  of  an  administrator.  Such 
research  has  been  aided  hy  various  theoretical  formulations  and  models 
(8,  9«  10)  which  serve  to  order  some  of  the  earlier  studies  and  to 
stimulate  new  enquiry.  One  of  the  concepts  or  variables  to  which 
these  draw  attention  is  the  expectations  which  different  individuals  in 
the  administrator's  life  space  hold  for  his  behavior.  In  discussing  the 
importance  of  these  various  expectations  Campbell  (3)  states: 

An  understanding  of  these  expectations,  often  conflicting 
in  nature,  may  appear  most  frustrating.  Only  by  such  under¬ 
standing,  however,  can  the  administrator  anticipate  the  reception 
of  specified  behavior  on  his  part,  Such  anticipation  seems 
necessary  if  the  area  of  acceptance  is  to  be  extended  and  the 
area  of  disagreement  minimized.  Moreover,  such  understandings 
are  necessary  if  a  program  of  modifying  expectations  is  to  be 
started,  (3,  p.  264) 

Before  an  administrator  can  understand,  before  he  can  modify,  in  fact 
before  he  can  act  in  any  way  with  respect  to  the  expectations  of  others, 
he  must  have  adequate  knowledge  about  the  nature  of  these  expectations. 

So  far  research  has  given  some  general  clues  as  to  what  these 
expectations  may  be,  but  this  same  research  has  also  shown  that  there 
are  wide  variations  in  expectations  from  situation  to  situation.  One 
of  the  purposes  of  this  study  was  to  explore  a  means,  scale  analysis, 
for  assessing  the  expectations  of  people  who  are  of  importance  to  an 
administrator  in  such  a  way  as  to  give  him  greater  understandings  upon 
which  to  base  his  actions.  It  was  hoped  that  the  results  of  the  study 


would  not  only  give  administrators  information  about  the  structure  of 


»  '  '  '  • '  ’  i  " '  j  : 

'  ''  '  :  ' 

"  o  '  '  . 

'■  •  '•  : 

.  '  '  T  : 

1 

, 

•  '  '  '  ' 

■ :  :  ’  • 

’  ■  « "  -  '  ' 

'■  •  ■'  ■ '  '•  ■  •  -  '  . 


-  > '  , 


•  r  ■  ■  ’•  •  •  r  .  i 


•  i . .V  •  •J  JFj'l  o  -U 


5 


expectations  which  are  held,  for  their  behavior,  but  that  these  would 
also  suggest  how  expectations  might  be  studied  comparably  in  a  variety 
of  situations. 

A  second  purpose  of  the  study  was  to  explore  the  possibility  of 
determining  from  self-report  instruments  variations  in  the  abilities 
of  administrators  for  dealing  with  the  frustrations  which  result  from 
the  fact  that  different  people  hold  different  expectations  for  the 
behavior  of  an  administrator.  If  this  were  a  normal  state  of  affairs 
in  the  reafity  of  administration,  and  research  seems  to  indicate  that 
it  is,  then  it  would  be  of  obvious  value  during  the  selection  process 
to  know  which  administraf ors  are  better  able  to  deaf  with  this  situa¬ 
tion.  This  study  subjected  the  concept  of  ambivalence  to  more  intensive 
testing  than  was  done  in  previous  research  and  also  studied  the 
relationship  of  this  variable  to  the  way  in  which  teachers  described 
the  leader  behavior  of  the  administrator . 

A  third  purpose  of  the  study  was  to  utilize  the  results  of  the 
scale  analysis  for  testing  a  number  of  hypotheses  concerning  relation¬ 
ships  among  the  leader  behavior  of  an  administrator,  the  extent  to 
which  there  is  agreement  between  the  administrator  and  his  subordinates 
on  role-expectations,  and  the  degree  of  consensus  within  the  group. 

In  this  particular  study  the  administrator  who  was  under  investi¬ 
gation  was  the  principal  of  a  school,  and  the  expectations  studied 


were  those  of  the  teachers  in  his  school. 


..  .  -  i; •'  .  :: 

:■  .  -x.  dx  -  c  •  .  •  •  n  •  I 


J 2  *'  ..  d  ,  ;;  hr:  oor. 

.  1>  ...  .:.  ■  >.'•  .  •  *  c,  •  .  ::  J:  ,  .5". 

•  •  o  .ex  d  v  .  r  ■■  x  .  . 

:  V.  ,  id  .  e  . ; 

.  X  .  .  '  >  '  ■  u"  '  X  X  L  ■.  ‘  '  .. .  -LVo  .  1 

x  ■  V  o'  c x  ;  xd  '  \  xx.'  .x  x 

x  .  .  ■.  . 

-  ■  xx  x  .  •.  •  .  ;  x  •  ex  .  x  x 

x  .  o'  .  d  c-  '  1  x  ■  '  .  xx 

"•  ox 

o  .  c  >0  G.r  v  .f  xd  •  j' .  \r 

>  L 

J  ..  •  '  J  •  x  1  .  ■.  ‘ :  .  <  1  ■  x  ■ 

i  o  X  '  xx  ■  J!  ■ ; 

?•  "  i  •  r  ■  Xf.  <  '  ri.'r  •  .  i  J: 

J  .  .  •  .  .  ■  o  ■■  .. 

•  '  > :  ...  •  .  .&  .  „ 

'  ■  .  ' 

-  d  5  x  ■  \  'x  .  i  r 

.  ?  ,.i:  ■  •  .  o  ■  x  J  '  .u  o  lc! 


II.  NEED  FOR  THE  STUDY 


4 


Reference  to  the  need  for  studies  such  as  this  has  already  "been 
made  hut  should  he  elaborated.  If  principals  need  to  know  what 
expectations  are  held  for  their  behavior,  and  if  the  expectations  change 
from  situation  to  situation  and  from  time  to  time,  then  there  is  a  need 
for  methods  which  will  facilitate  comparable  research  under  these 
conditions.  Such  research  may  be  carried  out  using  an  item  by  item 
approach  as  has  been  done  in  past  studies,  but  this  has  all  the  dis¬ 
advantages  of  opinion  polling  which  uses  only  unrelated  items.  There  is 
a  need  for  studies  which  will  explore  the  structure  of  the  expectations 
and  show  directions  in  which  further  progress  might  be  made. 

The  importance  of  having  adequate  knowledge  about  the  expectations 
which  are  held  for  a  role  incumbent  has  been  illustrated  by  studies  such 
as  that  carried  out  by  Femeau  (7)  who  investigated  the  expectations  for 
the  consultant  role  as  perceived  by  administrators  and  by  the  consultants 
themselves  and  then  related  the  degree  of  similarity  between  the  two  to 
the  extent  which  consultant  services  were  considered  to  be  effective.  The 
study  revealed  that  when  there  was  general  agreement  between  administrators 
and  consultants  on  the  consultant  role,  the  services  were  rated  as  more 
favorable  than  when  there  was  a  large  measure  of  disagreement  between  the 
expectations  of  the  administrator  and  the  consultant.  Femeau  concluded 
that  consultants  and  administrators  must  perceive  each  other  as  function¬ 
ing  according  to  their  expectations  if  the  consultation  is  to  be  effective. 

This  problem  may  not  be  overcome  as  easily  as  it  might  appear  for  the  study 


. 

•  .  rf 

.  '  '  ■  ' 

'  ;  '  :  '  ■  r  • 

'  '  '  i  s': 

....  .  .  • 

'■  •:  • 

-v  u  C  -  i  :  ut 

■  .•  '•...!  v  )r.a  2"  • 

,  ’  ■  .  • :  V  •  :  ■  ■  '  '  j  •  ‘ 

rf  ;i\  •  t  •  •  ’  .' 

...  '•  •  /  '/•;  r. 

.  '•  v '  '  ^  ' 

■  .  ;  '  -  '  '  •  •  ’ 


5 


revealed,  the  presence  of  regional  differences  in  the  expectations  which 
were  held  for  the  role  of  the  consultant.  This  suggests  that  a  consultant 
cannot  assume  that  he  has  acquired  a  thorough  knowledge  of  the  role  as  a 
result  of  his  experiences  in  one  situation;  he  may  well  have  to  adjust 
his  ideas  about  the  role  when  he  moves  to  a  different  position. 

The  importance  of  the  expectations  which  subordinates  and  super- 
ordinates  hold  for  each  other  is  also  illustrated  in  the  study  reported 
by  Moyer  (13).  Using  only  a  small  sample-  which  places  serious  limitations 
on  the  interpretation  of  his  findings-  Moyer  investigated  the  relationship 
between  teachers'  attitudes  toward  leadership  and  teacher  satisfaction.  He 
suggests  as  a  tentative  conclusion  that  teachers  do  have  a  mental  picture 
of  an  ideal  leader  and  that  the  closer  this  approaches  the  type  of  leader¬ 
ship  which  they  perceive  to  exist,  the  higher  will  be  their  satisfaction. 

He  states  further  that: 

...  the  principal  or  superintendent,  to  be  the  leader,  must  be 
aware  of  the  attitudes  of  the  teachers,  their  individual,  sub¬ 
group,  and  collective  differences  and  similarities.  Equipped 
with  this  knowledge  of  the  nature  of  his  group,  he  could  increase 
the  effectiveness  of  the  group  and  his  function  by  seeking  to 
unify  and  harmonize  the  prevailing  differences  among  them  and, 
at  the  same  time,  attempt  to  bring  his  own  leadership  attitudes 
into  a  compatible  relationship  with  his  group.  His  position 
of  leadership  will  probably  be  even  stronger  if  his  attitudes 
are  beyond  the  aspirational  tendencies  of  the  group.  (l3>  p.3) 

Moyer  also  suggests  that  in  order  to  become  keenly  aware  of  his 

subordinates'  attitudes  toward  leadership,  a  school  administrator  should 

use  some  system  for  determining  what  the  expectations  are  as  a  first  step 

toward,  unifying  these  and  modifying  his  own  behavior  and  the  expectations 


of  subordinates. 


■ 


It  ;  t'.:' 

■  ■ 

. 


■ 


.  ' 


. 


' 


»  •.  . 

' 

.... 

'•  '  '  ' 

.  >')  •' 


,  : ' 


6 


Chase  (4,5)  has  reported  on  a  number  of  studies  from  which  a 

general  conclusion  may  be  drawn  to  the  effect  that  when  teachers' 

expectations  with  regard  to  leadership  are  fulfilled  their  morale 

rises  but  when  they  are  not  fulfilled  their  morale  drops.  These 

same  studies  also  indicate  that  the  problem  is  complicated  by  the 

fact  that  the  expectations  vary  from  community  to  community.  Yet, 

if  he  is  to  fulfill  expectations  . . . 

The  administrator  needs  accurate  information  regarding  the 
teachers'  expectations  as  to  what  functions  should  be  carried 
out  by  the  school,  who  should  carry  out  particular  functions, 
and  how  the  person  or  persons  should  perform  these  functions. 

(5,  P-  1) 

Chase  suggests  that  the  expectations  can  be  determined  through  such 
procedures  as  day  to  day  contacts,  the  setting  up  of  grievance 
committees  and  such  devices  as  questionnaires  on  which  subordinates 
indicate  their  expectations  for  the  behavior  of  the  administrator. 

It  would  seem,  though,  that  before  this  approach  can  be  truly  useful, 
the  administrator  needs  to  know  what  questions  to  ask  and  how  to  ask 
them.  Both  of  these  are  problems  for  empirical  research  before  they 
can  be  utilized  by  practitioners. 

Two  other  investigations  carried  out  independently  by  Valenti  and 
Smith  (summarized  in  6)  support  the  observation  that  the  process  oi 
determining  expectations  is  complicated  by  variations  from  situation  to 
situation.  Valenti  developed  a  scale  to  measure  attitudes  toward  seventeen 
personnel  problem  areas;  his  investigation  revealed  that  different  teachers 
take  different  approaches  to  their  work  and  that  there  is  really  nothing 
like  a  "typical"  teacher.  Nor  was  he  able  to  predict  teachers'  attitudes 


'  ,'V  A. 


'  ■  '  x 


.  V-  ^ 


■  •  J  >.  .  . 


<  -  j  ■  >  • 


;i '  o 


•'  i  ..V 


z 


X 


.  .  :  . : 


V.  .  .  *  • . .  . 

.  •  ..  f 

.  : 

.  .  .• :  :■ ..  ■  •>.!’ 1  ■  ■ 


.  !'■ :  ' .  r.  ;  • 


c,  ;  .  >.  ■ '  .  ' '  ■ 

.  c 

■  •  •: 


' 

1  c 

.  . 

^  •- ; 

/  . 


■  "  :  -■  ,  .  r  ■.  '  ■  •:  ,•  ^ 


.  ■  ; 


-  -  ■  -  ■  •  ioxv  i 

o . '  x;.J 


■  ■  r.-:  •  ;  •  ■' 

».  . .  i. 

■  •  ••  j.\.  X 

.  .  . 


'  ■  '  .  r  • 

. 

v  •  -  •  ‘  y  ... 


•  r  .  ,  IX 


■ :  f  c  •  •  .  ,  .  ;  r 


'  :u  a  :  ;'  r. 


:.y  jr 


7 


from  other  characteristics  of  the  teachers.  As  a  result  of  this 
information  one  tends  to  agree  with  the  conclusion  . . . 

The  school  situation  in  which  a  teacher  or  administrator 
works  appears  to  shape  his  attitudes  far  more  than  his  personal 
characteristics.  Wide  differences  were  found  in  the  responses 
among  school  systems  and  among  schools  in  the  same  systems. 

(6,  p.  2) 

The  Smith  study  supported  the  findings  and  the  conclusion  stated  above. 

When  he  investigated  the  attitudes  of  teachers  in  a  seven-school  system, 
Smith  found  that  the  composite  attitudes  of  the  staff  in  one  school  could 
not  be  inferred  with  accuracy  from  the  knowledge  of  the  total  system  score. 
This  emphasizes  the  need  for  developing  methods  and  instruments  for  assess¬ 
ing  these  various  attitudes  and  expectations. 

The  above  studies  and  others  such  as  those  by  Bidwell  (l), 

Campbell  (2),  and  Hal pin  (ll)  tend  to  support  the  need  for  further  research 
into  the  expectations  which  are  held  for  the  incumbents  of  administrative 
positions.  Furthermore,  training  programs  are  in  need  of  predictors  which 
will  aid  in  the  selection  of  those  individuals  who  may  be  better  able  to 
cope  with  this  reality  of  administrative  life;  it  is  hoped  that  the  study 
of  leader  ambivalence  might  give  some  insights  into  this  problem. 

There  has  also  been  much  conjecture  but  comparatively  little  evidence 
on  the  extent  to  which  educational  administrators  are  able  to  influence  the 
expectations  which  3taff  members  hold  for  their  behavior.  The  first  step 
toward  utilizing  knowledge  in  this  area  for  the  preparation  of  adminis¬ 
trators  would  be  to  determine  the  extent  to  which  practicing  administrators 
differ  in  ability  to  do  this.  For  a  field  of  study  which  is  oriented 


. 

'  >J 

' 

'  • 

■  ■ 

•  .  >  '  ’O ' 

■  :  •'  ■  •  1 


■ 

•  •  ■ 

. 

....  ,  : 

' 

•  .  ■  [»;■•  (j 

'  '  1 


8 


toward  practice  to  such  an  extent  as  is  administration,  it  is  not  enough 
to  theorize  about  influencing  expectations.  The  theory  must  suggest 
hypotheses  and  the  hypotheses  must  then  be  tested  so  that  the  implications 
of  theory  for  practice  can  be  determined. 

III.  LIMITATIONS  OF  THE  STUDY 

One  of  the  problems  with  which  this  research  project  was  forced  to 
cope  involved  the  application  of  methods  which  had  not  been  used  in  this 
particular  type  of  investigation  but  which  had  been  employed  success¬ 
fully  in  related  social  sciences.  It  was  felt  that  if  the  same  methods 
appeared  to  be  useful  in  the  study  of  administrative  phenomena,  then 
what  was  initially  a  limitation  might  become  a  major  contribution. 

This  research  was  also  plagued  by  another  problem  which  is  common 
to  all  the  social  sciences  in  that  it  sought  to  find  significant  relation¬ 
ships  between  a  few  variables  from  among  the  many  which  no  doubt  were 
present  in  the  situations.  One  can  assume  that  all  other  things  were 
equad,  but  in  the  complex  field  of  human  relations  one  can  rarely  place 
much  confidence  in  such  an  assumption.  It  was  recognized  that  if  the 
hypothesized  relationships  did  exist  at  all,  it  was  likely  to  be  at  a 
low  level  of  significance .  However,  even  such  slight  relationships  may 
serve  to  indicate  fruitful  directions  for  more  intensive  and  more  care¬ 
fully  controlled  investigation. 

When  hypothesized  relationships  are  found  to  exist  there  is  also  a 
great  temptation  to  attribute  cause  and  effect  without  due  caution.  This 


i'  ■  •  •  ; ' 


„C; 


■ 


i  i ! . ). 


. 


i  : 


)  '  .xi  1  up 


.  •:  :  c  .  •  '■  :  .  .u  vo  -  V.  • 

•  i;  -  i  h  ©i.  -  >'  vI 


.  -  V.  '  -  .  0 


r  f,r 


?  > 


:  :  ■;  •  • 


:  i  i  > 


;  ..... 

•. 


i  - 


: .  .  \ .  ■ 


•  j:  ;. 


T 


' )  i.  . _  J  '  -  ...  ■  \ 

j 


•  :  ::  bst 


.'  v.  ■  .  .  '  .  t  : 

.  J  '  C  '  U  .  , 


' 


.  1  ,, 


:  c . 

c  .  :  e  :  ;  . 

j; 

v  r 

.V'  .  v  L"  ■  1  ‘ . 


:  '  ' 


:  ■  r. 


9 

is  particularly  true  in  the  case  of  an  exploratory  study  which  does 
not  have  an  extensive  grounding  in  theory.  In  this  study  there  has  been 
a  deliberate  attempt  to  avoid  such  a  pitfall,  but  there  is  always  the 
danger  that  not  enough  caution  has  been  observed. 

The  exploratory  nature  of  the  study  did  not  make  possible  a  design 
as  rigorous  as  might  be  desirable.  This  has  its  attendant  problems  but 
some  possible  advantages  as  well.  It  may  be  that  what  is  a  limitation  in 
one  respect  is  an  advantage  in  another.  Merton  has  indicated  that 
empirical  research  may  serve  some  functions  in  addition  to  the  testing  of 
hypotheses  (12,  p.  102).  One  of  these,  the  serendipity  pattern,  suggests 
possible  reformulations  in  existing  theory.  Merton  has  defined  seren¬ 
dipity  as  involving 

...  the  unanticipated,  anomalous  and  strategic  datum 
which  exerts  pressure  upon  the  investigator  for  a  new 
direction  of  inquiry  which  extends  theory  (12,  p.  105). 

Although  serendipity  may  enter  the  scene  even  if  a  study  is  designed  in 

such  a  way  as  to  control  carefully  certain  variables,  there  is  a  greater 

possibility  for  it  to  operate  if  not  all  other  variables  are  completely 

restricted  and  the  eyes  of  the  investigator  closed  to  interesting 

variations.  There  was  room  for  the  serendipity  pattern  in  this  study  and 

to  some  extent  serendipity  did  appear  as  will  be  revealed  in  the  text  of 

the  thesis.  It  may  well  be  that  this  finding  is  much  more  significant 

than  the  hypotheses  originally  stated;  however,  it  is  in  such  an  instance 

that  there  is  a  great  temptation  to  attribute  a  causal  relationship 

without  stipulating  that  this  is  only  a  hypothesis  for  further  investi¬ 


gation. 


/ 

' 

■  ■.  •  '  ■  '■  '  •  '  J:.  ' 

■ 


v-  .  ■  '  :< v 


■ 


'  ' 


’ 


‘  ••  '  ■  ■•T 


' 


. 


10 


Finally,  the  extent  to  which  the  results  of  the  study  can  he 
generalized  is  limited  by  a.  number  of  factors.  The  study  dealt  with 
only  one  type  of  administrative  position,  namely,  the  principalship 
and  this  in  a  particular  type  of  formal  setting.  Although  evidence 
will  be  presented  to  show  that  this  sample  wa,s  generally  representative 
of  the  population  of  principals  in  similar  settings,  it  does  not  justify 
making  generalizations  to  principals  in  different  formal  settings  let 
alone  different  administrative  positions.  This  study  can  only  suggest 
what  needs  to  be  done  in  other  situations  in  order  to  make  full  use  of 
existing  theory  in  empirical  research. 

IV.  ORGANIZATION  OF  THE  THESIS 

The  following  chapter  deals  to  a,  greater  extent  with  the 
theoretical  bases  of  the  study  and  includes  discussions  and  definitions 
of  various  terms  used  such  a,s  role  conflict,  ambivalence,  and  leader 
behavior.  Then,  significant  related  research  is  reviewed  and  its 
implications  for  the  study  reported  herein  are  discussed.  Consider¬ 
able  space  is  devoted  to  a  consideration  of  the  theory  and  methodology 
of  scale  analysis  which  is  of  central  importance  to  a  major  portion  of 
the  investigation.  The  uses  to  which  this  procedure  has  been  put  are 
examined  critically  in  order  to  establish  its  suitability  for  the 
purposes  of  this  study. 

In  the  next  section  of  the  thesis  the  methods  ol  the  study,  the 
results  of  the  scale  analysis,  and  some  descriptive  statistics  are 


:  •  d  .  .  J  ' :  •'  ,  AA 

.  :  A  d;  •  I:  .L  . 

d:  :  '  A  •  J  A;-  A  .  c  •  d  . 

,  -  -  »  d.A :  .  A  A  ;  '  J  ■ "  .  c  d:A-  •  A  '•  -  > 

■  ■  ■.  ■  '  '  -j.  .Ad. 

'  •  .  l  d:A  -A  '■  (  :L 

d:  A  xi  A  d  '  dd  •  A  :  : 

■  .x  '  .  \dd'  \A'  d:  d  d  ■.  /  A' 

„c  id-  '  -  d  x  onof  d  ■  d  ■ 

.  o  d  ‘  .  AA 

:  ;  : 

;d  -  c  d:  A.  A  : . 

x  ’  n  L  x  or.x  .  ;  x  >  &  fid  d  I&i  xd  ■  d 

c  ?  A c.dAA  :  .  •  :  d  •.  ■ 

0*100  r  oxi.i:  1  ..  .  b ' 

;  d;  \  .(  ■  ;  \d"  d  •  A  •  A.  A  oil;:  •  J: 

d  j  A  A  Ad  q  A  d  ■  ■  ax  0G(  A  1 

c  A  -'.A'  :  A  ■  d:  ;  d:  .  ■■  A  ;  •  ;.J  ;  ,  'r'i 

o  ■  f.d:  A-  cd;  d  ,  d:d  •  dd  d  d 

d:I  J'  c  did  d.  d:  ■ 

J* ;  d:  d 

A  ■  d  d:  A  •  .  ■  A  ;  . 

d  o  »  ,  d;  c 


11 

presented.  The  analysis  of  the  data  concludes  with  a  detailed  description 


of  the  testing  of  the  hypotheses  and  the  interpretation  of  this  statis¬ 
tical  treatment.  The  final  chapters  of  the  report  contain  general  con¬ 
clusions  as  well  as  the  implications  of  the  findings  for  the  practice  of 
administration  and  for  future  research. 


•  •  ■  ,  ;  ::  '  '  • 

'  '  ■  dr  "■  . '  ' 

’•  .  '  '  ' 

‘  ’ 


„  ' .  :  ■  ■ ' 


REFERENCES  FOR  CHAPTER  I 


Bidwell,  Charles  E.  "The  Administrative  Role  and  Satisfaction 
in  Teaching."  Journal  of  Educational  Sociology.  29:41-47, 
September,  1955 • 

Campbell,  Merton  Y.  "Teacher- Principal  Agreement  on  the 
Teacher  Role,"  Administrator's  Notebook,  Vol.  7>  No.  6, 
February,  1959* 

Campbell,  Roald  F.,  and  Russell  T.  Gregg  (eds.).  Administrative 
Behavior  in  Education.  New  York:  Harper  and  Brothers,  1957. 

Chase,  Francis  S.  "Professional  Leadership  and  Teacher  Morale," 
Administrator’s  Notebook,  Vol.  1,  No.  8,  March,  1955. 

Chase,  Francis  S.  "How  to  Meet  Teachers'  Expectations  of 
Leadership,"  Administrator's  Notebook,  Vol.  1,  No.  9* 

April,  1953* 

Chase,  Francis  S.  and  William  W.  Savage,  "Attitudes  of  Teachers 
and  Administrators, "  Administrator's  Notebook,  Vol.  3?  No.  1, 
September,  1954 • 

Femeau,  Elmer  F.  "Which  Consultant?"  Administrator's 
Notebook,  Vol.  2,  No.  8,  April,  1954* 

Getzels,  J.  W.  "A  Psycho-Sociological  Framework  for  the  Study 
of  Educational  Administration,"  Harvard.  Educational  Review, 
22:235-246,  Fall,  1952. 

Griffiths,  Daniel  E.  Administrative  Theory.  New  York: 
Appleton-Century-Crofts,  Inc.,  1959- 

Hal pin,  Andrew  W.  (ed.).  Administrative  Theory  in  Education. 
Chicago:  Midwest  Administration  Center,  1958* 

Hal  pin,  Andrew  W.  (ed.).  The  Leadership  Behavior  of  School 
Superintendents .  Second  Edition.  Chicago:  Midwest 
Administration  Center,  1959* 

Merton,  Robert  K.  Social  Theory  and  Social  Structure. 

Glencoe,  Illinois:  The  Free  Press,  1957* 

Moyer,  Donald  C.  "Leadership  that  Teachers  Want," 
Administrator's  Notebook,  Vol.  3»  No.  7>  March,  1955* 


I 


CHAPTER  II 


THEORETICAL  BASES  OP  THE  STUDY 

The  major  purposes  of  the  study  have  Been  outlined  in  the  previous 
chapter.  Before  the  problem  can  be  stated  more  specifically  in  the  form 
of  hypotheses  to  be  tested,  it  is  necessary  to  develop  some  of  the  theo¬ 
retical  framework  in  which  the  problem  in  general  is  grounded.  In  the 
sections  that  follow  a  number  of  concepts  will  be  discussed  and  developed 
in  such  a  way  as  demonstrate  their  relevance  for  the  study  reported  herein. 

I.  A  THEORETICAL  MODEL 


To  a  large  extent  the  theoretical  basis  of  this  study  lies  in  the 
model  developed  by  Getzels  and  Guba  ( 3 ?  4)  to  which  an  indirect  reference 
has  already  been  made.  The  original  model  (4)  and  the  extensions  of  the 
model  (5)  which  have  been  enunciated  more  recently  emphasize  the  social 
nature  of  the  tasks  of  an  administrator.  As  a.  theory  of  human  behavior 
it  has  been  used  widely  in  discussing  organizational  behavior,  and  it 
has  also  proven  to  be  valuable  in  the  development  of  testable  hypotheses 
about  the  interrelationships  of  the  variables  to  which  it  draws  attention. 

Getzels  has  shown  the  relevance  of  the  model  for  the  study  of 

administration  if  we  approach  the  problem  in  such  a  way  as  to 

...conceive  of  administration  structurally  as  the  hierarchy  of 
subordinate -super ordinate  relationships  within  a  social  system. 
Functionally,  this  hierarchy  of  relationships  is  the  locus  for 
allocating  and  integrating  roles  and  facilities  in  order  to  achieve 
the  goals  of  the  social  system.  It  is  here,  in  these  relationships, 
that  the  assignment  of  statuses,  the  provision  of  facilities,  the 
organization  of  procedures,  the  regulation  of  activity,  and  the 
evaluation  of  performance  take  pla  ce.  (5>  P •  151 ) 


.  •  ; 

X  .  ■  . .  •  7  • 

,  ■  ...  '  .  .• 

..  ■  :  X  .  X  e  ■  !  ■  77  ...  ■  'll  '  ' .XX  • 

.  ■ 

■  .  ' 


•  7:  •  X  oXX  o 

X  ■ 

:  ;■  •  X  ;  .7 

X  »  ■ 

X.-  'X' 

' 

: 

X; 

X 

.  X  X  .  .  •  .  X  .'  '  .  -  ■ '  :r 

.  •  -  X  '  ■  •  ■  c  i:  X  /  '  i"’  .X  "  ■  X  J X 

<  oX  j.C  ■  ■:  -  c  1  '■  rn  so  ct;'  c-bXs  B.ci: 

XX  ■  '  X  i  X  .  •...  XX  •  X X'r  -  ■ 

c  .. ...  -  .  X  o  ••  ;  ■  ...  '...I  "  ,':J  "  X 

:  •  •  .  .  . 

.'  :  :■  .V  j  •  i  J:.!.  ;  ••.  r  ~  •  X  . 

.  O'  .  :  .  •• 

.  :  •  vXX'."7 ,  -X '  :J:  ■"  .  77  c  11  ■ 

.  '  .  X.  '. 

7  ■  V.o  X  .  '  X 

<  ?  X  ' 

.  ,  .  :  .  0 


7  7  /  -  '  /  •'  X  ..  ..  i  \  .  •' 

X'  7:  '  IJQSO 

-  d  ’  ;  ; 


14 

The  nature  of  the  social  relationship  is  the  crucial  factor  because 
these  various  functions,  which  are  those  of  the  administrator,  can  be 
carried  out  effectively  only  insofar  as  they  are  accepted  by  subordinate 
members.  The  central  concept  of  the  social  system  draws  further  attention 
to  the  importance  of  these  relationships. 

The  social  system  in  question  may  be  any  unit  of  interpersonal 
relations;  it  may  be  a,  classroom,  a  school,  a  school  system  or  even  the 
broader  community.  Within  the  social  system,  whatever  it  may  be,  there 
are  in  general  two  classes  of  phenomena  which  must  be  dealt  with.  The 
first  of  these  consists  of  the  roles  and  expectations  that  will  fulfill 
the  goals  of  the  system,  and  the  second  consists  of  individuals  with 
certain  personalities  and  need-dispositions  who  inhabit  the  system.  It 
is  postulated  that  the  behavior  of  individuals  within  the  system  is  a 
function  of  the  role  and  personality  dimensions.  That  is,  the  indi¬ 
viduals’  behaviors  can  be  attributed  to  the  expectations  which  are  held 
for  them  and  to  their  own  need-dispositions.  The  model  can  be  con¬ 
veniently  diagrammed  in  the  following  manner: 


.Institution 


Social 

System 


^Individual- 


■>  Role 


■^Expectation 


Observed 

Behavior 


^Personality — ^Need-Disposition 


FIGURE  1 

THE  SOCIAL  SYSTEMS  MODEL 

This  illustrates  that  the  social  system  can  be  viewed  as  consisting 


. 


■ 

'  '  '  '  '  ■  ■ 

* 

'■  ' 

■  ;  r‘\  .  .  .  ■" 

'■  - 1  ,  "  •  .  • '  ' 

■  '  '  ■  '•  ’ 

- 

,  '•  '  :  '  '  •"  '  f  ' 

'  .  ■ .  '  v .  i  >  ‘ 

'  ,  ■  -  \ 

.  .  '  '■  :■  •  ' 

' :  v  j  '•  •  ■: "  :'v 


. 

■ 

;  f>  ' 

' 

! 

;  i  >ni: 


r  ■ 


' 


15 


of  institutions  which  are  defined  by  a  series  of  complementary  or  inter¬ 
locking  roles.  Each  role  is  in  turn  defined  by  the  expectations  which 
are  held  for  the  role  incumbent.  The  social  system  can  also  be  thought 
as  consisting  of  individuals  with  their  unique  personalities  as  determined 
by  their  peculiar  need-dispositions.  Consequently,  the  behavior  of  an 
individual  within  a  social  system  is  conceived  as  being  a  function  of 
both  his  role  and  his  personality. 

One  thing  which  the  model  does  not  show  is  the  proportion  of  any 
given  act  which  can  be  attributed  to  each  of  the  two  dimensions.  It  is 
fairly  evident  that  this  will  vary  from  individual  to  individual  and  from 
situation  to  situation.  Two  individuals  occupying  similar  positions  within 
the  social  system  will  be  influenced  to  varying  degrees  by  their  needs  and 
the  expectations  held  for  them.  Similarly,  positions  vary  in  the  amount  of 
behavior  which  is  specified  by  expectations  and  the  extent  to  which  this  is 
permitted  to  be  determined  by  individual  needs.  For  this  reason  certain 
behaviors  such  as  those  of  the  artist  might  stem  largely  from  his  person¬ 
ality  while  behaviors  which  one  observes  in  a  total  institution  may  stem 
almost  entirely  from  the  formal  role  prescriptions. 

It  is  also  fairly  evident  that  individual  behavior  can  be  influ¬ 
enced  by  expectations  only  to  the  extent  that  and  in  the  way  that  the 
individual  perceives  the  expectations.  These  may  be  totally  different 
from  the  expectations  as  defined  by  an  external  observer.  To  account  for 
the  observation  that  the  incumbents  of  certain  complementary  roles  do  not 

always  agree  in  their  perceptions  of  mutual  obligations,  Getzels  has 


' 


■  ■■  ■ 


•  •  ■ 


■- 


■ 


•  1 


"  "  '  '  '  •  ‘ 

f*  n 


' 

' 


16 


introduced  the  concept  of  selective  interpersonal  perception  (3,  p.155)* 

That  is,  each  role  incumbent  perceives  and  organizes  the  expectations  for 

his  and  complementary  roles  in  a  unique  way.  Some  preliminary  research 

has  led  Getz els  to  suggest  that 

...  the  functioning  of  the  administrative  process  depends  not 
only  on  a  clear  statement  of  the  public  expectations  but  on  the 
degree  of  overlap  in  the  perception  and  private  organization  of 
the  expectations  by  the  specific  role  incumbents.  (3,  p.  156) 

If  the  degree  of  overlap  between  the  expectations  held  by  the  incumbents 
of  complementary  roles  is  so  important,  one  might  well  ask  why  so  much  of 
this  should  be  left  to  chance.  Indeed,  in  formal  organizations  there  are 
deliberate  attempts  to  outline  and  to  communicate  appropriate  expectations, 
but  not  much  is  known  about  ways  in  which  congruence  of  perceived  expec¬ 
tations  among  role  incumbents  might  be  fostered. 

The  model  has  proven  to  be  useful  in  the  derivation  of  hypotheses 
in  the  area  of  selective  interpersonal  perception  and  in  such  areas  as 
conflict  between  and  within  each  of  the  dimensions  which  will  be  discussed 
in  the  following  section.  This  study  focused  on  the  institutional 
dimension,  particularly  on  expectations  held  by  subordinates  for  the 
behavior  of  a  superordinate.  Some  of  the  studies  illustrating  the 
importance  of  the  way  in  which  perceptions  of  expectations  enter  into 
administrative  relationships  have  already  been  outlined;  these  serve  as 
an  example  of  the  usefulness  of  the  model  for  organizing  existing  research 


evidence. 


»  9 

' 

' 

' 


' 

wm 

,  r  •  -  ••..r 

,  •  i  •  •  Y  - 

' 

*  *  * 

■;  ;•  ; 

■Kjmg 

' 

•  '  ' 

-  • : 

•  O  :  ,  ■  •  ■ 

loiriv  enoisi  i 

•  •  .  '  ■  '  «  '  ' 

' 

•  •.  .  ...  ■  '  "  ' 


II.  INSTITUTIONAL  AND  INDIVIDUAL  CONFLICT 


17 


Further  derivations  from  the  model  suggest  three  main  sources 
of  conflicts,  or  as  Getzels  says  "interference  of  reactions,"  (3,  p.l6l) 
in  administrative  settings.  These  are  (l)  role-personality  conflicts, 

(2)  personality  conflicts,  and  (3)  role  conflicts  which  stem  from 
incongruence  in  the  interaction  between  the  two  dimensions,  within  the 
personal  dimension,  and  within  the  institutional  dimension  respectively. 
When  such  incongruence  exists  there  is  likely  to  be  a  loss  in  individual 
and  institutional  productivity;  for  this  reason  each  of  these  becomes  an 
area  for  concern  in  the  study  of  administration.  Empirical  research  has 
shown  that  each  of  these  conflicts  do  exist,  and  so  this  is  more  than  an 
attempt  at  theory  -  building.  The  present  study  was  concerned  with  role 
conflicts  and  so  this  aspect  of  interference  will  be  discussed  in  greater 
detail . 


Role  Conflict 


Numerous  role  studies  have  shown  that  in  general  there  is  a  lack 
of  consensus  among  the  various  groups  and  individuals  who  may  hold 
expectations  for  the  behavior  of  a  particular  role  incumbent.  This  is 
usually  referred  to  as  role  conflict  and.  has  been  defined  in  this  way: 

Role  conflicts  occur  whenever  a  role  incumbent  is  required 
to  conform  simultaneously  to  a  number  of  expectations  which  are 
mutually  exclusive,  contradictory,  or  inconsistent,  so  that 
adjustment  to  one  set  of  requirements  makes  adjustment  to  the 
other  impossible  or  at  least  difficult.  Role  conflicts  in  this 
sense  are  situational  givens  and  are  independent  of  the  person¬ 
ality  of  the  role  incumbent.  (3»  p.  I6l) 

This  indication  of  disorganization  in  the  nomothetic  or  institutional 


’ 

' 

. 

■  .  •  .  •••••;••'  ■  • 

' 

*  ‘  ’■ 

' 

f  -  '  •'  p. 

r  '  '• 

' 

, 


18 


dimension  can  arise  in  a  number  of  ways.  A  type  of  conflict  which  one 
encounters  more  frequently  in  popular  writing  is  that  arising  from  the 
fact  that  an  individual  occupies  two  or  more  roles  which  may  he  in  conflict. 
One  example  of  this  would  be  the  teacher  who  also  has  a  son  or  daughter  as 
a  student;  at  times  there  may  be  conflict  between  the  parent  and  the  teacher 
role.  Such  conflict  is  not  encountered  too  often  in  formal  organizations 
for  the  same  individual  is  seldom  required  to  fulfill  expectations  for 
more  than  one  role.  Although  the  case  of  the  superintendent  who  is  expected 
to  act  as  a  consultant  for  teachers  and  at  the  same  time  carry  out  teacher 
evaluation  for  the  Department  of  Education  may  be  a  case  in  point. 

A  second  type  of  role  conflict  occurs  when  an  individual  is  faced 
with  disagreements  among  several  alter  groups  who  each  may  have  legitimate 
reasons  for  defining  the  expectations  for  the  role.  Thus  a  principal  may 
be  faced  with  the  conflicting  expectations  of  his  staff,  of  the  parents, 
of  the  students,  and  of  the  superintendent.  This  is  an  example  of  what 
Gross  has  called  interposition  conflict  (7,  p.  116)  and  Seeman  has  referred 
to  as  conflict  between  criterion  groups  (14,  p.  4-0).  The  present  study  was 
not  concerned  with  this  type  of  conflict. 

A  third  type  of  conflict  must  be  considered  because  seldom  is  there 
found  to  be  complete  agreement  in  the  expectations  held  by  the  individuals 
within  any  one  criterion  group.  This  may  be  a  particularly  difficult  problem 
for  an  administrator  because  some  of  the  means  of  conflict  resolution  which 
are  available  in  the  other  cases  are  not  as  easily  applied  here.  An  example 
of  this  type  of  conflict  is  found  in  the  situation  where  the  principal  of  a 
school  is  expected  by  some  of  the  teachers  to  give  them  considerable  assistance 


■ 


;  ■  :9  Hi  xi  ex 


gj 

1  ' 

' 

.  exriiT  .slox  erti  sol  aoojtf«f 

' 

.  ■  a  £  .d  ■.  lo  < 

OH  .0-  ,Y)  tomnoo  rtoxJ-I.  0 

.  : 

,&o':£ir:oo  ?co  ocrvl  ■■  0  :.t  tIv  >orr:s  :v  or 

' 

' 

. 


■ 


19 

with  their  problems  and  by  others  to  give  them  a  great  deal  of  professional 
freedom  and  room  for  experimentation.  The  principal  cannot  adopt  the  same 
style  of  behavior  with  all  the  teachers  but  must  try  either  to  reconcile 
the  differences  or  act  differentially  with  the  teachers  on  his  staff.  In 
any  event  he  may  be  perceived  as  behaving  inconsistently  by  his  staff 
members  and  perhaps  be  subjected  to  considerable  strain  as  a  result  of 
having  to  adjust  his  behavior  continuously.  The  importance  of  role  conflicts 
suggests  that:  the  dimensionality  of  conflicting  expectations  should  receive 
more  attention  in  empirical  investigations. 

III.  DIMENSIONS  OP  CONPLICTING  EXPECTATIONS 

Seeman  has  proposed  that  the  study  of  leader  behavior  within  a 
situational  frame  of  reference  should  not  be  restricted  to  either  the 
interpersonal  network  view  or  the  group  dimensions  approach  (14,  p.  4)» 
Instead,  he  suggests  that  there  is  much  to  be  gained  by  adopting  a  total- 
cultural  view  which 

...  stresses  the  fact  that  both  the  leader  and  his  specified 
group  operate  in  a  larger  cultural  situation.  The  basic  question 
is:  In  what  way  is  leader  and  member  behavior  ...  a  function  of 
conceptions  or  social  structures  which  characterize  the  culture 
of  which  both  the  group  and  its  leader  are  a  part?  (l4>  p.  4) 

This  approach  attempts  to  bring  the  studies  of  leadership  closer  to  such 

concepts  as  status  and  role  and  to  relate  them  to  such  broader  concerns  in 

sociology  as  the  study  of  stratification.  Studies  which  have  been  carried 

out  using  this  view  have  adopted  the  general  proposition  that  leadership 

behavior  and  ideology  are  to  a  significant  extent  functions  of  status 


,  •;  a  ■  ■ 

'  '  '  ,  -  : 


■ 


l:  IT 


....  .  : 


/ri*  i  :• I  j  :  '  'iF.ff  Hi® 

•  I  .  v  ■  ;  .  •  0/.'-  •  '  ' 


■  ; 

.  •;  i;  0?orij'  - 


... 

'  :  •'  ■  J:  v  '  < 

.  •  •  •  •  '  '  ;■  ■;  \  >'v : C’ •  '■ 

... 


20 

considerations  which  stem  from  the  community  and  culture  surrounding 
the  given  organization  (14,  p.  6). 

Value  C  ontradic  t ions 

One  particularly  useful  aspect  of  the  total -cultural  orientation 
is  that  it  suggests  some  dimensions  for  the  study  of  role  conflicts  in 
leader  behavior.  If  the  culture  is  characterized  by  certain  contradictory 
value  emphases,  it  may  very  well  be  that  these  contradictions  are  special 
problems  for  leaders  in  institutionalized  positions.  People  who  do  not 
hold  leadership  positions  would  probably  encounter  the  same  conflicts  but 
not  to  the  same  extent  as  would  leaders.  Observers  of  the  American  culture 
have  pointed  out  a  number  of  contradictory  value  emphases,  and  it  is  likely 
that  the  same  contradictions  are  to  be  found  in  any  western  society.  The 
four  dimensions  which  Seeman  has  identified  in  the  writings  of  Myrdal  (13), 
Williams  (19),  Fromm  (2),  and  Stouffer  (l6)  are  the  status,  the  authority, 
the  personal,  and  the  means-end  dimensions  (14,  pp.  8-ll). 

The  status  dimension  includes  the  conflicting  values  between  hierarchy 
on  the  one  hand  and  the  equality  ideology  on  the  other.  The  culture  glori¬ 
fies  success  and  rank  at  the  same  time  as  it  also  denies  the  significance  of 
status.  The  authority  dimension  involves  the  conflict  between  the  values 
of  dependence  and  independence.  On  the  one  hand  there  appears  to  be  a 
general  passivity  and  dependence  on  leadership,  while  on  the  other  hand 
there  is  a  demand  for  participation  and  self-determination.  The  personal 
dimension  involves  the  choice  between  universalistic  and  particularistic 


criteria  for  behavior;  the  universalistic  obligation  is  applicable  to 


■ 


■ 


V 


’ 

•• 


21 

anybody  while  the  particularistic  is  limited  to  persons  who  stand  in  some 
special  relationship  to  the  actor.  The  real  problem  enters  when  an  actor 
must  decide  between  acting  in  a  universalistic  manner  because  of  an  obli¬ 
gation  to  society  and  acting  in  a  particularistic  manner  because  the  object 
is  a  friend.  Finally,  the  means -ends  dimension  involves  the  conflict 
between  emphasis  on  getting  a  job  done  and  emphasis  on  the  process  of 
achievement.  It  is  in  this  area  that  the  question  of  means  justifying  ends 
is  often  raised. 

Relation  to  Role  Conflict 

If  it  can  be  assumed  that  different  individuals  take  up  different 
positions  between  each  of  these  polarities,  and  if  it  is  true  that  incumbents 
of  institutionalized  leadership  positions  are  particularly  vulnerable  to 
such  conflicts,  then  principals  of  schools  should,  experience  conflicting 
expectations  within  each  of  these  dimensions.  Not  only  might  one  expect 
there  to  be  a  lack  of  consensus  among  the  groups  with  which  a  principal 
works,  but  also  a,  lack  of  consensus  within  the  group  of  teachers  in  his  own 
school.  This  is  a  particularly  important  type  of  conflict  for  it  is  one 
with  which  he  must  contend  each  and  every  day.  Being  aware  of  this  conflict 
and  being  able  to  cope  with  it  may  be  central  to  the  effect  that  he  has  on 
people  in  the  institution  and  the  extent  to  which  the  institution  achieves 
its  objectives. 

The  usefulness  of  these  dimensions  for  the  analysis  of  role  conflict 
in  principal  leader  behavior  depends  upon  the  extent  to  which  it  is  possible 

to  speak  of  particular  teacher  expectations  in  each  of  these  dimensions. 


■ 


•  • 


■  •  ■ 

•-  ,  -  ■  ?  :  •-  •/  •  ■  ■  :■ 


.....  , . 


22 

The  following  is  a  description  of  the  kinds  of  expectations  which  might  be 
placed  in  each  of  these  areas; 

(1)  In  the  status  dimension  there  may  be  differences  in  the  expec¬ 
tations  of  teachers  concerning  the  extent  to  which  the  principal  should 
emphasize  his  rank  in  the  formal  structure.  At  the  extremes  some  may  expect 
him  to  adopt  the  behavior  of  one  who  has  achieved  a  position  of  superiority 
while  others  may  expect  him  to  consider  himself  on  an  equal  level  with  the 
teachers  in  his  school. 

(2)  Some  teachers  may  expect  that  they  should  be  allowed  to  conduct 
the  affairs  of  their  classroom  with  a  minimum  amount  of  interference  on  the 
part  of  the  principal.  These  teachers  may  look  upon  the  principal  as  one 
who  facilitates  the  operation  of  the  school  and  their  classroom  and  who 
need  do  little  more  than  this.  Others  may  look  upon  the  principal  as  one 
who  should  assume  a  great  deal  of  responsibility  for  what  goes  on  in  the 
school  and  even  in  their  particular  classrooms.  In  adopting  this  point  of 
view  they  show  their  willingness  to  accept  dependence  upon  the  principal 
rather  than  to  strive  for  independence.  If  this  situation  does  exist,  then 
there  will  be  conflicting  expectations  for  the  behavior  of  the  principal  in 
the  authority  dimension. 

(3)  There  may  also  be  disagreement  among  teachers  about  the  extent 
to  which  a  principal  should  look  upon  teachers  as  members  of  a  class  rather 
than  as  individuals.  Some  teachers  may  expect  particularistic  treatment 
with  special  attention  being  paid  to  their  personal  requests  for  special 
considerations.  Others  may  be  more  in  favor  of  a  universalis  tic  approach 
on  the  part  of  the  principal;  these  teachers  would  not  agree  that  some 


.  t  •  if  :f{'' 

•  • :  V  '  v  ' 


'  ,'i  "»  '  '  .fo-l 

-  .  •  ■  ■  '  ■  •.  '.ft!  '  " 

•  ■  '  ,  ' 

.  .  ... 

■  \  ' 

. 

* 

■ 


23 


members  of  the  staff  should  be  given  special  considerations  when  duties 
are  being  assigned  or  at  any  other  time.  This  would  be  indicative  of 
conflict  in  the  personal  dimension. 

(4)  If  there  is  conflict  in  the  means -ends  dimension,  it  may  center 
around  differences  in  opinions  concerning  the  extent  to  which  a  principal 
should  focus  attention  on  the  form  of  procedures  rather  than  the  purposes  of 
those  procedures.  At  one  extreme  might  be  the  expectation  that  standardized 
procedures  should  be  followed  regardless  of  whether  they  are  really  necessary 
or  suitable;  at  the  other  extreme  may  be  the  expectation  that  there  should  be 
more  emphasis  on  purposes  and  the  adaptation  of  procedures  according  to 
particular  circumstances . 

It  would  appear  that  these  areas  do  form  useful  categories  to  the 
extent  that  it  is  possible  to  state  that  certain  expectations  fit  into  one 
rather  than  another.  The  framework  would  be  much  more  useful  if  it  were 
demonstrated  that  it  was  possible  to  derive  a  quantitative  variable  with  which 
to  characterize  an  individual's  expectations  in  each  dimension  and  which  would 
enable  one  to  predict  with  some  certainty  what  the  individual's  expectations 
were  on  the  range  of  items  in  that  area.  This  would  be  the  case  if  it  were 
possible  to  construct  uni dimensional  scales  for  each  of  these  areas  of  expec¬ 
tations.  This  was  a  major  aspect  of  the  empirical  problem  with  which  the 
study  outlined  herein  concerned  itself. 

IV.  LEADER  AMBIVALENCE 

The  concept  of  leader  ambivalence  which  is  discussed  in  this  section 
is  closely  related  to  the  intensity  of  attitudes;  that  is,  to  the  intensity 


.  '  •  '  '•  '-e 

- 

■ 

„ 

■ 

■  :•  i ■  ■ 

' 

1 


.« 

■ 


■  '  ' 


24 

with  which  the  leader  holds  his  self-expectations.  For  this  reason  it 
is  necessary  to  elaborate  upon  the  intensity  component  of  attitudes. 

Attitude  Intensity 

Most  of  the  research  on  attitudes  of  various  kinds  seeks  only  to 
establish  an  ordering  of  people  along  a  content  continuum  ranging  from 
favorable  to  unfavorable.  A  question  which  might  be  asked  concerns  the 
position  of  the  zero  point  where  the  change  from  favorable  to  unfavorable 
opinion  takes  place.  In  trying  to  establish  an  answer  to  this  problem, 
Guttman  (8,  9)  has  proposed  both  argumentatively  and  mathematically  that 
the  solution  lies  with  the  intensity  component  of  the  attitude.  The 
intensity  of  the  attitude  refers  to  the  fact  that  people  differ  not  only 
in  their  position  along  the  content  continuum  but  also  differ  in  the  in¬ 
tensity  with  which  they  hold  their  opinions. 

The  relationship  which  might  be  expected  between  the  content  and  the 
intensity  of  the  attitude  would  not  be  linear.  It  might  be  argued  that  as 
we  move  along  the  attitude  continuum  from  unfavorable  to  favorable,  we 
would  find  that  those  who  are  at  the  extremes  would  tend  to  hold  their 
attitudes  or  opinions  more  intensely  than  would  those  who  are  somewhere 
mear  the  middle.  That  is,  a  U-or  J-shaped  curve  could  be  expected  if 
intensity  is  plotted  against  content  (8,  p.  263).  Empirical  studies  have 
tended  to  substantiate  the  theoretical  and  mathematical  derivations.  The 
bend  in  the  intensity  curve  indicates  the  region  on  the  content  scale  in 
which  there  is  a  change  from  "favorable"  to  "unfavorable"  attitudes  or 


opinions. 


.  ;  ..  ‘  :•  i:  ;;  :  r.  • :  ■ 

■ 

- 

■ 

' 

' 

•  .  ■  '  '  "  ’’ 

■ 

■ 

•  „  ,  .  ;  • 

...  •  ' 

-|  "  r  ;  .. 

. 


25 


In  general,  two  techniques  have  been  used  for  determining  the 
intensity  with  which  attitudes  are  held:  (l)  the  fold-over  technique,  and 
(2)  the  two-part  technique.  When  the  fold-over  technique  is  used,  the 
content  questions  are  rescored  in  such  a  way  that  responses  a.t  the  extremes 
are  given  the  same  weight,  the  adjacent  categories  given  an  identical  lower 
weight,  and  so  on.  Thus,  in  a  question  which  has  five  response  alternatives 
the  SA  and  SD  might  be  scored  two ,  the  A  and  D  might  be  scored  one  and  the 
U  would  probably  be  scored  zero . ^  That  is,  the  original  response  categories 
are  in  fact  folded  over  and  scored.  The  two-part  technique  involves  asking 
a  question  about  intensity  after  each  content  question  such  as,  "How  strongly 
do  you  feel  about  this?"  with  answer  categories  ranging  from  "Very  Strongly" 
to  "Hot  so  Strongly".  This  series  of  questions  forms  the  basis  for  ranking 
individuals  according  to  intensity.  It  is  the  opinion  of  Guttman  that  the 
latter  technique  is  theoretically  more  justifiable  than  the  former  (8,  p.  262). 

Torgerson  ( 18 )  and  Green  (6)  both  suggest  that  the  two-part  question 
be  used  rather  than  the  fold-over  technique.  Green  in  particular  feels  that 
any  justification  for  accepting  the  low  point  on  the  intensity  curve  as 
indicating  the  zero-point  must  be  made  on  empirical  rather  than  theoretical 
grounds  (6,  p.  558)*  This,  however,  does  not  negate  the  value  of  exploring 
the  intensity  component  for  other  purposes.  Mehling  (12)  has  suggested  a 
simple  means  for  determining  the  zero  point  when  a  semantic  differential 
scale  is  used;  he  obtained  a  U-shaped  curve  using  a  ridiculously  small 
sample  of  students.  The  number  was  ridiculous  in  the  face  of  the  several 
thousands  which  Suchman  used  in  some  of  his  analyses  (l7>  PP*  225,  230). 

1 

Symbols  represent  Strongly  Agree,  Agree ,  Uncertain,  Disagree, 
and  Strongly  Disagree. 


,  -  -  '  5 

^  •••'  '• 

.  '  .  d  ■  n  .  . 

c  i  •  .  :  ■  ■  •  •  -  ;  :- 

iBi  -  z  - 

/r*i  •  *  :  -  • 

i:  -  c  .  ■  .  '  -  - .  -  '  ■  ' 

.  ;  f  ■  '  : 

.•  ..  --  0"  -  ' 

.  •  -  noxd-  d  ;  c  ■ :  ' .  ■  •  -  ... 


. 


.d  ■  ’.o  ...  ..  .  J:  : \  d  ;  >C  .  ...  •  yjy  >. 


^  •  j:  x  !  i'  ii’X  .  x 


. 


oi 


‘  ,!  '  Y  ,.  ...  '.xiJ  .  "  1  .•  ^  1 

:  ,  ■  x  .  ^  rpxn'/loi  ■  v  .  •  •  .  -  "  d  1 

,  .  '  .  .  •  rx  xx  ;  ■'  '  ' 

■  Xd'  ■  .. 

y  .  ?  •  ,  c 


.  .  ,(  J  c  o  •!'.  .. 

cx  '  •  d' 

’  x  ■  <  ••  ■  '  ■  x 

c  '■  J'  :.  X 

,  .  .  .  1 


Generalized  Intensity 


26 


Suchraan  discusses  the  errors  associated  with  the  measurement  of 
intensity,  and  gives  examples  which  suggest  that  different  individuals 
may  attach  different  meanings  to  the  responses  (17,  p.  253)*  Thus,  an 
individual  may  indicate  in  the  second  question  that  he  feels  very  strongly 
about  an  attitude  question  after  having  responded  that  he  was  undecided 
about  content.  This  may  be  an  indication  that  the  question  is  very 
important  to  the  respondent.  Suchman  also  suggests  that  there  may  be 
differences  because  people  have  different  verbal  habits,  and  that  some  will 
state  that  they  feel  very  strongly  about  something  when  they  are  only 
slightly  favorable  to  it.  The  fact  that  people  may  differ  in  their  verbal 
habits  introduces  errors  into  the  intensity  curve.  When  the  sample  in  an 
empirical  study  (17,  p.  257)  was  divided  into  three  groups  of  high,  medium, 
and  low  generalized  intensity,  three  U-shaped  curves  with  the  same  general 
shape  but  at  different  levels  were  obtained.  The  curve  for  those  of  high 
generalized  intensity  was  in  the  upper  intensity  ranks  while  that  for  the 
lower  generalized  intensity  group  was  in  the  lower  intensity  ranks. 

In  considering  this  same  problem  Brim  (l)  suggests  that  this  gener¬ 
alized  intensity  may  be  more  than  just  verbal  habit.  It  is  his  opinion 
that  the  differences  in  responses  are  not  just  verbal  habits  but,  instead, 
that  these  indicate  different  degrees  of  motivation  to  escape  from 
uncertainty.  In  a  study  of  the  relationship  between  content  and  intensity 
in  probability  expectations.  Brim  theorized  in  the  following  manner:  All 
human  beings  have  a  need  for  security  or  certainty.  In  an  effort  to  achieve 
this  security,  there  is  a  tendency  to  deny  the  ambiguity  of  the  situation 


by  indicating  that  a  person  really  knows  what  the  outcomes  of  a  situation 


•  •  '  •  ■ 

X  i '  ■  X.  Jt  .  '  •  ' 

' 

....  '  ' 

•  .  •  '  .t-  '  '  ’ 

. 

...  ..  ..  .  •  "  ' 
■ 

*•  , 

- 

-L  « 


27 


will  be.  If  it  can  be  assumed  that  individuals  differ  in  the  need  for 
security,  this  may  express  itself  in  differences  in  tolerance  of 
ambiguity.  It  may  very  well  be  that  these  differences  in  the  need 
for  certainty  are  associated  with  the  concept  of  generalized  intensity. 

That  is,  those  individuals  who  have  a  high  need  for  certainty  would 
express  this  by  indicating  that  they  feel  very  strongly  about  various 
content  questions.  The  study  reported  herein  does  attach  a.  similar 
meaning  to  generalized  intensity  by  referring  to  it  as  ambivalence 
and  defining  it  as  is  described  in  the  next  section. 

Ambivalence 

In  this  study  ambivalence  was  defined  in  terms  of  the  intensity 
with  which  a  role  incumbent  holds  expectations  for  his  role.  A  prin¬ 
cipal  defined  as  being  high  in  ambivalence  was  one  who  did  not  hold  his 
self-expectations  very  intensely.  On  the  other  hand,  one  who  was  low 
in  ambivalence  held  his  self-expectations  with  much  higher  intensity. 

This  means  that  there  would  be,  in  terms  of  the  usage  of  these  words, 
an  inverse  relationship  between  ambivalence  and  generalized  intensity. 

It  was  theorized  that  a  relationship  would  exist  between  the  ambi¬ 
valence  of  the  principal  as  a.  leader  and  the  way  in  which  his  staff 
described  his  behavior. 

In  view  of  the  fact  that  there  are  conflicting  expectations  for  the 
behavior  of  the  principal  as  a  leader,  it  seems  reasonable  to  suggest  that 
a  principal  should  not  hold  his  self -expectations  too  intensely  but  must  be 
ready  to  adjust  to  various  situations.  On  the  other  hand,  it  might  not  be 
desirable  from  the  point  of  effectiveness  to  hold  self-expectations  too 


.. 


. 


c  •  Yli: 


' 

.  ■-  v 


»  ■  .  V 

.  X  '  .  --  - 


u 


: 


j; 


- 


X 


x  ; 


- 


•  J: 


28 


loosely,  that  is  to  he  too  ambivalent  for  the  principal  must  act  even  in 
the  face  of  conflicting  expectations.  Such  a  theory  suggests  a  curvilinear 
relationship  between  leader  ambivalence  and  effectiveness.  Although  this 
study  did  not  include  measures  of  leader  effectiveness,  it  did  involve 
measures  of  the  leader  behavior  of  principals  which  might  be  related  to 
overall  effectiveness. 


V.  LEADER  BEHAVIOR 

The  early  studies  in  leadership  focused  on  various  traits  or  qualities 
of  the  leader  in  an  effort  to  determine  differences  between  leaders  and  non¬ 
leaders  or  between  effective  and  non-effective  leaders.  A  survey  of  the 
relevant  research  by  Stogdill  (15)  caused  him  to  conclude  that: 

A  person  does  not  become  a  leader  by  virtue  of  some  combination 
of  traits,  but  the  pattern  of  personal  characteristic  of  the  leader 
must  bear  some  relationship  to  the  characteristics,  activities,  ana 
goals  of  the  followers.  (15,  p.  64) 

If  some  generalizations  are  made  from  Stogdill's  work,  it  is  usually  to 
suggest  that  leaders  tend  to  excel  in  intelligence,  scholarship,  depend¬ 
ability,  and  a  few  other  characteristics.  Because  of  the  failure  to  be 
supported  by  empirical  research,  the  so-called  traits  approach  has  been 
generally  abandoned. 

As  is  not  infrequent  in  such  circumstances,  the  first  stage  of  a 
new  theory  or  theories  is  to  move  too  far  in  the  other  direction.  For  this 
reason  the  extreme  situationist  view  which  looked  upon  the  situation  as 
the  sole  determinant  of  who  will  be  a  leader  did  not  prove  to  be  useful. 

More  recently  the  3 tructural -functional  approach  has  focused  attention 


~  ' ' 


■  '  !  '  ■ 

' 

;  .  '  '  £ .  j 


«  "■ 


'  - 

.  ■ ;  ;  r'-'i 

■ 

■- '  '  ■  • 

. 

.  ' 

j  '  '  "  ■ 

. 

'  ’ 

,  1  .  '  .  •  '  ‘  ' 

.  ■  ■  ,  i  ’  •  • 

' 


29 


upon  what  it  is  that  leaders  do  which  affect  the  operation  of  the  group. 

As  a  result  of  this  line  of  thinking  a  number  of  dimensions  have  been 
proposed  for  describing  the  behavior  of  a  leader.  Some  of  these  are  more 
applicable  to  leaders  in  informal  groups  while  others  are  more  useful  in 
describing  the  behavior  of  leaders  in  formal  organizations .  As  an  example 
of  the  former,  Hemphill  (ll,  pp.  51-35)  has  identified  fifteen  group 
dimensions  by  which  one  might  measure  the  influence  of  the  leader  on  the 
group.  These  include  such  things  as  size,  viscidity,  permeability,  and  so  on. 

The  Ohio  Leadership  Studies  have  resulted  in  the  development  of  an 
instrument  for  measuring  Leader  behavior  in  formal  organizations  on  two 
dimensions.  These  are  referred  to  as  Initiating  Structure  and  Consider¬ 
ation  and  are  defined  as  follows: 

Initiating  Structure  refers  to  the  leader’s  behavior  in 
delineating  the  relationship  between  himself  and  members  of  the 
work-group,  and  in  endeavoring  to  establish  well-defined  patterns 
of  organization,  channels  of  communication,  and  methods  of  procedure. 
Consideration  refers  to  behavior  indicative  of  friendship,  mutual 
trust,  respect,  and  warmth  in  the  relationship  between  the  leader 
and  members  of  his  staff.  (10,  p.  4) 

That  is,  the  first  mentioned  places  emphasis  on  getting  the  work  done  or 
goal  accomplishment  while  the  other  emphasizes  the  human  relations  or  group 
maintenance  aspect  of  leader  behavior.  In  research  using  the  Leader 
Behavior  Description  Questionnaire  four  styles  of  leader  behavior  have 
usually  been  discussed.  If  the  scores  for  the  group  being  studied  are 
divided  at  the  mean  or  the  median  for  both  dimensions,  it  is  possible  to 
describe  the  leaders  as  being  either  high  or  low  on  both  dimensions,  or  as 
being  low  on  one  and  high  on  the  other.  In  general  it  has  been  theorized 
that  the  most  effective  style  of  leader  behavior  is  that  characterized  by 


' 


■  ■  :r 


'• 


' 


;  •  ■  ■  ■  •  '  .  •  r  ■  ■ 

* 

•  ■  '  :L  ' . 

1  - 

- 

,  ■  •  . 

- 

- 

'  ■  ‘  ’•  ’  . 

■ 


30 


high  scores  on  both  dimensions. 

VI.  RESEARCH  HYPOTHESES 

The  general  theoretical  framework  of  the  study  has  been  presented 
in  preceding  sections  of  this  chapter.  The  main  ideas  which  have  been 
developed  will  be  summarized  here  through  definitions  of  the  terms 
which  have  been  used  as  well  as  definitions  of  some  new  ones.  A  state¬ 
ment  of  the  research  hypotheses  will  also  serve  to  organize  and  to 
summarize  the  preceding  sections. 

Definitions  of  Terms 

Role  Conflict.  In  this  study  role  conflict  refers  solely  to  the 
exposure  of  an  individual  to  incompatible  behavioral  expectations.  The 
conflict  with  which  the  study  is  most  concerned  is  that  which  stems  from 
the  lack  of  agreement  held  for  the  behavior  of  a  principal  by  the 
teachers  in  his  school. 

Ambivalence .  Ambivalence  refers  to  the  generalized  intensity 
associated  with  the  expectations  held  by  a  role  incumbent  for  his  role. 

The  more  intensely  a  role  incumbent  holds  his  self-expec tations ,  the 
lower  he  is  in  ambivalence. 

Teacher -Principal  Agreement  or  Agreement.  This  is  an  index  of 
the  degree  of  similarity  between  the  expectations  which  the  teachers  in 
a  school  hold  for  the  behavior  of  a  principal  and  the  expectations  which 
he  holds  for  himself. 

Consensus.  Consensus  refers  to  the  degree  of  similarity  among  the 
expectations  held  for  the  role  of  a.  principal  by  the  teachers  in  a.  school. 


->/_ .  '  .  X.0  1 


,  •  '  '  O 


■  1 


...  S  .  .  x  '  ~  •  '  ■'  J  oixN  •  ■  f  I 

too  ev  Ox  -  j oLx  i :  ■  =  ■ t .  x r  ■  o  . :  ox  •  o  a  c  e  a  ■  a 

■ j-  ©j  .j-  lo  dxj  ircxlsi  ,  -  V  f)9&:xtt!5©i  IIxw  qoIeveJb 

„  ■  . xcoixi  ■  x ... .  j  o  fir;  -  •  i  .9  ■  ‘  .  rv  .'  ci 

A/  jJJXB  9SX  '  '  ■  '  .  J  I.  A.  €  A  A  .  C  '  '  Z  '  A  t  t@ 

.Xu  O'  ■'  X  •  C  X(  •  ’ j  a;  A'.; 


-AO'  X 


■ ill  I' 

A  CA." . .  C  I  '  A  -u  A.:.  -  I  AAA1  .  A  I \  ; 

xd  ,  a  s  a  >xv®'/{9<  J  ia  qmoonl  #  Ib  ivi  -  i  to  ox 
j  .  ci  .  '  A  i  c  .  c  c  J  •  x  ax  '  '  a. 

i  '  t  •  -~r  o  a  ,vx 'I"  j 


(  .  d 


.o- 


A  r  ■  ::  >  x  ;xv  .  ;  ,  o 

,  1  j  c  i  a:  •  x  c  ■ .  J 

OdL  i  1  a  .  i  o'  a  ' :  o  I  vx  .  •  o  >vx 

.  ■  o  x  s .  0 ■  1  )±I  lo  X  ’  '  '  a':a  ' 

„  cox  i  -  x  .  ■  .  c  i  i 

.  c  '  . 

-  "  -  A  .  .  . 

xa  ,i;  •  L-  :±  ,x-  Ic  oaxaaIa  aIa 

y.'iv  ■ 1  ■  ,;d-  x  ’a  '  I<  ;  loorto  .  • 

„ '  x  •  ’  y  J  I 'Oi  l  o . ' 


-  A  . 


..A  O'.  A  X  A  x  .KOACrX.a 


XI  XI 


)  y 


a  ... 


«... 


XA  d  C  1 


c 


31 


Lea.der  Behavior .  A  dimension  of  leader  behavior  refers  to  either 
Initiating  Structure  or  Consideration  measures  obtained  through  the  use 
of  the  Leader  Behavior  Description  Questionnaire.  Style  of  leader 
behavior  refers  to  the  combination  of  high  and  low  scores  on  the  two 
leader  behavior  dimensions. 

Uni dimensional  Scale .  A  series  of  items  will  be  considered  to  form 
a  unidimensional  scale  if  they  closely  approximate  the  Guttman  model  (8). 

The  theory  of  unidimensional  scales  will  be  developed  in  succeeding  chapters. 

Scale  Analysis.  Scale  Analysis  or  scalogram  analysis  refers  to  the 
procedures  used  to  determine  whether  or  not  a  series  of  items  closely 
approximates  the  Guttman  model. 

Dimension  of  Conflicting  Expectations.  A  series  of  items  of 
expectations  for  the  behavior  of  a  principal  will  be  considered  to  form 
a  dimension  of  conflicting  expectations  if  these  closely  approximate  the 
Guttman  model. 

Scale  Analysis 

In  this  study  a  series  of  items  of  expectations  will  be  tested  for 
unidimensionality  by  moans  of  scale  analysis.  The  role  conflict  dimensions 
proposed  by  Seeman  will  be  used  ,  as  a  ba,sis  for  constructing  categories  of 
items;  each  series  of  items  will  then  be  subjected  to  scale  analysis  as  a 
test  for  uni dimensionality.  The  full  import  of  the  presence  of  a  uni¬ 
dimensional  scale  will  become  apparent  after  the  theory  of  scale  analysis 
has  been  developed  in  later  chapters. 

The  scale  analysis  procedures  will  be  used  as  a  basis  for 


J  x  ■ 


o ' :  J' 


r 

J  — 


"  •.) 


'  .  •  !  x.'  •  X  -OC  C  •  j*  xd  -i'  ±  X 

.  .'.'.'’O'1  i Xu  ■ ' ) 

c  $  J  -’v  I 

x  i.  ■  '  J:  '  v-  ■  '  ■  X 


j  •  v-  o  XXx  xx'x  'J'.-'  i: 

j:”  •  ; 

j-  ;iio  xJbcooo  x  >sqoI  ve  5  IIx  £ 

o  o  ■  x  '  - 


» ■  x  x. :  _ 

o  x  c.  1  .  XJX  x : 


x  ‘  i  '  To  T  x  ;'d 

}  0  -  X  -  '  .  '  v 


!+  ,  >  ' 


Is  Dio  &i  Xoxi  *io  lerld  X  -  x  ■  ■  x  c  ■ 

_  i  ■  x .'J'1  ■  (  ■ 

1  „  .  j.\v  .  o  ;  x.  >_  .  •>  --  _ 

X  bx  DO  t  -  H  -  '  -O  X  (  £  -  o  -  1  X.'  m 

jj-  yj  ■  (  -  Xo  d  tx  x  o  x#oxil  o  .  .  t  >x  tear! 

„  .  i  ,  •  i  ■  J  .  •! 


■OC 


..X  ,'x  xc  '  x,  '  Xx  Xo  i  o  :  '■  "  T  •'  x.x  '.J  i  !. 

ol  x  x  .  X  J  •  x.’  O’  lo  ■  i  '  '  d-x.'  x  x  x  x  x  v 

j  c  j  g  d  g  -  x  1  ;  <  1  *  ■' 

.  o  <  O’.  •  ■  X.  X  . i'X  ■  XX  Xo"X  •  : 

c  x  -  -  ti  x  x  >x  •  .  i 

c  •  x  o'  ■  d"  d  ■  ■  ■  ■  oo  (  ’  XX  .  -  O'  /  •  .. 

_dv  O' "  d*  -  x  t  :  < 

x  !  x  •  •  m  o  >  <  ••  X!  ,1  '  x  •  X  ■  o  Oj  !  j 


32 


quantifying  the  expectations  which  teachers  hold  for  the  behavior  of  a 
principal.  In  the  same  way  the  ambivalence  of  principals  will  also  be 
quantified.  Prom  the  quantitative  variables  mentioned  formerly  will  be 
developed  indices  of  consensus  and  teacher-principal  agreement  within  any 
one  school  so  that  it  will  be  possible  to  test  the  hypothesis  listed  below. 


Hypotheses 

1.  Principal  ambivalence  will  be  related  to  the  way  in  which 
teachers  describe  the  leader  behavior  of  a  principal. 

1.1  There  will  be  a  positive  relationship  between  ambivalence 
and  the  Consideration  dimension. 

1.2  There  will  be  a  negative  relationship  between  ambivalence 
and  the  Initiating  Structure  dimension. 

1.3  Bach  of  the  relationships  above  will  be  curvilinear. 

2.  The  degree  of  teacher-principal  agreement  will  be  related  to  the 
way  in  which  teachers  describe  the  leader  behavior  of  a  principal. 

2.1  Agreement  will  be  positively  related  to  Consideration. 

2.2  Agreement  will  be  positively  related  to  Initiating  Structure. 

2.3  Agreement  will  be  higher  in  schools  where  the  principal  is 
high  on  both  dimensions  of  leader  behavior. 

3.  The  degree  of  consensus  in  schools  will  be  related  to  the  way 
in  which  teachers  describe  the  leader  behavior  of  a  principal. 

3.1  Consensus  will  be  positively  related  to  Consideration. 

3.2  Consensus  will  be  positively  related  to  Initiating  Structure. 

3*3  Consensus  will  be  higher  in  schools  where  the  principal  is 
high  on  both  dimensions  of  leader  behavior. 

4.  Ambivalence  will  be  related  to  the  degree  of  consensus  on 
expectations . 

4.1  The  relationship  between  ambivalence  and  consensus  will  be 
positive. 


' 

'•I  >CT  BX  "  :  0.  : 

- 

'■  ‘ 

. 

,  ■ 

„ 

, 

. 


4.2  The  relationship  between  ambivalence  and  consensus  will  be 
curvilinear. 

Ambivalence  will  be  related  to  the  degree  of  teacher- principal 
agreement . 

5.1  The  relationship  between  ambivalence  and  agreement  will 
be  positive. 

5.2  The  relationship  between  ambivalence  and  agreement  will 
be  curvilinear. 


' 

* 

. 


■■ 

. 


34 


REFERENCES  FOR  CHAPTER  II 


(1)  Brim,  Orville  G.,  Jr.  "Attitude  Content-Intensity  and 

Probability  Expectations,  "American  Sociological  Review, 

20:  68-76,  February,  1955* 

(2)  Fromm,  Erich.  Escape  from  Freedom.  New  York:  Farrar  and 

Rinehart ,  Inc . ,  1941 • 

(3)  Getzels,  Jacob  W.  "Administration  as  a  Social  Process,"  in 

Andrew  W.  Hal pin  (ed.).  Administrative  Theory  in  Education. 
Chicago:  Midwest  Administration  Center,  1958,  pp.  150-165 • 

(4)  Getzels,  Jacob  W.  and  E.  G.  Guba.  "Social  Behavior  and  Adminis¬ 

trative  Process,"  The  School  Review,  65:423-441?  Winter,  1957* 

(5)  Getzels,  Jacob  W.  and  Herbert  A.  Thelen.  "The  Classroom  Group 

as  a  Unique  Social  System,"  in  Henry  B.  Nelson  (ed.). 

The  Dynamics  of  Instructional  Groups .  Chicago:  The  National 
Society  for  the  Study  of  Education,  i960.  Pp.  53-82. 

(6)  Green,  Bert  F.  "Attitude  Measurement,"  in  Gardner  Lindzey  (ed.). 

Handbook  of  Social  Psychology,  Vol.  I.  Reading,  Massachusetts: 
Addison-Wesley  Publishing  Company,  Inc.,  1954*  Pp.  335-569* 

(7)  Gross,  Neal,  Ward  S.  Mason,  and  Alexander  McEachem.  Explorations 

in  Role  Analysis :  Studies  of  the  School  Superintendency  Role. 

New  York:  John  Wiley  and  Sons,  Inc.,  1958* 

(8)  Guttman,  Louis.  "The  Cornell  Technique  for  Scale  and  Intensity 

Analysis,"  Educational  and  Psychological  Measurement,  7:247-279? 
Summer,  1947* 

(9)  Guttman,  Louis.  "The  Principal  Components  of  Scalable  Attitudes," 

in  Paul  F.  Lazarsfeld  (ed.).  Mathematical  Thinking  in  the  Social 
Sciences.  Glencoe,  Illinois:  The  Fhee  Press,  1954*  Pp*  216-257* 

(10)  Hal pin,  Andrew  W.  The  Leadership  Behavior  of  School  Superintendents . 

Second  Edition,  Chicago:  Midwest  Administration  Center,  1959* 

(11)  Hemphill,  John  K.  Situational  Factors  in  Leadership.  Columbus, 

Ohio:  The  Ohio  State  University,  1949* 


'  •  ■  i:  ]■ !  . 

‘ 


... 


?  * 


. 


! 


.  . 

.  ,  ' 

.  . 


.  ,  „  .  * 

•  •  .  ■  •  ■  ,  '•  •■■■ 

3  .  , 
.  ,  '  ■ 

. 


55 


(12)  Mehling,  Reuben.  "A  Simple  Test  for  Measuring  Intensity  of 

Attitudes,"  Public  Opinion  Quarterly,  23:576-578,  Winter,  1959* 

(13)  Myrdal,  Gumar,  Richard  Sterner,  and  Arnold  Rose.  An  American 

Dilemma:  The  Negro  Problem  and  Modem  Democracy.  Vol.  II. 

New  York:  Harper  and  Brothers,  1944* 

(14)  Seeman,  Melvin.  Social  Status  and  Leadership.  Columbus, 

Ohio:  The  Ohio  State  University,  I960. 

(15)  Stogdill,  Ralph  M.  "Personal  Factors  Associated  with  Leadership: 

A  Survey  of  the  Literature,"  Journal  of  Psychology,  25 : 35—71 , 
January,  1948. 

(16)  Stouffer,  Samuel  A.,  et  al.  The  American  Soldier:  Adjustment 

During  Army  Life.  ToT7~I  of  Studies  in  Social  Psychology 
in  World  War  II.  4  vols.  Princeton,  New  Jersey:  Princeton 
University  Press,  1949* 

(17)  Suchman,  Edward  A.  "The  Intensity  Component  in  Attitude  and 

Opinion  Research,"  in  Samuel  A.  Stouffer,  et  al.  Measurement 
and  Prediction.  Vol.  IV  of  Studies  in  Social  Psychology  in 
World  War  II.  4  vols.  Princeton,  New  Jersey:  Princeton 
University  Press,  1950*  Chapter  7>  PP*  213-276. 

(18)  Torgerson,  Warren  S.  Theory  and  Methods  of  Scaling. 

New  York:  John  Wiley  and  Sons,  Inc.,  1958. 

(19)  Williams,  Robin  M. ,  Jr.  American  Society:  A  Sociological 

Interpretation.  New  York:  Alfred  A.  Knopf,  1951* 


.  ■  '  . 

* 

■»  » 

- 

. 

»  .  '  '  ' 

.  '  .  '  "  ■  • 
,  ,  ■  •  ■ 

,, 

.  , 

■  , 

■ 

, 

-  -  ■ 

.  . 

-  '  1 

- 


CHAPTER  III 


RELATED  RESEARCH 

Although  much  has  heen  written  about  role  conflict,  there  is  in 
reality  very  little  research  evidence  other  than  that  this  is  a.  problem  in 
a  number  of  areas.  Some  examples  of  research  on  this  particular  problem 
and  related  problems  are  reviewed  in  this  chapter.  The  research  evidence 
presented  here  builds  directly  upon  the  theoretical  bares  of  the  study  as 
outlined  in  the  previous  chapter. 

I.  RESEARCH  OH  ROLE  CONFLICT 

The  research  reviewed  in  this  section  is  divided  into  two  main  parts: 
that  carried  out  in  the  field  of  education  and  that  carried  out  in  other 
areas.  In  order  to  provide  a  perspective  for  the  research  in  education, 
research  in  other  fields  is  reviewed  first. 

Hon-Educ ation  Role  Conflict  Studies 

Most  of  the  research  in  these  other  fields  concerns  conflict  resulting 
from  the  fact  that  essentially  different  expectations  can  be  held  for  the 
occupant  of  a  particular  position.  As  an  example  of  this  Burchard  (l)  studied 
the  role  conflicts  in  military  chaplains.  He  saw  a  possibility  of  conflict 
between  the  military  and  the  religious  ideology  and  hypothesized  that  the 
chaplain  seeks  to  reconcile  the  two  either  throiigh  rationalization  or  through 
compartmentaliz ation  of  role  behaviors.  Furthermore,  he  hypothesized  that  if 
reconciliation  were  through  rationalization,  this  would  tend  to  strengthen 


;  •  i&j:  y.  ;  ■  a  \  v,  -  • 

■  .  x ■  ■ ' 

.  •  .  ~L  . . 

-  ;  •  ■  .  .  x  •  ' 

;  Id-  d  x  osn:i  lx  ;  :  - 


'  d-  i  d  ’  iv  c 

X-  -  '  ,■  X  X  .  O  'X  '.tO  ■  X.Xv 


d'  .  ..":."x 


:  ]  ,  :  0.  .  • 


x  .  .  ;  ..  xvx  xdx  x  x  ".x  .  :  xv 

±  d-1  ■  x  X  V  r  %  .cl  x  f  ■  '  ;  "  " 

«  c  ;  vxd  xv  i  ■  - 

•.  .  J/..  ■  d  •  x  t  •  • 


•  X  •  X.  <■  •  X- 

;  t  ;  :  X  lie  x  1  d  ) 

C  dx  :  :  x  •  .  .xx-  .  .  &  :  x  &c  ':  c  x  -  1 

d  .  ±d‘x  •  x 

.  ifJi  .  -  ;  •  x  '  . '  X 

,  '  ■  x  :  •  J-  x.:x  >'  ■  : 

x  ■  •  x.‘;  x  .  ■  .  xo.  c  x 


<  ex 


„  .xx  ■  . 

.  x.  xd  •  ;' 


fJCll  x 


37 

the  role  as  military  officer.  The  data  which  he  collected  tended  to 
support  the  hypotheses,  hut  he  does  suggest  two  hypotheses  for  further 
testing.  These  are  (l)  that  the  role  which  provides  the  primary  identi¬ 
fication  for  the  individual  takes  precedence,  and  (2)  that  for  military 
chaplains  the  primary  identification  is  that  of  military  officer.  Burchard 
suggests  that  from  a  theoretical  point  of  view  the  military  chaplaincy  would 
seem  to  he  possible  of  fruitful  analysis. 

The  problem  of  conflict  between  the  expectations  held  for  an 
individual  by  the  external  and  internal  systems  in  an.  organization  was 
studied  by  Wisp£  (30)  who  carried  out  a  sociometric  survey  of  choices  among 
forty- three  agents  in  an  insurance  company.  According  to  the  analysis,  the 
traits  which  made  for  success  as  an  insurance  salesman  precluded  acceptance 
as  a  friend.  In  other  words,  there  was  a  conflict  between  the  expected 
pattern  in  the  external  system  and  the  expected  pattern  in  the  internal 
system.  Although  this  conflict  may  have  been  due  to  the  special  circum¬ 
stances  within  this  organization,  it  does  bring  to  mind  the  circumstances 
with  which  every  schoolboy  is  familiar.  In  the  case  of  the  organization 
investigated,  one  might  have  supposed  that  the  amount  of  personal  identifi¬ 
cation  with  the  goals  of  the  organization  would  have  been  great  enough  to 
forestall  any  great  amount  of  conflict  between  the  two  systems. 

Perry  and  Wynne  (23)  studied  the  conflict  which  arose  between  the 
researcher  and  therapist  roles  in  a  research  hospital  which  accepted 
patients  only  if  their  condition  fitted  the  research  program  currently 
under  way.  During  the  course  of  the  treatment  there  was  occasion  for 
conflict  to  arise  between  the  doctor's  role  as  a  therapist  for  the 


' 


' 


v 


' 


’ 


■ 


38 


individual  and  the  role  of  a  researcher  within  the  organization.  In  this 
particular  instance  there  were  no  institutionalized  guides  for  resolving 
the  conflict,  and  so  the  individual  was  fairly  well  on  his  own.  It  was 
found  that  some  doctors  made  the  conflict  explicit  for  a  patient  while 
others  left  it  implicit,  and  in  general  two  main  ways  of  avoiding  or 
resolving  the  conflict  were  "being  adopted  by  different  doctors.  Some 
built  up  a  hierarchy  of  obligations  as  far  as  individual  patients  were 
concerned  and  dealt  with  one  patient  on  a  research  basis  and  with  another 
on  a  therapeutic  basis.  Another  means  of  avoiding  the  conflict  was  becoming 
institutionalized  in  the  segregation  of  roles;  some  doctors  became  more 
research  oriented  while  others  became  more  therapy  oriented.  This  was  in 
fact  a  division  of  labor  very  similar  to  that  between  a  superintendent  and 
a  principal  as  far  as  the  evaluation  of  teaching  is  concerned. 

Some  interest  in  measurement  has  also  been  shown.  One  aspect  of  this 
was  discussed  by  Motz  (21 )  who  proposed  an  inventory  for  studying  husband- 
wife  roles.  Prom  an  analysis  of  open-end  questionnaires  she  decided  that 
there  were  two  types  of  roles,  the  traditional  and  the  companionship,  and 
then  she  set  out  to  study  six  areas  in  each.  The  final  instrument  consisted 
of  twenty-four  statements  concerning  these  two  roles  arranged  in  random 
order.  The  score  was  simply  the  number  of  items  in  each  category  to  which 
an  individual  gave  a  positive  response.  Motz  states  that  she  did  not  want 
to  use  scales  because  she  wanted  to  study  inconsistency  and  thus  to  leave 
room  for  inconsistency.  This  would  not  appear  to  be  a  very  logical  approach 
for  no  doubt  if  these  are  viewed  as  somewhat  opposite  roles,  the  use  of  scales 


would  be  almost  essential  for  a  true  picture  of  the  role  relationships.  It 


.  •  '•  ' '  '  ’  ’ 

.  ...  ...  '  • 

- 

.  -  '•  - 

•;  •-  V  >  '  .V  -  ’  '  •  :  i  "  f  ' ' 

...  ;  ,r‘  ■■  -n 

’  '  ■  1  - 

■  ■•■t:  '  -  f  ''  -  ''  . 

'  •  '  • 

■ 

' 

’  ’ 

-  • 

...  ,  .  ’•  '  D 

' 

'■  •  ,  •  • .  uc  : .  :  1 

•  .  :-K. n;  £}rfa  oa  -• 

•  '•  ' 

•  ri 

.  - 

■'  /  ,  r  •  '  K-y 

39 


would  be  interesting  to  attempt  to  construct  scales  in  such  dimensions 
as  housework,  employment,  financial  support,  etc.  The  inconsistency 
could  be  determined  even  more  readily  if  it  could  be  shown  that  individuals 
expected  different  degrees  of  traditionalism  in  different  dimensions.  The 
methodology  suggested  by  Motz  is  indicative  of  the  item  by  item  approach 
that  has  been  prevalent  in  role  studies. 

Gullahorn  (7)  was  also  interested  in  studying  role  conflict  and  in 
particular  the  resolution  of  role  conflict.  He  presented  subjects  with  a 
hypothetical  role  conflict  situation  such  as  that  of  a  worker  who  must 
decide  between  accepting  a  position  as  union  chief  steward  or  a  responsible 
position  in  the  employer’s  club.  The  subject  must  decide  whether  to  act 
according  to  the  wishes  of  the  employer  or  of  the  fellow  workers.  The 
instrument  asked  the  subject  to  state  whether  he  would  accept  one  position, 
the  other,  or  whether  he  would  attempt  to  keep  both  under  a  variety  of 
combinations  of  external  pressures  from  employer  and  fellow  workers. 
Gullahorn  suggests  that  the  individual  feels  the  conflict  more  if  he  favors 
one  of  the  roles  while  reference  group  pressure  builds  up  in  favor  of  the 
other  role.  He  also  offers  some  research  evidence  to  support  the  notion 
that  as  the  pressure  builds  up  the  individual  tends  to  react  more  unreal¬ 
istically  and  tries  to  do  more  than  he  should  or  can. 

Dimensions  of  Conflict.  The  question  of  how  choice  is  made  in  role 
conflict  situations  has  been  studied  and  commented  upon  by  a  number  of 
investigators.  These  studies  are  closely  related  to  the  dimensions  of 
conflicting  expectations  mentioned  in  the  previous  chapter  in  that  the 

conflict  between  universalistic  and  particularistic  behavior  has  been 


,  .  •  .  ' 

,  '■  •  ■  .  c  '■  t;  i" 

.  1  '  '  i  : 

,  I  '  '  ■  ■  '  ’’  ’  '  ' :  '  ’ 


-r 

* 

.  • 


.  . 


40 


studied  a  number  of  times.  Sutcliffe  and  Haberman  (29)  suggested  that 
social  distance,  social  sanction,  and  publicity  interacted  to  determine 
universalistic  responses  but  overall  friendship,  privacy,  and  weak  sanctions 
tended  to  favor  particularistic  responses.  A  factor  analysis  revealed  that 
when  social  factors  were  well-balanced  there  was  a  personality  bias  to  respond 
either  universalistically  or  particularistically  in  role  conflict  situ¬ 
ations. 

The  results  of  the  study  carried  out  above  were  similar  to  those 
obtained  by  Stouffer  and  Toby  (27)  some  years  earlier.  These  investigators 
also  hypothesized  that  when  individuals  are  faced  with  conflicts  between 
obligations  to  a  friend  and  obligations  to  society  in  general  some  will 
tend  to  be  more  particulatistic  while  others  will  be  more  universalistic 
in  their  behavior.  They  concluded  from  their  study  that  it  is  possible  to 
classify  people  according  to  which  horn  of  the  dilemma  they  select.  One  of 
their  suggestions  of  interest  here  is  that  role  conflict  studies  be  carried 
out  in  the  analysis  of  various  leadership  positions. 

Laulicht  (18)  based  his  study  of  role  conflict  on  the  pattern  variable 
theory  (22)  which  suggests  that  there  are  five  dilemmas  of  choice  which  must 
be  resolved  before  an  individual  can  act  in  a  situation.  In  particular,  he 
studied  the  degree  to  which  different  individuals  were  consistent  in  the  way 
that  they  resolved  role  conflicts.  Laulicht  constructed  a  number  of  situ¬ 
ations  which  seemed  to  pose  one  or  the  other  of  the  pattern  variable 
dilemmas  under  study.  He  concluded  that  he  had  been  successful  in  obtaining 
scales  which  measured  dimensions  of  role  conflict  and  suggests  that  this 
might  be  a  fruitful  way  for  studying  the  role  of  the  leader  or  of  the 


bureaucrat. 


.  ,  r...-  ‘  ‘  '•  •  '  ‘  v 


„ 

‘ 

■ ■  -  t 

•  •  ,  r  , 

' 

1  i  t 

■ 

- 

.  !| 

-  '  '•  ■  '  ‘ 


41 

Variability  in  Expectations,  In  an  earlier  study  Stouffer  (26) 
investigated  the  amount  of  variability  in  expectations  which  was  present 
in  different  situations  when  role  conflict  was  defined  as  a  conflict 
between  roles  due  to  the  fact  that  the  same  individual  occupies  different 
positions  in  different  groups.  He  hypothesized  that  expectations  for  a 
particular  aspect  of  role  behavior  can  be  described  more  appropriately  in 
terms  of  a  range  rather  than  a  point.  In  other  words,  within  each  group 
of  role-definers  there  is  a  range  of  expectations  and  a  range  of  acceptable 
behavior  so  that  even  if  the  role  incumbent  does  not  act  in  accordance  with 
the  expectations  of  some  group  members,  there  will  still  be  other  indi¬ 
viduals  from  whose  point  of  view  there  has  been  little  or  no  departure 
from  expected  behavior.  According  to  Stouffer  this  variability  in  expec¬ 
tations  may  be  very  important  for  making  any  action  possible  because  it 
tends  to  weaken  the  sanctions  which  are  applied.  In  an  empirical  investi¬ 
gation  of  the  expectations  held  for  a  proctor  who  sees  a  student  cheating, 
the  hypothesis  of  the  existence  of  this  range  was  upheld.  It  seems  evident 
that  the  existence  of  a  range  is  functional  only  up  to  a  point,  and  that 
beyond  this  point  the  slippage  would  be  so  great  as  to  lead  to  a  degree  of 
di s  o  rgani z  at i on . 

The  expectations  which  group  members  hold  for  the  role  of  the  leader 
and  the  relationship  of  this  to  other  group  characteristics  was  investigated 
by  Hall  (10).  One  of  his  hypotheses  was  that  the  greater  the  cohesiveness 
in  an  aircraft  crew,  the  greater  would  be  the  agreement  on  role  prescriptions. 
Furthermore,  it  was  hypothesized  that  the  extent  to  which  both  the  commander 
and  the  group  were  motivated  to  remain  together,  the  closer  would  role 


/ 

' 

» 

- 

■ 

■ 

■  ,  .  r 

„ 

’•  ■')•••  ■  -  ' 

. 

.  ■  ••  4.,  -  v 


' 

.  '  '  1  n'i 

-  •  ■  ■  ■  ,  -I; 

,  •  ...P- 


42 

performance  be  to  actual  expectations.  In  general  these  hypotheses  were 
supported  on  a  dimension  defined  as  intimacy,  which  is  not  surprising  in 
view  of  its  close  relationship  to  cohesiveness,  but  the  same  hypotheses 
were  not  supported  on  other  dimensions.  Hall  suggests  that  the  reason 
for  this  was  that  norms  with  respect  to  these  other  dimensions  had  as  yet 
not  developed  within  the  group. 

Res  earch  in  Education 

Some  of  the  research  mentioned  in  the  introductory  chapter  served 
to  indicate  that  conflicting  expectations  were  a  fact  in  the  lives  of 
people  in  educational  positions  particularly  if  these  happen  to  be 
administrative  positions.  The  research  which  is  reviewed  in  this  section 
will  be  studies  which  attempted  to  probe  a  little  more  deeply  into  the 
problem  than  those  which  were  cited  previously. 

Role  Conflict  in  Teaching.  Getz els  and  Guba  (4)  investigated  to 
some  extent  the  nature  of  roles  and  role  conflict  in  the  teaching  situ¬ 
ation.  After  intensive  interviews  with  a  number  of  teachers  they  came 
to  the  conclusion  that  there  were  three  major  aspects  of  the  teacher’s 
role  which  were  characteristized  by  conflict:  (l)  the  socio-economic  role, 
(2)  the  citizen  role,  and  (3)  the  professional  role.  On  the  basis  of  these 
interviews  an  instrument  was  constructed  to  measure  the  extent  to  which 
conflicts  existed  in  different  situations  and  the  extent  to  which  teachers 
were  personally  troubled  by  the  conflicts.  An  analysis  of  the  data  led  to 
the  conclusion  that  there  were  some  expectations  which  were  situationally 
independent;  that  is,  the  conflicts  exist  everywhere,  but  there  were  also 


-  •_  ■  -  ••  '■ 


' 

„ 

■ .  yv  •  •  . 

'  '■  >  >:;■  ■  ; rx :  :  ■  ■  ''  n I:  < >cr 

.  '■ :  J  ’  '  ■ 

. 

, 

- 

■  . 

■  '  ■  ■  '■  '  .  '< 

'•  ’  '  <  '  ' 

.  '  '  •  ;  • 

,  •  '  '  '  , 

:ik>  •  • 

■  ■  >  • 


45 


other  conflicts  which  were  much  more  closely  related  to  specific  school 
situations . 

The  data  also  revealed  that  the  extent  to  which  teachers  are  troubled 
by  conflicting  expectations  seems  to  be  systematically  related  to  personal 
characteristics  of  teachers.  Some  teachers  are  extremely  troubled  by  role 
conflicts  while  others  in  the  same  situation  are  not  greatly  concerned  about 
.  them.  The  investigators  suggested  that  this  lack  of  perfect  integration  in 
the  role  of  the  teacher  may  result  in  ineffective  performance  by  the  teacher. 

Role  Conflict  and  Effectiveness.  The  relationship  between  role 
conflict  and  effectiveness  was  studied  more  specifically  in  another  research 
project  (3).  It  was  theorized  that  actually  two  variables  were  involved: 
the  extent  to  which  there  was  role  conflict  in  the  situation  and  the  extent 
to  which  there  was  role  conflict  in  the  individual.  Getzels  and  Guba 
hypothesized  that  the  extent  to  which  an  individual  recognized  the  existence 
of  conflicting  expectations  would  be  a  function  of  the  number  and  magnitude 
of  incompatible  expectations.  Furthermore,  the  greater  that  an  actor  became 
involved  in  the  conflict,  the  less  effective  he  would  be  in  at  least  one  of 
the  roles.  A  third  hypothesis  was  that  the  degree  to  which  different  actors 
became  involved  in  role  conflict  would  be  systematically  related  to  certain 
personal  and  attitudinal  characteristics.  When  the  latter  two  hypotheses 
were  tested  using  instructors  in  a  military  situation  they  were  supported 
in  general;  those  instructors  who  were  more  personally  involved  in  role 
conflict  also  tended  toward  ineffectiveness.  The  idea  of  personal  involve¬ 
ment  in  role  conflict  is  closely  related  to  the  idea  of  ambivalence  or  the 
intensity  with  which  self-expectations  are  held  by  a  role  incumbent  which 


’ 

. 

* 

. 


'  ' 

•  '  ' 

- 

. 


■ 

•  '• :  - 


44 


was  developed  in  the  previous  chapter. 

Typology  of  Conflicts .  Hencley  (15)  attempted  to  set  up  a 
typology  of  the  conflict  patterns  in  the  role  of  the  superintendent. 

The  general  approach  wa.s  through  attempting  to  determine  the  extent  to 
which  superintendents  as  a  group  could  anticipate  the  expectations  of 
other  groups  commonly  considered  to  be  reference  groups.  .Among  the 
typologies  suggested  were  such  categories  as  trouble-seeking,  innocent, 
overlooked  support,  keen,  and  reverse  polar.  These  correspond  to  con¬ 
flicts  which  were  anticipated  but  non-existent,  existed  but  were  not 
anticipated,  non-existent  and  not  recognized,  existed  and  anticipated, 
and  finally  completely  misjudged.  It  may  be  that  some  of  the  items  of 
expectations  were  of  value  in  studying  the  role  of  the  superintendent, 
but  such  a  typology  is  actually  of  very  little  value  either  for  the 
development  of  theory  or  the  improvement  of  practice. 

The  research  reported  by  Gross  et  al_  (6)  studied  the  role  of  the 
superintendent  particularly  from  the  viewpoint  of  analyzing  the  role 
conflicts.  The  investigation  revealed  that  on  about  two-thirds  of  the 
items  used  in  the  study  there  were  sizable  differences  between  superin¬ 
tendents  as  a  group  and  the  board  members  ac  a  group  on  the  expectations 
which  were  held  for  the  behavior  of  a  superintendent.  On  nearly  one-half 
of  the  total  items  the  differences  were  differences  in  direction  of 
expectations  and  not  just  intensity  (6,  p.  141).  It  was  shown  also  tha.t 
there  were  different  degrees  of  consensus  on  specific  items  within  each 
of  the  groups.  The  writers  suggested  that  one  reason  for  differences  in 
these  expectations  is  that  the  two  groups  occupy  different  positions 


• 

- 

'  y  ’  X 

.  *  r 

O!  X  ! 

X  i 

(  lev 

l.  ;V.r 

i  j 

o 

. 

-  ■ 

o': 

§  . 

«  \ 

*  .  J* 

-  .  ;  ■  . 

:  X  r-i 

X 

o  J 

o'  cx . 

.  .  0 

rf; 

^  ‘  vJ 

Xi 

€  X  : 

.  o' 

O.'  .  _ .  * .  1-'  .  V 

.  :'  X 

o 

r<  '  '  X 

L  "  ' 

'  c; 

■ 

Ox?.. 

t  .. 

o  . 

...  X  .  C 

I'Kv  Ml.' 

o 

I'.O'C'C 

;  . 

x  -  ) 

S  d  - 

,_X  'o 

o  .  ':  ■: 

o'..’ 

.  '  0 

’ 

. 

X 

i  r  o 

. 

- 

t.(  :  r. 

' 

0 

--  C 

-  - 

„  !■  j.: 

i:  . 

gX. 

0  0 

- 

, '  .  J  .. 

r  '  )X 

;  •  x 

.0 

)X  X 

.  < 

■  IOO  .  ' . o.  ' . 

3 1  H'.  b£ 

/•:  > 

c  - 

c  a  - 

t  C 

• 

V 

c,  ‘  ~  ,w 

0 

-  C  ...  ^  d 

.:  J' 

X  .  - 

X  (  . .  ■  ■ 

loll 

' 

o 

j- 

r  loll. 

' .  ±:  a. 

5  .  i  I  )0‘. 

X  .  £  1  . 

'  :  X 

X 

'  ;.  .  .  . 

;  .  9rao  srfX .  '  l  ■  .  .  I  J  .  o  ‘  I.  X  - 

•  X  i.  o'  X  £  1  '.r  •  xi  >  :  <  •  • 

X  ■  XI  mj  :  IXd  X.  '  -  IX  Xs  -  .  X  i  0  i  ■ 

.  o  ■  .  .  .  •  X  - .  •  'J:  •'  ■.<  I- v 

..  sX  ,.o  .  X  J'  v  -  o  '  •  ■  '  o-I  ;  (  ,r0‘  err 

XX..  Xiixoowexv  .  Ix-3 1  old  <  X  ■  '  '  " 

■ 

J  X  j  '  1  X  .  xX  '  x  v.  I  jtE  .ad  oil'  £ 

c  .  x  X  IcBcix  d  nc  .  i 

•  O  •  X  .  .  X 

o'  <  -  '  XV.  X  '  ' 

.  -  x  t  i  0i  •  Xx  IeXoX  eriX  Xo 

J  .  .  .  .(  c  ■ '  XI  Xi  J'  ■  Xu::  <  :  ■  >  -tOX  f>Xo  t 

x  X .  X  o'  '.O'  •  o  >  1  X  v  I 

.  '  X  Xes;  rio$ & 

X  j  oo  X  ■  -X  o'  XX  .-..i;  „i  oj:i  oo  <  .  ::  t.r  I 


45 


within  the  organization  and  that  superintendents  differ  among  themselves 
because  they  occupy  the  same  position  in  organizations  which  differ  in  size 
(6,  p.  136).  Microscopic  role  consensus,  that  is  the  amount  of  consensus 
within  any  one  school  board,  was  also  explored.  Attempts  were  made  to  relate 
this  to  the  homogeneity  of  the  board  but  in  general  these  hypotheses  were 
not  supported  (6,  p.  191). 

II.  INTENSITY  FUNCTION  AND  AMBIVALENCE 

The  previous  chapter  has  already  given  some  indications  that  the 
intensity  of  attitudes  concept  has  been  subjected  to  empirical  investigation. 
Suchman  (28)  has  outlined  some  of  the  studies  carried  out  by  the  Research 
Branch  of  the  U.  S.  War  Department  during  World  War  II.  In  general  these 
studies  were  carried  out  in  connection  with  determining  the  zero  point  on 
an  attitude  scale.  Although  studies  such  as  those  by  Getzels  and  Guba  (3) 
have  touched  upon  personal  involvement  in  role  conflict,  the  only  attempt 
which  has  been  made  to  approach  this  through  the  intensity  with  which  self¬ 
expectations  are  held,  and  which  is  known  to  this  investigator,  is  the  study 
carried  out  by  Seeman  (24,  25)  on  leader  ambivalence. 

Attitude  Intensity 

The  intensity  function  has  been  subjected  to  some  further  analysis 
since  the  time  of  the  Research  Branch  studies.  McDill  (20)  investigated 
the  generality  and  replicability  of  Suchman' s  conclusion  concerning  the 
invariance  of  the  zero  point.  In.  a  study  of  the  attitudes  of  5500  students 
toward  quitting  school  and  toward  the  honesty  of  policemen,  McDill  made  use 
of  three  techniques  for  obtaining  measures  of  intensity.  One  of  these  was 


'  ■  ,  . 

. 

. 


... 

•  •'  :  i-  '  •  i'.C  '  ■; ■■■■•>  ‘  '  '• '.  Ji''< 

'  '■  :i:  b.r: 

.;  '  .  .'  '  .  ..  ,  . 


r.  & 

. 

•  (  ;•  •:  1  '  •  ■ 

,  '■ 

.  '.  lH  .rt  iti  :  '  ■  i ' 

. 


46 


the  fold-over  technique,  the  second  was  through  asking,  "How  hard  was  it 
for  you  to  decide?"  after  each  question,  and  the  third  was  asking,  "How 
sure  are  you?"  after  each  question.  When  he  plotted  his  measures  of 
intensity  against  his  measures  of  content  he  did  obtain  J-shaped  curves 
but  there  was  considerable  error  around  the  curves.  The  interval  of 
inflection  at  the  lowest  point  had  a  large  range,  and  it  was  difficult 
to  decide  where  the  zero  point  might  be  located.  McDill  found  that  the 
foldover  technique  came  closest  to  producing  a  desirable  curve;  however, 
even  McDill  agrees  with  Suchman  that  there  is  some  possibility  of  spurious¬ 
ness  because  intensity  measures  obtained  in  this  way  are  not  entirely 
independent  of  content. 

Although  McDill  does  recognize  the  limitations  of  his  study,  he  also 
suggests  that  the  zero  point  may  vary  with  the  way  in  which  intensity  is 
measured.  He  cautions  that  the  findings  of  Katz  (l6),  Guttman  (8,  9),  and 
Suchman  (28)  may  possess  limited  generality  and  replicability.  This  is  in 
one  sense  an  unjustified  conclusion  because  McDill  has  not  shown  that  his 
techniques  were  comparable  to  those  used  by  other  investigators  let  alone 
superior  to  them. 

Ambivalence 

It  is  unfortunate  that  there  is  almost  no  research  evidence  concern¬ 
ing  the  psychological  meaning  of  the  intensity  function.  Seeman  (24,  25) 
adapted  the  technique  to  studying  the  difficulty  of  choice  on  various 
behavioral,  alternatives  as  reported  by  principals  and  superintendents.  In 
his  study  he  defines  the  concept  as  follows: 


, 

■ 

..  ,  ■-■•••• 

* 

■ 

■'I  ‘  i ■  '* 

8  '  {•  '  '  f.  \:'J  •  '  '  '  '•  '■  ■=  : 

,  1 

' 

•;  •  ' 

' 

. 

; 


47 


The  concept  of  ambivalence  is  used  here,  not  in  the  sense  of 
deep-lying  unresolved  personality  conflicts,  but  simply  as 
difficulty  of  choice.  The  statistical  data  on  ambivalence 
were  obtained  by  asking  the  question  after  each  of  the  ten 
i+oms  "How  hard  was  it  for  you  to  make  your 


In  spite  of  the  fact  that  the  definition  leaves  much  to  be  desired  and 
that  the  number  of  items  was  rather  small,  Seeman  did  draw  some  conclusions 
from  his  data.  He  observed  that  ambivalence  tended  to  be  lower  among 
leaders  than  teachers,  that  there  was  little  relationship  between  difficulty 
of  choice  for  leaders  and  non-leaders  on  specific  items,  and  that  questions 
of  status  seemed  to  give  the  leaders  the  most  difficulty  of  choice  (25,p.53)« 
Even  when  total  scores  were  used,  leaders  consistently  had  lower  ambivalence 
scores  than  non-leaders.  Seeman  suggests  that  this  lower  score  may  be  a 
response  to  the  pressure  for  clarity  which  is  demanded  of  the  leader  in 
carrying  out  the  expectations  for  his  role.  That  is,  leaders  tend  to  reply 
the  way  that  they  think  leaders  should  reply.  This  suggests  the  concept  of 
inauthenticity,  that  is,  of  individuals  acting  or  responding  in  a  fashion 
which  fits  the  stereotype  of  the  position  rather  than  responding  as  they 
actually  think  and  feel. 

A  hypothesis  which  was  tested  in  the  research  was  that  even  though 
the  expressed  ambivalence  could  not  be  taken  at  its  face  value,  the  sub¬ 
ordinates  evaluation  of  the  leader  might  in  some  way  be  related  to  the 
expressed  ambivalence.  The  first  testing  of  the  hypothesis  was  based  on 
the  notion  that  the  leaders  who  expressed  high  ambivalence,  that  is,  those 
who  seemed  to  be  aware  in  some  way  that  there  was  a  natural  conflict  in 
some  of  their  choices  would  be  described  more  favorably  by  their  subordi¬ 


nates  than  those  who  denied  this  problem  by  expressing  little  difficulty 


.  '  ■  '  ' 

’ 

■ 

' 

'  . 

' 

■ 

- 

'  .  • .  '  . 

-  ■  "j [  ‘p  "  ’  '  > 

.  ' 

'  '  ' 


48 

in  making  their  choices.  This  linear  relationship  was  not  supported  by 
the  data.  However,  the  data  did  suggest  that  the  relationship  might  he 
curvilinear  in  that  both  those  who  expressed  high  ambivalence  and  those 
who  expressed  low  ambivalence  were  not  evaluated  as  highly  as  those  who 
expressed  moderate  ambivalence. 

To  test  this  relationship  the  correlation  ratio  rather  than  linear 
correlations  was  used,  but  again  the  relationships  between  leaders’  evalu¬ 
ation  and  ambivalence  scores  were  not  significant.  In  the  next  attempt 
the  relationship  of  ambivalence  scores  to  descriptions  of  leader  behavior 
on  the  dimensions  of  communication,  separation,  change,  and  domination 
was  explored.  Here  again  the  correlation  ratios  were  low  and  only  one  of 
them  reached  significance  at  the  .05  level.  Seeman  did  indicate  that  from 
the  shape  of  the  distribution  there  were  some  grounds  for  believing  that 
both  very  high  and  very  low  ambivalence  is  associated  to  some  extent  with 
what  teachers  describe  as  being  poor  leadership  behavior  (25,  p.  56). 

These  and  other  observations  lead  Seeman  to  suggest  that  perhaps  there  is 
such  a  thing  as  the  realist  pattern  in  leader  behavior.  Part  of  this 
pattern  is  the  realistic  appraisal  of  the  situation  in  which  the  leader 
finds  himself  which  includes  neither  denying  his  difficulty  of  choice  nor 
exaggerating  it.  The  leaders  who  can  do  this  are  the  ones  who  should  be 
described  more  favorably  by  their  subordinates. 

III.  SOME  STUDIES  OF  LEADER  BEHAVIOR 

The  Leader  Behavior  Description  Questionnaire  which  was  developed  in 
the  Ohio  Leadership  Studies  has  been  used  in  numerous  situations.  Although 


I 


, 

. 


;  '•  .  '  ' 
„  ’  ‘  ‘  *.  •  •  : 


> 


. 


■ 

■ 

■ 

■ 


■ 

* 

. 

. 


49 


the  present  instrument  consists  of  fifteen  items  for  each  dimension, 
some  of  the  earlier  studies  used  varying  numbers  of  items.  Fleishman  (2) 
found  the  instrument  useful  in  industrial  situations  and  Halpin  (ll)  made 
use  of  it  in  studying  the  leader  behavior  of  aircraft  commanders.  Hemphill 
(14)  studied  the  leader  behavior  of  university  department  heads,  and  the 
general  conclusion  that  might  be  reached  from  all  of  these  studies  is  that 
the  more  desirable  type  of  leader  behavior  is  that  which  is  characterized 
by  high  scores  on  both  dimensions. 

Halpin  also  studied  the  "Ideal"  and  "Real"  leader  behavior  of  a 
number  of  school  superintendents  (12).  He  concluded  that  the  "Ideal" 
leader  behavior  from  the  point  of  view  of  both  teachers  and  board  members 
indicated  that  the  superintendent  should  be  high  on  both  dimensions.  The 
"Real"  descriptions  proved  to  indicate  significant  differences  between 
the  leader  behavior  styles  of  the  superintendents  who  were  under  investi¬ 
gation.  The  results  of  these  studies  lead  Halpin  to  offer  the  instrument 
as  an  aid  for  the  evaluation  of  school  superintendents  (13).  Since  that 
time  there  has  been  some  additional  evidence  relating  descriptions  of  the 
behavior  of  leaders  to  the  outcomes  of  group  effort. 

The  questionnaire  was  adapted,  by  McBeath  (19)  for  use  in  the  class¬ 
room  situation  to  study  the  relationship  of  the  leader  behavior  of  teachers 
to  effectiveness  rating  by  pupils,  principals,  superintendents,  fellow 
teachers,  and  self-ratings.  McBeath  concluded  that  part  of  teaching 
effectiveness  was  related  to  the  leader  behavior  style  of  the  teacher. 
Teachers  who  were  rated  as  being  high  in  effectiveness  tended  to  score 
above  the  median  on  the  two  dimensions  more  often  than  did  those  teachers 
who  were  rated  as  being  low  in  effectiveness. 

Using  this  same  adaptation  of  the  LBDQ,  Greenfield  (5)  examined  the 


......  •-  •  "  '  ■  •  . '  "  :  v' 

'  f  ■■  ■  ■  ' 

.  •  '  ■  I  "  ,  - 

. 

' 

' 

, 

. 

' 

'  .  ••  -  '  '  1  ' 

;  ■  ■  :  :  •  ■  ■  •  ■  '  • 

' 


50 


relationship  between  the  leader  behavior  of  teachers  and  measures  of 
pupil  growtn  determined  from  Grade  Nine  examination  results  for  classes 
instructed  by  these  teachers.  He  concluded  that  Initiating  Structure  was 

I 

significantly  related  to  pupil  growth,  and  that  Consideration  was  also 
related  to  this  measure  but  did  not  appear  to  be  as  important  as  was  the 
other  dimension. 

The  validity  of  the  LBDQ,  as  a  measure  of  the  effectiveness  of  a 
principal  is  also  aupported  in  part  by  the  research  carried  out  by 
Keeler  (17)*  The  relationships  investigated  were  those  between  principal 
leader  behavior,  staff  morale,  and  productivity  of  the  school  as  determined 
from  province-wide  examinations  at  the  Grade  Nine  level.  Keeler  hypothe¬ 
sized  that  leader  behavior  would  be  significantly  related  to  staff  morale 
and  that  staff  morale  would  in  turn  be  related  to  productivity.  Since  the 
data  would  also  permit  testing  the  relationship  between  leader  behavior 
and  productivity  an  appropriate  hypothesis  was  also  formulated  for  this. 
Keeler  concluded  that  the  research  evidence  indicated  a  relationship  between 
leader  behavior  and  productivity  and  also  between  morale  and  productivity 
but  not  between  leader  behavior  and  morale.  If  the  relationship  between 
leader  behavior  and  productivity  were  a  causal  one,  it  would  give  evidence 
of  the  validity  of  the  instrument  for  measuring  effective  leader  behavior. 

The  LBDO  was  used  in  this  study  for  two  main  reasons.  The  first  of 
these  was  that  it  did  enable  the  investigator  to  describe  styles  of  leader 
behavior.  The  second  wa3  that  there  is  some  basis  for  suggesting  that  the 
style  of  leader  behavior  which  is  characterized  by  high  scores  on  both 
dimensions  could  be  considered  as  effective  leader  behavior.  A  further 


;  o 


' 


- 


. 


,  ’  ;  '  '  '  ■ 

.  :  j  tsj 

- 

. 

: 


51 


reason  is  that  the  use  of  the  instrument  in  this  study  will  serve  to 
link  this  research  with  the  other  research  which  has  made  use  of  the 
instrument. 


REFERENCES  FOR  CHAPTER  III 


Burchard,  Walo  W.  "Role  Conflicts  in  Military  Chaplains," 
American  Sociological  Review ,  19:528-535*  October,  1954. 

Fleishman,  Edwin  A.  "The  Measurement  of  Leadership  Attitudes 
in  Industry,"  Journal  of  Applied  Psychology,  37 : 153-158 , 
June,  1953* 

Getzels,  J.W.  and  E.G.Guba.  "Role,  Role  Conflict,  and 
Effectiveness,"  American  Sociological  Review,  19:164- 
175,  April,  1954.  "  " 

Getzels,  J.W.  and  E.G.Guba.  "The  Structure  of  Roles  and  Role 
Conflict  in  the  Teaching  Situation,"  Journal  of  Educational 
Sociology,  29:30-40,  September,  1955* 

Greenfield,  Thomas  Barr.  "Teacher  Leader  Behavior  and  its 
Relation  to  Effectiveness  as  Measured  by  Pupil  Growth." 
Unpublished  Master’s  thesis,  The  University  of  Alberta, 
Edmonton,  1961. 

Gross,  Neal,  Ward  S.  Mason,  and  Alexander  McEachem. 
Explorations  in  Role  Analysis:  Studies  of  the  School 
Superintendency.  New  York:  John  Wiley  and  Sons,  Inc., 

1^585 

Gullahom,  John  T.  "Measuring  Role  Conflict,"  American 
Journal  of  Sociology,  61:299-303,  January,  1956. 

Guttman,  Louis.  "The  Cornell  Technique  for  Scale  and 
Intensity  Analysis,"  Educational  and  Psychological 
Measurement ,  7:247-279,  Slimmer,  1947* 

Guttman,  Louis.  "The  Principal  Components  of  Scalable 
Attitudes,"  in  Paul  F.  Lazarsfeld  (ed.).  Mathematical 
Thinking  in  the  Social  Sciences.  Glencoe,  Illinois: 

The  Free  Tress,  1954*  Pp.  2i6-257* 

Hall,  Robert  L.  "Social  Influences  on  the  Aircraft  Commander’s 
Role,"  American  Sociological  Review,  20:292-299,  June,  1955* 

Hal pin,  Andrew  W.  "The  Leadership  Behavior  and  Combat 

Performance  of  Airplane  Commanders,"  Journal  of  Abnormal 
and  Social  Psychology,  49:  19-22,  January,  1954* 


•  •  ’ . 

' 


i r 


•  ■ 


, 

■ 


. 

;• 


' 


/ 


-  -  -  .  ' 


' 

. 

’ 

.  :  .r  ' 

•-  • 


. 

' 


.  . 


»  * 


•  .  '  , 


. 

' 


.  '  "> 

■ 


,  ■  i  '  -  to:..'  { 

. 


(12)  Hal pin,  Andrew  W.  The  Leadership  Behavior  of  School 

Superintendents .  Second  Edition.  Chicago:  Midwest 
Administration  Center,  1959 • 

(13)  Hal pin,  Andrew  W.  "The  Superintendent's  Effectiveness  as 

a  Leader,"  Administrator's  notebook,  Vol.  7,  Ho.  2, 
October,  1958* 

(14)  Hemphill,  John  K.  "Leader  Behavior  Associated  with  the 

Administrative  Reputation  of  College  Departments,”  in 
Ralph  M.  Stogdill  and  Alvin  E.  Coons  (eds.).  Leader 
Behavior;  Its  Description  and  Measurement.  Columbus, 

Ohio:  The"  'Ohio  State  University,  1957*  Pp*  74-85* 

(15)  Hencley,  Stephen  P.  "The  Conflict  Patterns  of  School 

Superintendents,"  Administrator' s  Notebook,  Vol.  8, 

No.  9,  May,  I960, 

(16)  Katz,  Daniel.  "The  Measurement  of  Intensity,"  in  Hadley 

Cantril  (ed.).  Gauging  Public  Opinion.  Princeton,  New 
Jersey:  Princeton  University  Press,  1944*  Chapter  3* 

(17)  Keeler,  Bernard  T.  "Dimensions  of  Leader  Behavior  of 

Principals,  Staff  Morale  and  Productivity,"  Unpublished 
Doctoral  thesis,  The  University  of  Alberta,  Edmonton, 
1961. 

(18)  Laulicht,  Jerome.  "Role  Conflict,  the  Pattern  Variable 

Theory,  and  Scalogram  Analysis,"  Social  Eorces,  33 2 
250-254,  March,  1955* 

(19)  McBeath,  Arthur  Groat.  "Teacher  Leader  Behavior  and  its 

Relation  to  Teacher  Effectiveness."  Unpublished  Master's 
thesis,  The  University  of  Alberta,  Edmonton,  1959* 

(20)  McDill,  Edward  L.  "A  Comparison  of  Three  Measures  of 

Attitude  Intensity,"  Social  Forces ,  38:95-99, 

December,  1959* 

(21)  Motz,  Annabelle  Bender.  "The  Role  Conception  Inventory: 

A  Tool  for  Research  in  Social  Psychology,"  American 
Sociological  Review,  17:465-471,  August,  1952* 

(22)  Parsons,  Talcott,  and  Edward  A.  Shils  (eds.).  Toward  a 

General  Theory  of  Action.  Cambridge,  Massachusetts: 
Harvard  University  Press,  1951* 


(23)  Perry,  Stewart  E. ,  and  Lyman  C.  Wynne.  "Role  Conflict, 

Role  Definition,  and  Social  Change  in  a  Clinical 
Research  Organization,"  Social  Forces ,  38:62-65? 

October,  1959* 

(24)  Seeman,  Melvin.  "Role  Conflict  and  Ambivalence  in  Leader¬ 

ship,"  American  Sociological  Review,  18:373-380, 

August,  1953* 

(25)  Seeman,  Melvin.  Social  Status  and  Leadership.  Columbus, 

Ohio:  The  Ohio  State  University,  I960. 

(26)  Stouffer,  Samuel  A.  "An  Analysis  of  Conflicting  Social 

Norms,"  American  Sociological  Review,  14:707-717? 

December,  1949* 

(27)  Stouffer,  Samuel  A.  and  Jackson  Toby.  "Role  Conflict  and 

Personality,"  American  Journal  of  Sociology,  56:395-406, 
March,  1951  •„ 

(28)  Suchman,  Edward  A.  "The  Intensity  Component  in  Attitude  and 

Opinion  Research," , in  Samuel  A.  Stouffer,  et  al.  Measurement 
and  Prediction.  Vol.  IY  of  Studies  in  Social  ""Psychology 
in  World  War  II.  4  vols.  Princeton,  New  J  ers  ey :  Princ e  t  on 
University  ’Press ,  1950.  Chapter  7?  Pp.  213-276. 

(29)  Sutcliffe,  J.  P.  and  M.  Habeiman.  "Factors  Influencing  Choice 

in  Role  Conflict  Situations,"  American  Sociological  Review, 
21:695-703?  December,  1956. 

(30)  WispS,  Lauren  G.  Sociometric  Analysis  of  Conflicting  Role- 

Expectations,"  American  Journal  of  Sociology,  61:134-137? 
September,  1955*  Errata  61:363?  January ,  1956 . 


, 

:  . 

, 

. 

- 


. 


. 


. 


, 


.  « 

. 


, 


' 


* 


CHAPTER  IV 


SCALE  ANALYSIS:  THEORY  AND  METHODOLOGY 

Since  the  theory  of  scalogram  analysis  and  the  procedures  involved 
in  it  are  central  to  a  major  portion  of  this  study,  each  of  these  will  "be 
developed  in  some  detail.  This  degree  of  development  seems  necessary 
"because,  to  the  "best  knowledge  of  the  investigator,  the  method  has  not 
been  used  previously  in  this  specific  area,  and  it  may  prove  to  be  a 
neglected  approach  which  could  be  utilized  to  a  much  greater  extent  than 
it  has  been.  Although  the  techniques  employed  are  straightforward  and 
apparently  simple,  it  will  become  evident  that  the  implications  of  the 
existence  of  a  scalable  area,  in  the  sense  that  the  term  is  used  here, 
are  far  reaching.  Much  of  the  material  in  this  chapter  is  drawn  from 
relevant  sections  in  Measurement  and  Prediction  (7*  PP»  3  -  361 )  which 
contains  the  most  definitive  outline  available  on  the  theory  amd  method 
of  scale  analysis. 

I.  THE  THEORY  OF  SCALE  ANALYSIS 

The  Universe  of  Attributes 

The  problem  with  which  scale  analysis  primarily  concerns  itself 
is  the  question  of  whether  or  not  a  series  of  items  of  qualitative  data 
can  be  considered  as  belonging  to  the  same  dimension  (7a,  p.  46).  In 
other  words,  scale  analysis  asks  the  question  of  whether  or  not  one  can 
consider  the  responses  to  a  set  of  items  as  ordering  individuals  along 
the  same  continuum.  The  process  begins  with  a  definition  of  a  certain 


■ 

lov&b 

. 

y-1  ■'  ' 

« 

■  ■  ■  4  ■  :  ,./■  •  ’  •  "  • 

■  ■ 


’ 

' 


* 


56 


universe  of  interest  or  universe  of  attributes  which  is  defined,  prior 
to  any  attempts  at  data  collection  (7b,  pp.  80-88).  Once  the  universe 
has  been  defined,  the  items  constructed,  and  the  responses  obtained,  scale 
analysis  investigates  whether  or  not  this  is  scalable  in  the  sense  that 
only  one  variable  is  involved  and  whether  or  not  individuals  can  be  ranked 
in  accordance  with  the  extent  to  which  the  variable  is  present  in  each 
individual. 

It  is  assumed  that  there  are  numerous  items  which  might  be  selected 
from  the  universe  but  that  for  a  number  of  reasons  it  is  neither  possible 
nor  necessary  to  include  all  of  the  items;  consequently,  the  universe  is 
sampled.  Items  are  selected  solely  on  an  a  priori  judgment  that  their 
content  is  such  that  they  belong  within  the  defined  universe  of  attributes. 
Once  the  items  have  been  selected,  scale  analysis  tests  the  hypothesis  that 
a  group  of  respondents  can  be  ranked  in  a  meaningful  way  with  respect  to 
this  area.  The  theory  on  which  the  procedures  are  based  suggests  that  if 
the  area  is  scalable,  it  does  not  matter  which  items  are  selected  because 
the  same  rank  ordering  of  people  will  result  regardless  of  which  items 
are  used  (7b,  p.  8l). 

Example  of  a  Dichotomous  Scale 

A  clarification  of  what  is  implied  in  the  statement  that  a  scale 
exists  in  this  restricted  sense  can  best  be  achieved  by  the  use  of  an 
example.  Even  the  definition  of  what  constitutes  a  scale  can  await  the 
completion  of  the  illustration.  Let  us  suppose  that  we  have  defined  as  a 
universe  of  interest  the  amount  of  formal  education  completed  by  a  number 


of  individuals.  For  this  purpose  we  design  the  following  three  questions: 


. 

'  i  ••••/ 

■ 

. 

-  '  ' 


’ 

. 

i  • 

’ 

' 


' 

* 

. 


•  1 


57 

(1)  Did  you  complete  Grade  8? 

(2)  Did  you  graduate  from  High  School? 

(3)  Have  you  earned  one  or  more  university  degrees? 

Notice  that  there  are  a  total  of  eight  possible  response  combinations  if 
the  individuals  are  instructed  to  respond  either  yes  or  no  to  each  of  the 
questions.  But  one  would,  in  reality,  expect  to  obtain  only  four  of  the 
eight  possible  combinations.  That  is,  we  would  not  expect  to  have  indi¬ 
viduals  answer  yes  to  (2)  and  (3)  if  they  have  answered  no  to  (l);  nor  would 
we  expect  to  obtain  a  yes ,  no,  yes  combination  of  responses.  The  reason  for 
this  is  that  the  series  of  items  has  a  special  cumulative  property. 

For  purposes  of  comparing  the  formal  education  of  a  number  of 
individuals,  it  would  be  possible  to  assign  a  value  of  one  to  each  yes 
response  and  a  value  of  zero  to  each  no  answer.  If  we  then  totalled  each 
person's  responses  we  would  obtain  a  possibility  of  four  different  scores, 
namely,  3>  2,  1,  and  0.  The  person  with  a  score  of  2  should  have  answered 
questions  (l)  and  (2)  yes  while  one  who  obtained  a  score  of  0  would  have 
answered  no  to  all  three  questions,  and  so  on.  The  possible  responses  for 
each  score  could  be  diagrammed  as  follows: 

Items  Answered  Items  Answered 


Scale 

"Yes" 

"No" 

Scores 

3 

2 

1 

5 

2  1 

3 

X 

X 

X 

2 

X 

X 

X 

1 

X 

X 

X 

0 

X 

X  X 

FIGURE  2 

RESPONSE  PATTERN  FOR  THREE  DICHOTOMOUS  ITEMS 


i;  .T: 

.  ;  -  j;  ■  err  ■'  ...  '  '. 

,  £TO  .-  ; 


. 


■ 

•  - 


n 

•  ‘  ■  ■  '  ’  •’ 1  ).• ’’i 

. 

....  ■  1 

,  .... 

. 


•  ■  >a 


58 


Such  a  picture  is  called  a  scalogram,  and  the  existence  of  the  parallelo¬ 
gram  pattern  for  a  number  of  dichotomous  items  is  a  necessary  and  sufficient 
condition  for  the  items  to  form  a  scale  because  the  responses  are  perfectly 
reproducible  from  the  scale  scores  (7b,  p.  67).  Stated  otherwise,  the 
responses  are  a  simple  function  of  the  quantitative  variable  which  has 
been  assigned  to  the  different  patterns  of  responses .  Figure  2  includes  the 
four  perfect  scale  types  which  we  would  expect  to  obtain  no  matter  how  many 
respondents  we  had  in  our  sample.  The  only  difference  in  the  scalogram 
picture  for  a  larger  sample  would  be  the  variations  in  the  number  of 
respondents  at  each  scale  type.  Figure  5  shows  how  such  a  distribution  of 
responses  might  look  for  ten  respondents. 


Items  Answered 

Items 

Answered 

Scale 

"Yes” 

»» 

No" 

Score 

3  2 

1 

3 

2 

1 

a 

5 

X  X 

X 

b 

3 

X  X 

X 

c 

2 

X 

X 

X 

d 

2 

X 

X 

X 

e 

2 

X 

X 

X 

f 

1 

X 

X 

X 

e 

1 

X 

X 

X 

h 

0 

X 

X 

X 

i 

0 

X 

X 

X 

j 

0 

X 

X 

X 

FIGURE  3 


SCALOGRAM  FOR  THREE  DICHOTOMOUS  ITEMS 
Here  again  the  existence  of  the  parallelogram  pattern  is  evidence 
that  the  ten  people  have  been  ranked  along  a  vertical  continuum.  There 
are  tied  ranks,  and  we  can  assign  only  the  four  different  scale  scores  on 
the  basis  of  the  combination  of  responses  as  in  Figure  2.  The  scalogram 


■ 

.  .  •  ,  ' . .....  ." 

■ 

■  .  . ■ . . ;  . ... 

■  x  :• 

,  .  ■  ■  •  :  ■  f  •zo  \  '  ’  ••  ■.  ]' ■rf;  . 


.  . 


59 


illustrates  a  number  of  properties  of  a  series  of  scalable  items.  The 
first  is  that  each  question  has  placed  a  cutting  point  in  the  continuum. 
Secondly,  all  those  who  gave  a  positive  response  to  a  specific  question 
are  above  a  cutting  point  between  scale  scores  while  all  those  who  gave 
a  negative  response  are  below  the  cutting  point.  This  is  the  same  as  saying 
that  any  person  answering  positively  to  a  particular  question  is  ranked 
higher  than  one  who  responded  negatively  to  the  same  question.  Furthermore, 
given  the  scale  score  for  a  respondent  it  is  possible  to  reproduce  that 
individual’s  responses  perfectly. 

Although  this  example  has  dealt  with  only  three  items,  and  with 
dichotomous  items  at  that,  the  same  rationale  and  procedures  are  used  for 
testing  the  scalability  of  any  number  of  items  with  any  number  of  response 
categories.  The  example  has  also  dealt  with  a  perfect  scale  mainly  because 
the  scale  model  is  deterministic  rather  than  probabilistic.  As  will  be 
discussed  later,  the  perfect  scale  is  not  found  in  empirical  research  and 
the  investigator  must  usually  be  content  with  approximations  to  the  model. 

Definition  of  a  Scale 

So  far  it  has  been  made  clear  that  scale  analysis  tests  the 
hypothesis  that  a  series  of  items  can  be  used  to  order  individuals  or  objects 
along  a  single  continuum.  Although  other  procedures  used  in  the  development 
of  various  "scales"  also  interest  themselves  with  ordering  people,  no  other 
approach  is  concerned,  with  the  dimensionality  of  the  "scale"  to  such  an 
extent  as  is  scale  analysis  nor  in  this  particular  way.  Guttman  has  proposed 
three  equivalent  definitions  of  a  scale  (7h,  p.  62).  The  first  of  these  is 


that  a  series  of  items  will  be  called  a  scale  if  a  person  with  a  higher  rank 


....  .  • 

r '  _  '  '  '■ 

.  ■  !•  r  ;'•)  '  i  .'>■ : 

■ 

■ 

•  ■'  ’  ' 


* 

.  .  ■  '  , 


. 


60 


than  another  is  just  as  high  or  higher  on  every  item  than  the  other 
person.  This  is  the  same  as  saying  that  if  within  a  given  item  a  certain 
response  category  is  more  favorable  than  another,  then  all  the  people  in 
the  lower  category  will  have  lower  ranks  than  the  people  in  the  higher 
category.  The  third  definition  is  the  one  on  which  the  empirical  process 
of  scalogram  analysis  is  based,  and  this  is  that  the  ordering  of  individuals 
and  their  responses  shall  be  such  that  each  person's  responses  will  be 
reproducible  from  his  rank  alone.  This  has  been  stated  in  more  technical 
terms  by  Guttman: 

For  a  given  population  of  objects,  the  multivariate  frequency 
distribution  of  a  universe  of  attributes  will  be  called  a  scale 
if  it  is  possible  to  derive  from  the  distribution  a  quantitative 
variable  with  which  to  characterize  the  objects  such  that  each 
attribute  is  a  simple  function  of  that  quantitative  variable. 

Such  a  quantitative  variable  is  called  a  scale  variable.  (7b,  p.64) 

It  is  evident  that  according  to  these  definitions  of  a  scale,  a 
series  of  items  must  conform  to  a  rigorous  condition  if  they  are  to  be 
considered  scalable.  This  differs  to  a  considerable  extent  from  other 
approaches  which  do  not  in  reality  concern  themselves  with  this  question 
of  dimensionality.  For  the  purpose  of  this  thesis,  the  term  scale  will 
be  restricted  to  a  series  of  items  which  can  be  used  to  order  individuals 
in  such  a  way  that  their  responses  can  be  reproduced  from  their  ranks. 

If  this  cannot  be  done,  the  series  of  items  will  not  be  termed  a  scale. 

Measurement  of  Error 

In  very  few,  if  any,  of  the  scales  obtained  in  empirical  research 
is  it  possible  to  reproduce  the  responses  from  the  ranks  without  error; 
in  all  of  these  it  i3  almost  certain  that  some  responses  will  fall 


•  X  ' 

•  ■-  :  '  V  *  ,  : ! 


■  '  > 


' 

' 

*.  : 

* 

:  ■  ■  ••  r 

• 

• 

•>.»:  >.■' 


, ,  ■ 

.  •  ■ 


■ 


' 

■ 

’ 


61 


outside  the  pattern  of  perfect  scale  types.  In  order  to  give  an  index 
of  the  extent  to  which  a  certain  scale  deviates  from  the  ideal  pattern, 
the  coefficient  of  reproducibility  is  computed.  This  is  obtained  by 
simply  counting  up  the  number  of  errors  -  responses  which  fall  outside 
the  main  pattern  -  dividing  this  sum  by  the  total  number  of  responses  and 
subtracting  the  quotient  from  unity.  Thus  the  coefficient  of  reproducibility 
gives  the  proportion  of  responses  which  would  have  been  predicted  correctly 
from  a  knowledge  of  rank  or  scale  scores  alone.  Unfortunately ,  this 
coefficient  suffers  from  an  unknown  sampling  distribution,  but  a  level  of 
90 $  reproducibility  has  been  arbitrarily  set  as  a  lower  limit  for  consider¬ 
ing  a  series  of  items  to  be  scalable  (7b,  p.  77)* 

Although  the  coefficient  of  reproducibility  is  a  major  criterion  for 
scalability,  a  number  of  other  criteria  are  also  employed  and  must  be  used 
in  conjunction  with  it  (7b,  pp.  JQ-BO) . 

These  criteria  are: 

(1)  Marginal  Frequencies.  At  least  some  of  the  items  should  have 
close  to  equal  frequencies  of  positive  and  negative  responses 
because  extreme  marginal  frequencies  automatically  mean  a  high 
coefficient  of  reproducibility. 

(2)  Pattern  of  Errors .  The  pattern  of  errors  should  be  random  and 
not  more  than  about  five  per  cent  of  any  of  the  errors  should 
be  identical  or  grouped. 

(3)  Number  of  Items.  If  all  items  are  dichotomized  at  least  ten 
items  should  be  used  although  a  lesser  number  might  be  used  if 


some  of  the  items  have  from  thirty  to  seventy  per  cent  of  their 


. 

.  - .  .  •  •  . .  .  • 


■ 


■ 


'  • 


•  •  ■ 

■ 

I  n 


62 


responses  in  the  modal  category. 

(4)  Response  Categories .  Acceptance  of  the  hypothesis  is  enhanced 
if  not  all  items  are  dichotomized. 

Quasi  Seal es 

Empirical  research  has  revealed  that  even  this  arbitrary  standard 
of  ninety  per  cent  reproducibility  may  be  difficult  to  achieve  (7b,  p.  89)  • 
If  the  other  criteria  concerning  the  number  of  items,  the  pattern  of  errors, 
the  number  of  response  categories,  and  the  range  of  marginal  frequencies 
are  satisfied,  such  approximations  to  scales  or  almost  scales  are  termed 
quasi  scales  (7d,  pp.  159-163).  The  essential  features  of  such  approxi¬ 
mations  to  scales  is  that  the  pattern  of  errors  is  random;  if  the  errors 
are  grouped  this  is  considered  to  be  an  indication  that  more  than  one 
dimension  is  present  in  the  series  of  items.  Quasi  scales  indicate  that 
even  though  there  is  only  one  major  dimension  present,  many  other  less 
significant  variables  are  involved  in  the  responses.  This  is  concluded 
from  the  observation  that  the  errors  occur  in  a  sort  of  a  gradient  with 
the  number  of  errors  decreasing  as  one  moves  away  from  the  highest  density 
of  responses.  Concerning  the  utility  of  quasi  scales,  Suchman  has  stated: 

The  importance  of  a  quasi  scale  lies  in  how  it  is  used  for 
external  prediction  problems.  While  we  cannot  derive  a  person's 
responses  from  his  quasi-scale  score,  the  score  does  yield  a  zero- 
order  correlation  with  any  outside  variable  which  is  equivalent 
to  the  multiple  correlation  on  all  the  items  in  the  quasi  scale. 

The  prediction  of  the  external  variable  rests  essentially  on  the 
dominant  factor  that  is  being  measured  by  the  quasi-scale  score. 

Thus  a  quasi  scale  has  the  full  mathematical  advantage  of  a 
scalable  area.  (7d,  p.  162) 

This  property  of  a  quasi  scale  is  very  important  since  it  is  highly  likely 


. 

•  I  '  ’ 

- 

* 

•..  .  -  ... 

•  :  .  r  < 


- 


' 


■ 


. 

■  ' 

■ 

■ 

■  ... :• :  <  ■■  -  &  ••  '  - 


63 


that  in  practice  it  is  much  easier  to  obtain  quasi  scales  than  scales. 

And  if  in  the  problem  of  relating  the  attitude  area  to  an  outside  variable 
the  full  advantage  of  having  a  scale  exists,  then  there  is  not  much  greater 
value  in  having  a  scale  particularly  if  one  is  not  interested  in  reproducing 
responses  from  scale  scores. 

II.  THE  METHODOLOGY  OF  SCALE  ANALYSIS 

A  number  of  different  techniques  have  been  developed  for  testing  the 
scalability  of  a  series  of  items;  each  of  these  has  both  advantages  and 
disadvantages  when  compared  with  the  other  methods  which  are  available. 

There  are  certain  procedures  such  as  the  Cornell  technique  (4)  which  make  use 
of  graphic  methods  and  do  not  require  any  mechanical  equipment.  Rapid  scoring 
procedures  which  make  use  of  punched  cards  and  sorters  have  also  been 
developed  (l,  2,  5)*  More  recently  a  method  which  utilizes  summary  statistics 
has  been  proposed  (3).  The  Research  Branch  of  the  U.S.  War  Department,  which 
has  made  the  most  extensive  use  of  the  theory  of  scale  analysis  in  any  single 
series  of  investigations,  utilized  the  scalogram  board  technique  (7c). 

Although  this  method  has  not  become  too  popular  for  a  number  of  practical  and 
procedural  reasons,  it  is  the  method  which  was  used  in  the  study  reported 
in  this  thesis.  As  a  technique  it  most  clearly  illustrates  the  rigorous 
criterion  which  a  series  of  items  must  satisfy  in  order  to  fulfill  the 
requirements  for  a  scale.  Because  of  its  importance  for  this  particular 
research,  the  procedure  will  be  outlined  in  detail. 

The  Scalogram  Board 


Most  procedures  of  scale  analysis  involve  techniques  to  determine 


■ 

' 

.  •  r  : 


.  •  ■  ;  •  :  r  •/"'  • 

:  <"  -rr.  *tr>n:  ''to  t  : "•  : 

'  1  •'  '  '■  '  " ' 

,  '  '  ■  1  :  " 

•  ■■■'>■ 


. 

• 

• 

3 

• 

• 

:  - 

64 


whether  or  not  the  basic  parallelogram  pattern  is  present  in  a  series  of 
responses.  Louis  Guttman  designed  a  board  which  makes  it  possible  to 
determine  visually  whether  or  not  this  pattern  is  present.  Each  board 
consists  of  a  frame  for  holding  one  hundred  small  strips  or  slats  in  each 
of  which  are  drilled  one  hundred  holes  at  equal  distances.  The  purpose  of 
the  holes  is  to  hold  the  shot  which  is  used  to  indicate  an  individual's 
responses  to  each  question.  The  Research  Branch  used  two  boards  in  an 
analysis,  and  the  distance  between  rows  of  holes  was  made  equal  to  distances 
between  columns  of  holes.  This  was  done  so  that  the  shot  could  be  trans¬ 
ferred  from  one  board  to  the  other  by  superimposing  one  board  on  the  one 
containing  the  markers  in  such  a  way  that  the  slats  were  at  right  angles 
to  each  other  and  then  inverting  the  boards. 


Initial  Arrangement 

In  order  to  carry  out  the  scalogram  analysis  one  hundred  respondents 
are  selected  at  random  from  among  those  which  are  available.  In  the  initial 
arrangement  each  slat  will  contain  the  responses  of  one  individual  and  each 
column  will  contain  the  responses  for  one  answer  category.  The  description 
which  Suchman  gives  of  the  manner  in  which  the  responses  are  placed  on  the 
board  is  so  clear  and  concise  that  it  is  quoted  verbatim: 

The  pattern  of  our  initial  scale  setup  is  determined  by  the 
frequencies  of  the  "positive'*  and  "negative"  categories  of  each 
question.  The  scalogram  is  built  from  left  to  right,  column  by 
column,  each  column  containing  a  single  category  of  response  for 
a  single  question.  The  first  column  consists  of  the  "positive" 
category  with  the  fewest  responses,  and  so  on  until  all  the 
"positive”  responses  of  the  questions  in  the  series  are  represented. 

Here  a  column  is  left  vacant  to  set  off  the  "positive"  from  the 
"neutral"  responses,  which  are  the  next  group  placed  in  the  board. 


' 

■> 


••  ■ 

, 

)  03  »»i 

. 

•  •  >  f  •  y.  '■  ' 


■ 

, 


- 

■ 

■  ' 

. 

'  ' 

' 


65 


The  "neutral”  responses  may  follow  the  question,  or  column,  order 
of  either  the"positive"  or  "negative"  responses.  If  any  of  the 
questions  have  more  than  onp  "neutral"  response,  the  "Positive- 
neutral"  responses  of  all  the  questions  are  placed  first,  following 
the  column  order  assigned  the  respective  "positive”  categories. 
Immediately  following1  are  the  "neutral -neutrals" ,  and  then  the 
"n-egative-neutrals",  following.. the  column  order  assigned  the 
"negatives".  Another  vertical  column  is  left  vacant  and  is 
followed  by  the  "negative"  responses.  The  "negative”  category 
order  is  determined  again  on  the  basis  of  frequency  of  "negative" 
responses.  The  categories  in  the  "negative"  section  are  placed 
from  left  to  right  in  order  of  highest  frequency  to  lowest. 

When  "no  answers"  occur,  they  are  placed  a  few  columns  to  the  right 
of  the  scale  proper.  If  it  is  desired  to  test  whether  a  question 
of  different  content  correlates  with  the  scale,  its  responses  may 
be  set  in  further  on  the  right  side  of  the  board,  with  the  columns 
ranging  from  "positive"  to  "negative".  (7c,  pp.  102-105) 

No  pattern  of  responses  will  be  apparent  from  the  initial  arrangement 

because  the  respondents  are  in  random  order. 


Ranking  Respondents 

Ihe  slats  or  respondents  are  now  moved  up  or  down  to  arrange  them  in 
a  preliminary  rank  order.  Those  who  have  answered  the  most  number  of 
questions  favorably  or  "positively"  are  placed  at  or  near  the  top  of  the 
board  while  those  who  have  answered  the  most  number  -unfavorably  or 
"negatively"  are  moved  toward  the  bottom  of  the  board.  Between  the  extremes, 
respondents  are  ranked  in  such  a  way  as  to  approximate  a  parallelogram  in  the 
pattern  of  responses.  Various  rules  have  been  developed  to  bring  some 
uniformity  into  the  procedures  and  to  serve  as  a  guide  to  those  who  choose 
to  use  this  technique.  These  rules  of  thumb  include  weighting  responses, 
building  up  solid  streaks  in  each  column,  and  keeping  errors  close  to  the 
centre  column  (7c,  pp.  106-108).  The  aim  of  the  rules  and  the  arrangements 
is  to  minimize  the  errors  of  predictions  of  responses  based  on  ranks.  If  a 
number  of  response  categories  have  been  used,  it  will  probably  be  apparent 
at  this  stage  that  some  of  the  errors  could  be  reduced  by  combining  some  of 


' 


; 

' 

. 

' 

, 

; 

' 


. 


. 


. 

' 

*  : 

. 

: 

* 


■  • 


..  ■  :  1  ■ 


, 


I.  U 


66 


the  response  categories. 

Combining  Categories 

'  In  order  to  determine  which  categories  should  be  combined,  the 
balls  must  be  shifted  from  one  board  to  the  other  so  that  categories  can 
be  shifted  while  the  order  of  the  respondents  remains  unchanged.  The 
second  board  is  placed  over  the  first  one  so  that  the  slats  are  perpen¬ 
dicular  to  each  other;  the  boards  are  then  held  firmly  together  and  turned 
over  so  that  the  balls  fall  from  the  holes  in  one  board  into  the  correspond¬ 
ing  holes  in  the  other  board.  The  slats,  which  now  each  contain  one 
response  category,  are  then  shifted  so  that  the  categories  for  a  single 
question  are  adjacent. 

If  the  items  were  perfectly  scalable,  we  would  expect  that  when  a 
solid  streak  in  the  "positive"  category  ended  a  solid  streak  of  markers 
would  begin  in  the  "neutral"  category  and  when  this  ended  it  would  begin 
in  the  last  category.  In  the  perfect  scale  there  would  be  no  overlap  in 
adjacent  categories  and  no  ranks  would  be  omitted.  Empirically,  it  will 
probably  be  true  that  there  is  considerable  overlap  between  categories 
and  as  a  result  of  this  ranks  will  be  omitted  within  what  should  be  solid 
streams  of  markers.  This  will  be  particularly  so  if  there  are  more  than 
three  answer  categories.  What  overlap  and  missing  ranks  there  do  exist 
will  appear  as  errors  in  the  scalogram  pattern  and  so  an  attempt  is  made 
to  reduce  these  by  combining  categories.  In  many  instances  items  must  be 
dichotomized  in  order  to  establish  satisfactory  cutting  points  between 
categories.  Examining  the  pattern  visually,  plus  the  application  of 
criteria  suggested  by  Suchman  (7c,  pp.  108-115)  determine  which  of  the 


’ 

■ 

'  .  ■ 

■ 

■  ■:  : 

{ 

, 

' 

■  ■ 


- 

, 

' 

* 

' 

| 

'  «  • 

v:  nr!-  ;!' 

.  V  ‘ 

' 

. 

♦ 

67 


categories  should  be  combined. 

Final  Arrangement 

Because  response  categories  have  been  combined,  it  is  necessary 
to  reorder  the  respondents.  In  order  to  do  this  the  balls  are  transferred 
back  to  the  first  board  so  that  each  slat  again  contains  the  responses  for 
one  individual.  The  respondents  are  again  shifted  so  as  to  minimize  the 
deviation  of  the  obtained  pattern  from  an  ideal  pattern.  Now  that  the  number 
of  response  categories  has  been  reduced,  it  is  easier  to  see  what  effect  shift¬ 
ing  a  respondent  has  on  the  total  number  of  errors.  However,  the  same  criteria 
as  were  applied  in  the  initial  arrangement  can  be  used  to  determine  where  a 
particular  respondent  should  be  placed  if  there  is  some  doubt. 

Using  One  Board 

Although  there  are  obvious  advantages  to  having  two  boards  to  work  with, 
the  entire  procedure  can  be  carried  out  using  only  one  board.  When  only  one 
board  is  used,  it  is  not  necessary  to  have  it  constructed  so  precisely  nor 
does  it  have  to  consist  of  slats  with  one  hundred  holes  —  fewer  holes  and, 
indeed,  a  different  form  of  marking  responses  could  be  used.  It  does  mean 
some  additional  work  in  determining  which  response  categories  should  be 
combined,  but  the  final  result  should  be  identical  to  that  obtained  with 
the  use  of  two  boards. 

Testing  for  Scalability 

When  the  closest  approximation  to  a  parallelogram  pattern  has  been 
obtained,  the  scalogram  is  tested  for  the  scability  of  the  items.  This  is 
done  by  applying  the  criteria  which  have  been  outlined  in  a  previous  section 


•  -  ••  •  " 

,  .  •  '  .  '  .  •  ■ 

’  • 

■-  r  ’• 

■ 


•  ;;  •'  ■  •  ■;  . 


, 

.  ' 

■  ■  '  •;  ■  ’  '  ' 

t ....  -  •  ■  ■  ’  :  ■  ■  ■ 

■ 

■ 


6  • 


'  • 


'  ' 


68 

of  this  chapter.  The  number  of  errors  is  determined  by  counting  all  the 
markers  which  fall  outside  the  main  pattern.  In  order  to  determine  the 
coefficient  of  reproducibility,  the  following  formula  is  used: 

Coefficient  of  ^  _ Number  of  .Errors _ _ 

Reproducibility  Number  of  Number  of 

Questions  Respondents 

(7c,  p.  117) 

It  has  been  mentioned  previously  that  the  acceptable  lower  limit  for 
this  coefficient  has  been  placed  at  about  .90.  In  addition  to  the 
determination  of  this  coefficient  the  other  criteria  including  the 
number  of  answer  categories,  range  of  marginal  frequencies,  and  pattern 
of  errors  are  considered. 

Scoring  the  Questionnaires 

The  scalogram  picture  determines  how  the  entire  population  should 
be  scored.  Ideally,  responses  might  be  scored  on  the  basis  of  the 
perfect  scale  type  which  they  most  closely  approximate,  but  this  presents 
some  problems.  For  most  practical  purposes  it  has  been  found  that  a 
simple  weighting  system  correlates  to  a  sufficient  degree  with  the  more 
rigorous  procedure  mentioned  (7c,  p.  119)*  When  all  items  have  been 
dichotomized,  the  categories  can  be  weighted  simply  1  and  0.  Indeed, 
if  only  a  quasi  scale  exists,  this  is  the  only  alternative  for  the 
solution  of  the  scoring  problem. 

Discussion 


The  scalogram  board  technique  has  certain  limitations  in  terms 


1 


-CO 


-J& 


,  l  ■< 


.O.C  i  ■  : 


c  '  ...O'...  0 


£ 


: 


. 


■ 


££  '.  *  .  L  ■ _ LO  -  i  I 


J' 


c  £  ..  .  ..J'  ■  o  .  x> 


r  :  ■  j:  ■  fc-  '  ? '  :  •  c  •  ■:  h 


JLc- 


■ 


•  u'  -• 


o  C .  .  c 


19  >0  .  '  '  £1  •'  .  OC 


(  J.  r:.’c  o  on  \  15  ox  '  :•/  'o;  J'c  •  < 

;‘‘x  •  •  ■  c  .1-'  c  ■  5  >  ■>  ' 

O  '  :xoo  o.'  .  i  -  £ 

o  .  '  .  '  t  o';'  'r  n  '  '  C  •  .  •  '  Xo: 

o  £■  -  ,/  o  „  .  ..  >  c ..  1 

j:  ;■  c  •  .  3 

.  ■  .  £■  'i  i  o;  J  •'  x5  ;rj. 


o'  .  .  u  >  (**_'  _ L.  _  . 

'  ■  u  •  {l  ' : :  '  ! 

J  -  ■  :■  i 


-  u 


69 

of  the  physical  aspects  of  the  procedures  which  are  involved;  there  are 
also  inherent  limitations  in  basing  the  conclusion  that  a  series  of 
items  is  scalable  on  only  a  sample  of  the  population.  These  limitations 
have  been  overcome  to  a  considerable  extent  by  the  procedures  which  use 
cards  or  summary  statistics  and  are  thus  ahle  to  utilize  the  responses 
of  the  total,  population  available  in  determining  whether  or  not  the  items 
axe  scalable.  In  spite  of  its  inadequacies,  the  scalogram  board  does 
have  the  advantage  that  it  gives  the  researcher  a  visual  representation 
of  the  scale.  For  the  investigator  who  is  not  too  clear  about  the 
nalure  of  the  criterion  being  applied,  the  use  of  the  scalogram  board 
has  the  added  advantage  of  serving  as  an  invaluable  learning  activity. 

III.  THE  UTILITY  OF  SCALE  ANALYSIS 

The  process  of  scale  analysis  and  the  product  of  the  analysis 
have  a  number  of  distinct  advantages  over  some  of  the  more  traditional 
approaches.  It  is  these  advantages  which  maice  the  application  of  this 
particular  model  useful  in  dealing  with  qualitative  data.. 

Ie fining  a  Problem 

The  use  of  scale  analysis  can  help  to  define  a  problem  if  it  is 
found  that  a  universe  actually  consists  of  a  series  of  subuniverses 
rather  than,  the  single  one  hypothesized;  the  problem  can  then  be 
reformulated  in  terms  of  these  new  dimensions  (7d,  pp.  144-150)* 

In  this  way  scale  analysis  helps  to  show  when  there  are  differences 
of  kind  as  well  as  of  degree  between  items.  It  can  do  this  because 
scale  analysis  provides  a  powerful  test  of  the  internal  structure 
of  the  series  of  items  undergoing  analysis.  Furthermore,  if  it 


; 


r  v 


;r  • 


V  £ 


C'O;..' 


CO  co.:;cp:c:  -s  .vv.v  nx 

:  .  .  :  1  ;j 

CO  .  C .  >■;  ..  v  o  *  MV 

•  ■  t  ■  s  c  \  £  i 


v  v'v  .  .or  •.  •  :>  : 

:  t  >  i-  '  '  evB 

J;  J  '  u  c  v  v.  -  o 


o  ;  o  o  ■  .  c  !  v  .  ■  _wv  .  v  .  .  w  >  u'  :o 

;  0  -  lc  ■  £  £1  ££  I  .6-  •' 

.  Xu  :  I<-j  r  ■  ;  ■  &  d  1  \':  cv  £  d  £  Yb&  i 

.  .  .  o  J'  v;'  v  x  vo  o  .  o.  c  .  .  1  c  > 

.  : o  v  c  ire:  :  o'  ■. 

c  '  •  .  .  .  .  ;>  •  V)  ■  £  o'  .  - 


.  i 


- ,.r  •  b  c  ■  •  V  .o  •  x  ■  .!'  • 

h  x  .  ■-  '  ;  J  '  vr 

■  ?  c  '  ■  o'  :  c  J:  '  ■  •  c  -  ■  v  >  • 

,  J  .v.cc  i  r  ■ 


C  C  •  r  •  ■  M  V 

■■  c  J-  o  ■  vA  v  -  y;' 

v  ,  .  ■  c  •  vvqqs 

.  *  G  .  c.  1 


:x  /  ~  '  ■  v1  r  J"; 


o  v  c  i  j 


o  .  .  O  '.  '■  V 

r 


o’  . (.  !  i 


.  £  1  ’  0 


'  X  X  . 

£  •  ■  d 


•  -  J’.-.  or  c.  ■  J  ::  .. 

‘O  '  ;  ■  -•  ..  J  box  j.  ) 

>  q  5  • : '  V  ’  C  c 

.  .I  ■.  j:  ■  o' 


70 


can  Toe  shown  that  a  series  of  items  is  scalable  then  what  was  apparently 
complex  has  been  simplified  to  some  extent  by  the  uncovering  of  some  order 
or  structure,  and  knowing  that  the  items  are  scalable  gives  clues  as  to  how 
responses  to  items  used  individually  might  be  interpreted.  This  has  proven 
to  be  of  value  in  the  field  of  public  opinion  polling. 

Implications  of  Scores 

If  a  series  of  items  has  been  shown  to  scale,  then  immediately  there 
are  a  number  of  further  advantages  which  accrue  to  the  investigator  (7d, 
pp.  153-155)*  The  first  of  these  is  that  the  score  which  a  person  obtains 
on  the  items  is  an  accurate  description  of  his  attitude  with  respect  to  those 
items.  It  also  means  that  persons  who  obtain  the  same  score  have  responded 
similarly  and  could,  therefore,  be  expected  to  be  similar  in  the  character¬ 
istics  which  the  verbal  responses  might  represent.  Furthermore,  if  A  is 
higher  than  B,  it  means  that  A  has  answered  all  questions  favorably  which 
B  has  answered  favorably  plus  a  few  more.  The  general  implication  being 
that  all  individuals  have  been  ranking  along  the  same  continuum. 

Prediction 

A  scalable  universe  also  has  implications  for  prediction;  if  a 
universe  is  scalable  it  makes  sense  to  use  it  as  either  the  predictor  or 
the  criterion  in  a  prediction  problem.  What  is  more,  if  the  items  are 
scalable  one  should  obtain  the  same  result  by  using  one  question  from  the 
universe  a3  by  using  another;  however,  it  seems  reasonable  that  the  greatest 
accuracy  will  be  obtained  by  using  the  whole  scale.  Suchman  states  that  if 

an  outside  variable  is  to  be  predicted  from  the  scale  items,  the  multiple 


. 

. 

.  •  '  '  ■  " . 


71 

correlation  on  the  scalable  area  is  equivalent  to  the  zero-order  correlation 
with  the  scale  score  (7d,  p.  156).  This  is  true  as  long  as  there  is  only 
one  dominant  variable  present  even  if  the  reproducibility  is  not  high  as 
in  the  case  of  quasi  scales. 

IV.  SOME  PROBLEMS  IN  SCALE  ANALYSIS 

There  are  a  number  of  problems  which  enter  into  scale  analysis  and 
which  affect  the  utility  of  scales  both  in  describing  and  in  making  pre¬ 
dictions.  Not  the  least  of  these  is  the  extent  to  which  scales  can  be 
obtained  in  actual  practice;  it  may  be  that  in  much  work  investigators 
who  are  interested  in  using  the  technique  will  have  to  be  satisfied  with 
approximations  to  scales.  Even  when  an  area  is  found  to  be  scalable  there 
are  problems  related  to  the  reliability  of  the  sampling  of  people,  of  items, 
and  of  trials  (7e,  p.  277)* 

Sampling  of  People 

The  problem  of  the  sampling  of  people  enters  the  picture  because  the 
pretests  and  scalogram  analysis  are  carried  out  on  about  one  hundred  people 
and  from  this  the  assumption  is  made  that  the  same  results  would  have  been 
obtained  had  the  entire  population  been  included.  More  specifically,  by 
testing  the  scalability  of  a  series  of  items  for  a  sample  of  the  population, 
conclusions  are  drawn  concerning  the  scalability  of  the  area  for  the  whole 
population.  The  hypothesis  is  not  one  concerning  perfect  scalability  for 
in  this  case  any  sample  drawn  in  any  way  would  be  sufficient  to  refute  the 
hypothesis.  In  this  instance  the  hypothesis  is  one  of  approximate  scalability, 

and  since  we  can  be  fairly  certain  that  there  is  some  error  in  the  population, 


■  ' 

' 


, 

. 


. 

to  -  :r  ’  '  " 

' 

■  ;  '  ' 

.  ^ 

' 


•  •  ■  ;;  Vv  •  •  • 


- .  , . 

: 

■ 


■ 


. 

.  ' 


72 


there  will  also  be  some  error  in  the  sample  that  is  drawn  from  the  popu¬ 
lation.  The  major  concern  is  the  extent  to  which  the  scalability  of  the 
items  for  the  sample  represents  the  scalability  of  the  items  for  the 
population.  This  is  difficult  to  deal  with  because  the  sampling  theory  of 
scale  errors  is  as  yet  -undeveloped.  In  spite  of  this  difficulty,  there  is 
good  reason  to  believe  that  the  higher  the  coefficient  of  reproducibility 
for  the  population,  the  less  will  any  sample  drawn  from  the  population 
depart  from  this  coefficient  (7e,  pp.  279-280).  Guttman  offers  mathematical 
and  empirical  evidence  to  support  the  contention  that  if  there  is  an 
approximate  scale,  there  will  be  relatively  little  variation  in  the  amount 
and  the  pattern  of  errors  for  a  series  of  samples  drawn  from  the  population 
(7e,  pp.  280-283). 

Sampling  of  Items 

It  has  been  discussed  earlier  that  the  concept  of  a  universe  of 
attributes  from  which  items  are  selected  as  samples  to  test  the  hypothesis 
of  scalability  is  central  to  the  theory  of  scale  analysis.  There  is  then 
also  a  problem  of  how  reliable  is  this  selection  of  items.  The  ideas  of 
random  sampling  cannot  be  applied  here  because  the  items  are  not  selected 
by  any  random  process;  they  are  constructed  by  the  investigator  who  chooses 
a  particular  form  and  a  particular  wording  for  the  item.  The  main  question 
is  whether  or  not  the  items  which  have  been  selected  represent  the  universe. 
Again  it  is  not  difficult  to  refute  the  hypothesis  that  the  items  are  perfect, 
but  it  is  not  as  easy  to  refute  the  hypothesis  which  is  aimed  at  something 
less  than  perfection  (7e,  p.  286).  The  question  then  is  how  imperfect  an 

item  can  be.  In  order  to  determine  whether  or  not  an  item  should  be 


' 


'( 

:  .  ■  ■ : 

-  ■  ■  •  ■ 

. 

1  ■ 

. 

" 

•  . 

■  ■  ■■ ■  ' 

•  ' 


.  .  r. r-,',  .  -V;:  <  S  o;’  I  ■:  i?.  . 

' 

:  ,;o  :  '  t 

. 


73 

included,  two  criteria  have  been  used  in  practice  (7e,  p.  287).  The  first 
of  these  is  that  each  item  separately  should  have  high  reproducibility  and 
the  second  is  that  the  scale  error  must  he  at  the  most  half  of  what  it  would 
he  without  knowledge  of  the  scale  pattern.  Of  course,  these  two  are  closely 
related  for  if  the  reproducibility  is  high,  the  scale  error  is  bound  to  be 
low.  It  may  be  necessary  to  apply  both  criteria  in  cases  where  there  are 
few  responses  in  one  category.  By  employing  these  two  criteria,  it  can  be 
fairly  certain  that  the  items  do  represent  the  universe  which  is  under  investi¬ 
gation.  If  items  depart  considerably  from  these  and  in  addition  have  grouped 
errors,  i.e.,  non-scale  types,  then  it  can  be  fairly  certain  that  the  item 
does  not  belong  in  the  "universe  or  has  been  constructed  in  a  faulty  manner. 

Still  in  connection  with  items  is  the  question  of  the  extent  to  which 
the  ranks  obtained  by  people  on  the  sample  of  items  would  compare  with  the 
ranks  obtained  on  other  or  on  all  possible  items  (7e,  p.  296).  It  is  obvious 
that  the  correspondence  can  never  be  perfect  because  the  number  of  distinct 
ranks  depends  upon  the  number  of  items  used.  The  ranks  obtained  on  a  few 
items  are  groupings  of  the  ranks  that  would  have  been  obtained  on  the  universe 
of  items,  but  from  the  ranks  of  the  sample  it  is  not  possible  to  draw 
inferences  concerning  specific  ranks  on  the  universe  of  items. 

Guttman  advances  arguments  to  the  effect  that  if  the  reproducibility 
is  high,  then  the  ranks  obtained  by  people  when  different  items  are  used  must 
be  fairly  stable  (7e,  p.  296).  This  can  be  investigated  by  correlating  part 
scores  on  the  sampling  of  items;  that  is,  by  correlating  scores  on  half  the 
test.  Since  the  marginal  frequencies  differ,  it  is  not  possible  for  the 
correlations  to  be  perfect,  but  one  can  expect  the  reliability  to  increase  as 
the  number  of  items  increases  (7e,  p.  JOO).  The  number  of  items  which  are 
retained  in  the  final  study  depends  to  some  extent  upon  the  reproducibility 


> 

1  ■ 

y.:  '-  . 

.  :  '• 

'  '  .  ■  '  ■-  »  '  '  ' 

* 

... 

...  ■  ■ 

■  ,  ■  .  "  •  '  .  ■  /;  V  V  ; 

■»  ' 

;  •/  £ 

'  ' 

0  21 

:  '• 

.  • 

■  ’  •  • 

■ 


74 

and  upon  the  number  of  ranks  which  are  desired.  If  the  reproducibility  is 
high  and  if  relatively  few  ranks  are  required,  then  relatively  few  items 
will  need  to  be  included. 

Test-Re test  Reliability 

One  other  question  of  reliability  to  be  considered  is  that  due  to 
variability  within  a  given  respondent.  Although  the  question  of  variability 
due  to  different  items  has  been  discussed,  this  is  not  the  same  as  asking 
what  the  persons  score  would  be  likely  to  be  if  he  were  given  the  same  items 
over  again.  Guttman  advances  a  number  of  arguments  to  enable  him  to  conclude 
that: 

For  the  case  of  scale  scores  then,  as  well  as  for  the  qualitative 
responses  to  the  separate  items,  we  have  assurance  that  if  the  items 
are  approximately  scalable,  then  they  necessarily  have  very  substantial 
test-retest  reliability.  Scalogram  analysis  provides  as  an  automatic 
byproduct  the  assurances  that  responses  to  individual  items  and  total 
scores  both  have  relatively  little  error  of  measurement  if  the 
reproducibility  is  high.  ('Je,  p.  3 ll) 

Summary 

The  preceding  sections  have  indicated  that  according  to  the  theory  of 
scale  analysis,  if  the  coefficient  of  reproducibility  is  high  there  is  good 
reason  to  believe  that  there  cannot  be  much  error  in  the  sampling  of  people, 
in  the  sampling  of  items,  and  that  there  would  also  be  high  test-retest 
reliability.  When  approximate  scales  depart  considerably  from  the  ideal 
pattern  there  is  more  room  for  sampling  variation  and  each  of  these  factors 
of  unreliability  must  be  dealt  with  in  specific  ways  or  be  left  in  doubt. 


i  ' 


. 

. 


■ 

. 


, 

. 

, 


■ 

•> 


REFERENCES  FOR  CHAPTER  IV 


Ake,-  James  N.  "Rapid  Machine  Procedures  for  Determining 
Scalability  of  Any  Number  of  Questions,"  Public  Opinion 
Quarterly,  26:121-125,  Spring,  1962. 

Ford,  Robert  N.  "A  Rapid  Scoring  Procedure  for  Scaling 
Attitude  Questions,"  Public  Opinion  Quarterly,  14:507- 
552,  Fall,  1950. 

Green,  Bert  F.  "A  Method  of  Scalogram  Analysis  Using  Summary 
Statistics,"  Psychometrika,  21:79-88,  March,  1956. 

Guttman,  Louis.  "The  Cornell  Technique  for  Scale  and  Intensity 
Analysis,"  Educational  and  Psychological  Measurement ,  7: 

247-279,  Summer,  194? 

Riley,  Matilda  White,  John  W.  Riley,  Jr.,  and  Jackson  Toby. 
Sociological  S tudi es  in  Scale  Analysis .  New  Brunswick: 

New  Jersey:  Rutgers  University  Press,  1954* 

Stone,  Carol  Larson.  "A  Machine  Method  for  Scaling  as  Many 
as  Twelve  Dichotomies,"  Station  Circular  #29.  Pullman, 
Washington:  Washington  Agricultural  Experiment  Station, 
Institute  of  Agricultural  Sciences,  State  College  of 
Washington,  August,  1958.  (Mimeographed) 

Stouffer,  Samuel  A.,  et  al.  Measurement  and  Prediction. 

Vol.  IV  of  Studies  in  Social  Psychology  in  World  War  II. 

4  vols.  Princeton,  New  Jersey:  Princeton  University 
Press,  1950» 

(a)  Louis  Guttman,  "The  Problem  of  Attitude  and  Opinion 
Measurement,"  Chapter  II,  pp.  46-59* 

(b)  Louis  Guttman,  "The  Basis  for  Scalogram  Analysis," 

Chapter  III,  pp,  60-90. 

(c)  Edward  A.  Suchman,  "The  Scalogram  Board  Technique  for 
Scale  Analysis,"  Chapter  IV,  pp.  91-121. 

(d)  Edward  A.  Suchman,  "The  Utility  of  Scalogram  Analysis," 
Chapter  V,  pp.  122-171. 

(e)  Louis  Guttman,  "Problems  of  Reliability,"  Chapter  VIII, 
pp.  277-311. 


■ 

'  f  •  •  •  ’ 


.  '  .  " 


? 

- 

. 

. 


-  . 

, 


•  ;•  ■ 

» 


' 


: 

♦ 

. 


:  ■  ;  :  i  •,  O  :  &\  '  c  <;  l  f  ’• : 

. 


. 


. 


. 


CHAPTER  V 


SCALE  ANALYSIS:  CRITICISMS,  DEVELOPMENTS,  AND 
APPLICATIONS 

This  chapter  outlines  some  of  the  shortcomings  which  have  become 
apparent  in  the  scalogram  model  and  the  procedures  associated  with  it.  Much 
of  this  adverse  criticism  is  no  doubt  justified,  and  it  is  also  only  fair 
to  say  that  much  of  the  same  criticism  attempts  to  be  constructive  and  is 
designed  to  extend  the  utility  of  the  procedures.  Perhaps  the  most  that  can 
be  said  in  favor  of  the  entire  approach  rests  with  its  demonstrated  empirical 
value  which  is  discussed  in  the  final  sections  of  the  chapter. 

I .  SOME  CRITICISMS  OP  SCALE  ANALYSIS 


It  would  not  be  an  understatement  to  say  that  scale  analysis  has  been 

met  with  something  less  than  enthusiasm  in  many  circles.  One  reason  for 

this  may  be  that  those  who  have  done  extensive  work  in  psychometrics  may  feel 

that  there  are  more  appropriate  models  for  the  data  with  which  they  work. 

In  the  early  1950 1 s  when  there  appears  to  have  been  much  interest  in  the 

technique  among  some  social  scientists,  Guilford  (13)  indicated  the  various 

shortcomings  that  had  been  mentioned  by  other  writers: 

...  it  is  pointed  out  that  the  criterion  of  scalability  is  rarely 
achieved,  even  when  total  scores  reach  an  acceptable  level  of 
reliability.  The  criterion  is  variously  described  as  "unrealistic , " 
"useless,"  and  even  "harmful."  Even  when  the  criterion  is  achieved, 
it  is  not, certain  that  we  have  a  univocal  score.  The  score  may 
represent,  instead,  a  rather  uniform  combination  of  two  or  more 
factors.  There  is  no  really  effective  way  of  selecting  good  items 
by  this  approach.  Much  depends  upon  the  investigator's  wisdom  in 


' 

..  .  .,  ,r  £T  •  '  •  '  •  '  ' 

,  •  ;•  '•  '  '■  '  •..!  ::  i  •:  •  • '  '/  ’>f  ■ '  '  r’ 

■ '  o  '  ’■  v  ” ;  1  :  •  f 

.  ■  ■  ; '  ‘ :  c  '■  v  t 

<■(*  V'  ■  ’  '■  '  i  :  i-  ’ . ' 

.  -  •  '  '  ■  '  ••  • '  '  '  '  ’• 


r  ri  i  -  r  -  •-  ;-r 

,  '  ■  - 


.  .  - 


77 


regard  to  the  items  he  puts  in  the  analysis  process.  The  procedures 
themselves  lack  rigorous  rules  for  combining  response  categories  and 
for  counting  errors  of  reproduction.  Reproducibility  is  related  to 
response  popularity.  When  responses  pile  up  in  one  category, 
reproducibility  cannot  fail  to  be  high.  This  sets  a  lower  limit  to 
what  is  often  a  restricted  range  of  nonchance  reproducibility,  a 
limit  that  may  on  occasion  be  uncomfortably  close  to  the  criterion. 

The  method  also  favors  groups  of  items  that  turn  out  to  be  virtually 
rewording  of  the  same  content,  in  which  case  the  variable  emphasized 
could  well  be  a  specific  factor  rather  than  a  common  factor.  (13,  p.461) 

It  is  not  surprising,  in  view  of  these  criticisms,  that  Guilford  suggests 

that  the  Thurstone  and  Likert  approaches  to  the  measurement  of  attitudes 

are  superior  to  the  Guttman  technique. 

In  evaluating  the  Guttman  process  Peaks  (22)  points  out  that  there 

are  ambiguities  in  certain  steps  in  the  procedures.  For  example,  when  the 

scalogram  technique  is  used,  there  is  room  for  a  considerable  amount  of 

subjectivity  in  the  shifting  of  rows  and  columns.  In  addition  to  this  she 

states  that  it  must  be  recognized  that  the  coefficient  of  reproducibility 

is  in  itself  a  function  of  the  items  selected  in  the  first  place.  Peaks 

also  recognizes  the  problems  in  generalizing  to  some  universe  on  the  basis 

of  the  scalability  of  a  specific  set  of  items;  perhaps  all  that  can  be  said 

is  that  a  given  set  of  items  scale  under  a  given  set  of  conditions  and  for 

a  specified  sample.  A  further  question  concerns  whether  the  method  is 

suitable  for  the  inves tigation  of  all  sorts  of  psychological  variables  or 

perhaps  only  for  those  in  which  the  answer  to  one  question  implies  an  answer 

to  another  and  where  they  are  obviously  related.  If  this  is  so,  then  the 

unidimensionality  which  comes  through  may  only  be  the  felt  need  on  the  part 

of  the  respondents  to  be  consistent  (22,  p.  265)* 

Many  of  the  objections  and  criticisms  voiced  by  Peaks  and  Guilford 

have  been  raised  more  recently  by  other  writers  as  have  some  newer  ones. 


•  -  '  ■  ..  .. 

'  )'.  :• :'f.  turc/  :  r;  f. :: 

.  '  ' 


I  :  •  '■  ■'  ! 

'/  •;  h 

t '  1  • 


, 


,  -  ■ 

...  '  ♦ 


■ 


■ 


78 

Different  writers  have  taken  various  opportunities  to  attack  scalogram 
analysis  on  numerous  points,  "but  most  of  them,  such  as  those  which  will  be 
discussed  in  the  next  section,  have  attempted  to:  be  constructive. 

Universe  of  Attributes 

Although  Campbell  and  Kerckhoff  (4)  are  quite  impressed  with  the 
development  of  scale  analysis;  they  also  indicate  that  they  recognize  some 
attendant  difficulties.  These  writers  are  most  concerned  with  the  concept 
of  a  universe  of  attributes  and  are  doubtful  as  to  whether  a  universe  can 
be  said  to  be  scalable  if  a  sample  of  items  from  the  universe  is  scalable. 

They  note  that  the  judgment  of  content  is  not  subject  to  verification 
within  the  limits  of  scale  theory.  If  a  scalable  universe  does  exist  then 
it  should  be  relatively  simple  to  suggest  additional  scalable  items  once 
the  universe  has  been  identified.  Unfortunately ,  the  fact  that  judges  have 
often  been  unable  to  add  such  items  to  existing  scales  casts  some  doubt  on 
the  assumption  that  the  judgments  about  the  content  have  been  accurate.  In 
view  of  this,  it  may  be  difficult  to  speak  of  a  universe  of  items  beyond 
those  used  in  the  scale.  They  argue  further  that  since  it  is  the  respondents 
who  determine  the  meaning  of  a  certain  item  only  they,  through  their  responses, 
can  indicate  whether  a  new  item  belongs  in  the  scale.  Thus  it  seems  entirely 
possible  that  a  universe  of  content  might  be  scalable  but  is  not  found  to  be 
so  because  of  the  particular  items  which  are  selected  by  the  judges. 

This  may  mean  that  the  universe  of  attributes  must  be  defined 
operationally  in  terms  of  the  items  that  are  known  to  scale  plus  other  items 
as  yet  unknown.  Campbell  and  Kerckhoff  also  suggest  that  if  a  set  of  items 


■ 

■  ■  •.  ■  i 


_  ... 


:  ■  '  •'  r 


’ 


. 


■  .  .!( 


■  1 

■ 


79 


scale  they  do  have  a  single  meaning  for  the  subjects,  and  that  for  this 
reason  they  form  a  universe  of  attributes  for  the  subjects.  This  is 
opposed  to  scale  theory  which  states  that  content  alone  and  not  scalability 
defines  a  -universe.  But  it  would  not  seem  unreasonable  to  state  that  if 
a  set  of  items  have  been  judged  to  be  of  common  content,  and  then  if  they 
do  scale,  it  could  be  taken  as  further  evidence  of  their  unidimensionality. 
However,  as  Campbell  and  Kerckhoff  state  this  does  not  enable  us  to  say 
anything  more  about  the  temporal  or  spatial  dimensions  of  the  universe  of 
attributes.  In  spite  of  their  critique  the  writers  do  not  suggest  that 
they  are  in  any  way  challenging  the  usefulness  of  scale  analysis,  and  in 
fact  they  suggest  that  scale  analysis  may  be  more  useful  than  has  been 
claimed  heretofore  (4,  p.  303). 

Item  Construction  and  Arrangement 

One  of  the  suggested  values  of  scalogram  analysis  has  concerned  its 
usefulness  for  selecting  poll  questions.  Steiner  (30)  hypothesized  that 
the  scalability  of  a  series  of  items  presented  in  random  order  in  a 
questionnaire  would  differ  from  the  scalability  of  these  same  items  if 
they  were  grouped  and  presented  under  a  title.  To  test  this  hypothesis 
he  administered  three  sets  of  scales  with  the  items  grouped  and  then  also 
in  random  order  to  ninety-one  members  of  an  undergraduate  psychology  class. 
He  considered  this  to  be  a  conservative  test  of  the  hypothesis  for  all  were 
aware  that  they  were  being  asked  to  respond  to  the  same  questions.  Each  of 
the  scales,  whether  presented  in  homogeneous  or  mixed  context,  produced 
a  satisfactory  coefficient  of  reproducibility,  but  there  were  a  number  of 
non-scale  types.  He  found  that  the  coefficient  of  reproducibility  of  the 


'  \i 

»  ' 


.  •* 

; 

• 

• 

• 

v 

, 

■ 

80 

mixed  items  were  not  appreciably  different  but  that  the  rank  order  of 
the  respondents  obtained  by  the  two  methods  were  by  no  means  the  same. 

As  a  result  of  this,  Steiner  concludes  that  it  would  be  unwise  to  assume 
that  the  same  results  are  obtained  when  items  are  presented  isolated  as 
when  they  are  presented  in  context.  However,  he  does  ignore  the  fact  that 
he  never  did  satisfy  all  the  criteria  for  scalability,  but  it  is  interesting 
to  note  that  what  criteria  he  did  employ  were  not  changed  markedly  when 
items  were  presented  in  different  contexts. 

Scheussler  (28)  was  also  concerned  with  the  effect  that  the  investi¬ 
gator  has  on  determining  whether  or  not  an  attitude  area  is  scalable  through 
the  influence  that  he  has  on  the  construction  of  items.  In  his  view,  the 
selection  of  the  items  is  biased  in  the  sense  that  it  is  not  independent 
of  the  person  who  is  drawing  the  sample.  In  order  to  study  this  effect  he 
used  thirty- five  items  constructed  by  five  different  individuals  on  the 
topic  attitude  toward  the  Negro.  All  items  were  administered  to  146  students 
on  the  same  questionnaire.  Scale  analysis  of  the  responses  revealed  that  the 
hypothesis  of  scalability  was  not  confirmed,  by  any  of  the  five  sets  of  items. 
Scheussler  concludes  that  investigators  may  reach  contradictory  conclusions 
regarding  the  scalability  of  what  superficially  at  least  seems  to  be  the 
same  topic.  Among  the  special  investigator  tendencies  which  may  well 
influence  scalability  are  the  following:  social  perspectives,  tendency  to 
require  a  negative  reply  to  show  a  favorable  response,  asking  questions  on 
either  the  personal  or  institutional  level,  and  changing  the  order  of 
alternatives.  Scheussler  also  questions  the  practice  of  combining 
response  categories.  In  an  empirical  3tudy  the  responses  of  the  same 


' 

.  'X  / 

- 1  ■ 

V .  V 

)•  .  'j  ■  :<!  '  1  ■  ■'  ' 

• 

,  ,  '  .  I\;  - 

' 

r  "... 

■ 

'  re  •  ’’  '■/:  '  '  '  ■ 

'  -1  ■'  < 

'  .  .  - 

■  .  . 

•  ;•  :  '  '  ' '  ' 

*  '' 

- 

;>  ■  ■  >  ’  >’ 

V  ..  .  ■ y. 

, 

• 

81 


group  of  individuals  were  obtained  twice  for  the  same  questions;  the 
first  time  three  response  alternatives  including  "uncertain”  were  provided 
and  the  second  time  only  two  response  alternatives  were  given.  Each  set 
was  subjected  to  scale  analysis  and  in  this  process  the  first  set  of 
responses  was  dichotomized.  When  the  two  were  compared  it  was  evident 
that  in  only  two  of  the  seven  items  were  the  shifts  in  responses  in  the 
direction  anticipated  by  combining  categories.  As  a  result  of  this 
Scheussler  indicates  that  there  is  a  need  for  more  objective  means  of 
selecting  items,  and  that  combining  response  categories  may  make  it  difficult 
to  interpret  the  structure  of  the  attitude  area. 

The  effects  of  changing  the  response  alternatives  for  the  same 
questions  were  investigated  by  Stover  (34)  who  found  that  although  the  items 
he  used  scaled  in  each  instance,  there  were  marked  differences  in  the 
marginal  frequencies.  This  leads  him  to  suggest  that  the  response  alterna¬ 
tives  may  influence  scalability.  It  should  be  noted  that  the  theory  of 
scale  analysis  does  recognize  the  fact  that  marginal  frequencies  are  subject 
to  sampling  fluctuations,  and  so  Stover's  finding  may  be  only  a  verification 
of  this  rather  than  any  evidence  that  the  response  alternatives  provided 
make  a  difference  in  scalability. 

The  views  of  Stover  are  supported  by  Henry  and  Borgatta  (14)  who 
also  suggests  some  other  ways  in  which  the  scalability  of  a  series  of  items 
might  be  improved.  These  include  using  the  same  response  alternatives  for 
all  questions,  using  the  same  cutting  point  in  all  response  alternatives 
when  categories  are  combined,  etc.  Some  of  these  views  are  contrary  to 
the  requirements  proposed  by  scale  theory,  but  it  may  well  be  that 


: 

:  ::  '■  ' 

•  .  ;<  ■  '  ;  ' 

.  -  - 

"■  :  ;  /'..'•no  "'<.3  os! 

-■  > t '  '  '■  ;  ■  '  '  '  '' 

o  •;  r  •  ■  ■  .3  ,  •  >r  •  holdVQ  bOQfJ  brf 

'  '  .  •  i"  '  «  ■"  1 

.  ’  1 "O 

1 

. 

* 

’ 


82 

scalability  can  be  enhanced  by  the  use  of  these  techniques. 

Interviewer  Effect 

In  a  study  in  which  investigators  interviewed  lower  class  Peurto 
Rican  families,  Back,  Hill,  and  Stycos  (2)  examined  the  effect  which 
interviewers  had  on  the  scalability  of  the  responses.  Their  findings 
suggest  that  the  reproducibility  of  the  responses  obtained  by  interviewers 
who  were  careful  about  the  mechanical  aspects  of  the  recording  were  higher 
on  scales  requiring  only  such  recording,  but  that  interviewers  who  under¬ 
stood  the  purposes  of  the  study  obtained  data,  which  had  higher  repro¬ 
ducibility  on  the  more  subtle  scales.  Back  et_  al_  suggest  that  if  one  is 
aware  of  these  effects  one  might  well  select  the  interviewers  according 
to  the  types  of  scales  which  will  be  used.  The  value  of  keeping  inter¬ 
viewers  informed  as  to  the  nature  and  importance  of  the  study  is  also 
demonstrated. 

Coefficient  of  Reproducibility 

In  addition  to  the  criticisms  mentioned  above,  the  value  and 

meaning  of  the  coefficient  of  reproducibility  has  also  come  into  question. 

As  Green  (ll)  points  out,  the  Guttman  model  is  not  a  probability  model, 

and  thus  the  statistical  fluctuations  which  are  to  be  expected  remain 

obscure.  He  states  further: 

It  has  been  suggested  that  the  scale  be  established 
on  the  basis  of  a  sample  of  100  cases.  Since  the 
statistical  theory  of  the  random  fluctuation  in 
reproducibility  is  unknown,  this  seems  to  be  a 
dangerous  extravagance,  (ll,  p.  356) 

Other  writers  have  been  quick  to  suggest  different  procedures  which  might 

be  used  to  evaluate  obtained  scales.  Thus  there  have  been  proposed 

measures  such  as  the  coefficient  of  scalability  by  Menzel ,  ( 19 )  the  error 


u  i  ; 


.  -  ...  • 


'I'X  ■ 


x.  0  •  w  .... 


•  < 


,!  J;  ,  . ..  ■  1  t  ■  ■ . ■  •'  '.to 

i.i  .  /  ■  •  ti  - '  •  ■  J.  ox 

-  >  f  ■  ■  .  t  •  .  ■  ■  ; 


.  •  ■  '  .  >i'  .. 

,’:o  i  ■ 

.  X  .  •  . '  •/  c.  :  ■  t  '/  )!  i 


^  ' 


’  ■  J  -o; .  ko 


-  (  • .  ■  •  ,  :  i  j;  ■  ..  - 

,  ;x  j-  ■  ;  t  l  ,  .  ;  :  .  ;  v:  -to  ,  ■ 

too  •  ■  •  •  J  t.’:  civ  .  I.  '-  >' i  ...  .1  o  ..  .  o  o  :  >  '>t"< 

o.':  .  O  •< 

,  .  ■  1  $  i  lo  o  x  «  %  '  ’&  ■  '  ■ 

.  .  J-  -  -  t 


'  ■.  't  i  .  .':cx  i o  ■;  t  . 

o  of  I  l  XO.  O 

•  »  .  l  .  '  '  ‘  5  ^  u  .  -  - 

■  •  •  *  •  .  i  -L  I.  . 

0  C  '■  v  I  .  ...  JO'.  .  V  ...  .• 


..  y 


■■  •  •  o. 


..... 


: 


•  ;  .  - 


i  .;  ,  x  s;  . 


t  ,  j 


.  o 


, 


-V-  P 


o.:;  . 


O  -M  . .  .rr  ... 

Vi,  '  P  ' 


83 

ratio  by  Borgatta,  (3)  and.  the  plus  percentage  ratio  by  Pearson  (23).  Some 
of  these  innovations  do  not  really  solve  the  basic  problem  because  their 
sampling  fluctuations  are  equally  as  obscure  as  those  of  the  coefficient 
of  reproducibility.  Furthermore,  most  of  these  measures  have  not  been  used 
to  any  extent  in  reported  empirical  studies,  and  so  one  does  not  even  have 
the  opportunity  to  compare  one’s  results  with  those  obtained  by  other 
investigators. 

Davis  (7)  is  of  the  opinion  that  the  usual  things  in  which  an  investi¬ 
gator  is  interested  cannot  be  represented  by  a  single  measure.  Although  the 
coefficient  of  reproducibility  is  designed  to  give  some  indication  of  the 
extent  to  which  responses  are  reproducible  from  the  total  score,  investigators 
are  usually  not  concerned  with  being  able  to  reproduce  the  responses.  They 
are  more  generally  concerned,  with  problems  of  validity,  internal  consistency, 
and  reliability.  In  view  of  this  he  suggests  that  the  coefficient  of 
reproducibility  does  not  serve  as  useful  a  purpose  as  might  some  measures 
which  are  more  suitable  for  each  of  these.  Davis  believes  that  validity  is 
not  indicated  by  the  coefficient  of  reproducibility,  and  that  to  get  at  this 
it  is  necessary  to  use  an  empirical  correlation  with  some  outside  criterion 
or  perhaps  to  fall  back  on  some  operational  definition  based  on  the  items  used. 
Similarly,  he  suggests  that  some  test  of  association  might  be  used  to  test  the 
internal  consistency  of  a  series  of  items,  and  that  a  test  for  the  significance 
of  differences  between  correlated  proportions  might  be  used  to  check  the 
reliability  of  sets  of  items.  Which  specific  tests  will  be  employed  will 
depend  upon  whether  the  emphasis  is  on  testing  for  scalability  or  on  the 


construction  of  an  instrument. 


>  fi  > : 

'  ‘  ■  '  "  • 

' 

-  : 

• 

'  •  ;  '  J 

i  o  /  '•  1  .  ' 

l  '  r  i  •'  ..  r  ''  '  > o  ::  '  •  . 

r.  '■  - : ■  ■■  '  v-  :  r.  \  •  c  '■  ; 

» g  :  ' 

.  '  '  .  1  "  •  •  • :  » 

•  '  '  : '  • 

•;  •  :  :::  )  ■  /  :  r  j  c  ‘  9  r  5  j  >  '  -  'x  'xoirikrxq&x 


;  •  •  ■;>  .  •  :  ■ 

•  ■  '  ■  /. 


o  •  : 

. 


84 


II.  SOME  DEVELOPMENTS 

In  spite  of  the  early  criticisms  that  were  levied  against  it,  the 
idea  of  scalogram  analysis  has  not  disappeared  from  the  literature  on 
methods  in  the  scaling  of  attitudes.  It  ha,s  been  modified  to  some  extent 
and  changes  in  it  have  been  suggested,  but  there  is  every  indica.tion  tha,t 
theoretical  interest  will  be  maintained  at  least  as  long  as  the  method 
proves  to  be  useful  in  the  quantification  of  qualitative  variables. 

Edwards  (8)  considers  scale  analysis  not  so  much  as  a  means  for  con¬ 
structing  scales  as  for  evaluating  sets  of  sta/tements  to  determine 
whether  or  not  they  fit  the  particular  model  proposed  by  Guttman.  Thus 
he  thinks  of  it  being  used  in  conjunction  with  some  of  the  more  conventional 
methods  for  the  construction  of  scales.  Edwards  has  also  found  applica¬ 
tions  for  the  extension  of  scale  theory  into  scale  discrimination 
techniques  and  in  the  so-called  H-technique  for  the  improvement  of 
cumulative  scales. 

Torgerson  (38)  deals  with  the  entire  field  of  scaling  from  a 
theoretical  and  mathematical  approach.  He  considers  all  scaling  methods 
as  falling  into  two  broad  classes,  namely,  the  judgment  and  the  response 
methods.  In  the  former  variability  is  assumed  to  be  associated  only  with 
the  stimuli  while  in  the  latter  the  scaling  procedures  attempt  to  account 
for  variability  in  both  the  stimuli  and  the  respondents.  For  this  reason 
the  models  themselves  are  more  complex  than  those  associated  with  the 
judgment  methods.  The  response  methods  can  be  further  divided  into  those 
which  are  based  on  deterministic  and  those  which  are  based  on  probabilistic 
models . 

Deterministic  models  state  the  ideal  case  which  is  perhaps  never 
realized  in  practice;  there  is  no  room  for  error  variance  in  the  model, 


'  . 


.  •  -  d'.":.1  d'  •  ‘ .  ■  v  •  ■■  *  -- 

l'  .  d  :  !  .  0  •  1  ■  ■  '  ■  -  -  • 


d  d:  ■ 


U 


,  'd  ■  ,  .  d'  '  •  -  '  >'  •'  ■'  '  '  :  ....  C  i.' 

.  •  xx  v  id  5  •  ±Xo.  -  -  xli  '  --  - 

■  ;!•  .  L...  O'  0  .  .  ■  ■ 

'  •'  '  .  ;  .  .u  L  I' 

:  ■  :/c  .  V  ■  X .  :  .  •  •  -  ' 


L 


J  v:  .  I 


■■  Y  .  ;  ;■  X  1  J1  '■  - 


t  X' . 


' :  :  .n;r. 7.  o  ■  .  ■  . 

, :  xd'  '  o 

.  J-. ;  .  •  v  x  '  ■' 


■  it')' 


;  .  -  X.  O:  '  d 


Jo:  )  '  . '  x  d' 

•.  d"  uh  < 


:  ■  ■  x  .  ‘  v’  ' 


&  •  y.u  '  :  did' 


- : 


d  -  :  J'.’  O-  )"  .  '  o  0 

.  ■  X:"c  •  vxx  :ro 

,  d‘  d'X  .  :  v  .  ;  '  •'  ,! 

..  •  .•  y  0  d.d  •> 

c  i"  d  . : 


i;.1.  O'.  .  ■ 

)  j  0  ‘  ]  GO  -■  -  '  '  '  —  ■  ! 

*  ,  *  *  C  v.  0  o  -  >  / 

r  :  •  r  ■  -  .  x: do  d.  •'/  ■.  /  "  • 

d: 

■  Jo  o.Jo/  d  '.O;  r  d: 


.  .  ■■  .  o'  ..o o  .  ■  . 

J.  .  00  •  d  Jo  ■  d  d'*-'  gov  ;  •\U; 

.  ■:  '  r  '  .  " 


o  ■■ 


. x  d  :  ’ . 


. 


;  . .  0.0  . - 


vJ  —  I. ' 


:  :  .  1  1 


:  x  J:  d.  •  ; 


85 

and.  the  scaling  method,  determines  whether  the  data,  closely  approximates 
the  model.  The  probabilistic  models  have  provision  for  error  variance 
and  have  built  into  them  some  system  for  determining  goodness  of  fit. 

In  this  way  the  model  gives  some  objective  basis  for  accepting  or  reject¬ 
ing  the  scaling  hypothesis.  'The  Guttman  model  is  an  example  of  a  deter¬ 
ministic  model  while  Lazarsfeld’s  latent  structure  model  is  an  example  of 
a  probability  model.  It  is  interesting  to  note  that  the  Guttman  model 
can  be  considered  to  be  a  special  case  of  the  latent  structure  model 
(52,  p.  50). 

The  response  scaling  methods  suffer  generally  from  the  fact  that 
most  of  the  work  has  been  done  on  the  models,  and  very  few  applications 
have  been  made  to  empirical  data.  This  is  particularly  true  of  probabil¬ 
istic  models.  Torgerson  stales  that  empirical  use  of  the  Guttman  model 
involving  the  coefficient  of  reproducibility  and  the  auxiliary  criteria 
have  often  led  to  the  conclusion  that  a  series  of  items  is  approximately 
scalable.  .Among  the  reservations  which  he  has  is  that  this  has  often  been 
done  on  the  basis  of  only  a  few  items  and,  in  addition  to  this,  that  the 
coefficient  of  reproducibility  is  a  deceptive  index.  Often  coefficients 
which  are  only  halfway  between  unity  and  that  expected  by  chance  on  a 
hypothesis  of  independence  are  considered  to  indicate  scalability.  This 
suggests  that  the  ideal  Guttman  model  may  not  fit  empirical  data  too  closely. 
It  may  be  more  realistic  to  think  in  terms  of  the  covariance  among  items  as 
being  attributable  to  their  separate  relations  to  an  underlying  continuum 
rather  than  to  the  variance  of  each  (56,  p.  425)*  Torgerson  concludes  tha.t 
much  more  experimental  and  analytical  work  on  these  methods  is  needed. 

The  developments  which  have  taken  place  in  connection  with  handling 


-  •  ■  •  3  iX  oi  % 

l  x  \  •  '  •  :  .  :  0.:x  ■ 

.  /  ±7  '  if  ‘‘  .V  '  '■  X  V  X  ' 


-  •  X  :  \  -  x  o'  eX  -1  : 

-  ;  •  .  .  v.:-  x  .  1  •  ■'  •  ••  ■  ■  •••  ; 

.  io  ,  -  x.:.  -  ;  '  :  '  -  • 

•.  .  -  >  '  • . 

.  Xc  :■  ■  t 

*  «  5 

;;  o  .  '•  '  •  .  •  •  X  O  '  C  • 

..v  ;  ...  '  .  ,  .v  -  ‘  :  ’  ■  :  O':. 

•  ,  •  '  c  ..i '  '  x  .  «  :  ox  . -  ■  ■  •  ' 

■  .  .  cx  o  ■  '  •  °  '  ; 

•  ;  •  •  '  X.' X  .'A  '.  V. 

.  o  ■  j-ij  ■  b.c  i . ; i vu  x  j :  ■  v:  x  . 

;  .  j-  t  •  x  -  :  - 

r  \  ?  . 

,  ...  .  .  X  ■  ;  -  ■  x  X.':.  x.  xo  '  '  •  :  '  c 

£  ;  £  f  ■  ; 

X  .  X  ■  X.  x  '  ,  J:  X  ■  ■  x  c  ?.  '  ’’  w 

■  I  ■  s ..V  .  sp«atfd  J  t  •'  j  3  •  r,  ' 

.  _•  ;  .;  •.  J;  .  • 

o  ;  .  o  ,  "Xu 

.  .  .  •’/  ■' 

)  -!  o.:.  ,.  ■  . 

U  XXX  ■ 


86 


data  using  the  Guttman  model  have  been  mentioned  in  the  previous  chapter. 

These  include  the  machine  procedures  used  by  Stone  (31),  Ford  (9)?  and 
Ake  (l);  the  summary  statistics  approach  by  Green  (10);  and  also  a  data 
reduction  method  which  is  outlined  by  Danielson  (6),  With  the  model  itself 
there  appear  to  be  promising  developments  in  such  areals  as  its  relationship 
to  latent  structure  models  (32)  and  possible  extensions  into  multidi¬ 
mensional  scale  analysis. 

III.  APPLICATIONS  OP  SCALE  ANALYSIS 

Extensive  use  was  male  of  scaLogram  analysis  by  the  Research  Branch 
of  the  U.S.  War*  Department  during  and  after  World  War  II.  Suchman  gives 
evidence  of  scales  being  developed  in  areas  such  as  satisfaction  towards 
one's  army  job,  attitude  toward  officers,  attitude  towards  postwar  con¬ 
scription,  knowledge  of  current  events,  and  fear  symptoms  (32,  pp.  122-142). 

The  scales  which  were  developed  in  these  areas  consisted  of  as  many  as  eleven 
to  as  few  as  five  items  most  of  which  were  dichotomized;  the  coefficients  of 
reproducibility  which  were  obtained  ranged  from  .98  to  .89,  and  in  general 
those  scales  which  had  the  greater  number  of  items  also  had  the  lower 
coefficients  of  reproducibility.  This  is  evidence  contrary  to  the  assertion 
that  few  scales  are  found  empirically.  It  should  be  mentioned  that  the  final 
scales  were  obtained  after  a  number  of  pretests  which  were  not  always  successful 
in  identifying  scalable  areas.  Although  the  Research  Branch  did  attempt  to 
scale  morale,  it  found  that  this  was  not  a  unidimensional  area  nor  was  it  one 
that  could  be  investigated  with  a  quasi  scale.  Instead,  it  seemed  to  consist 
of  a  series  of  subareas  which  in  themselves  were  scalable.  The  inter- 
correlations  of  the  scale  scores  revealed  that  there  was  no  single  common 


/,  ^  ,  ,j  •,-{  .  i  -<.)•  :  ■  ju  ■•••I  J"  ii.  ' 

■  ■  ■  : v  C’l  :•  •  91  d'  j  ' 

;  (  ;  '■  1  d  ‘  ,  •  1  ■■  • 


^r 


«  .  - 


i  j;  ;■  .  v  1  '  '  ■  -  '  ■'  ^  1  0«. 


ion  ■  :  o*  •  i  -  '  ■  J  -  -  '  ' 

,  ■  v.‘.  .i  ■  i  oi  \  "  '  ■'  '  '  -  -  ■' 

,  J-  :  -  J:..;  o  •  c  '  >■  •  v  .  "  •  ;  .  •  ; 

c  •  ,  •  -  .  •  l-  ;  <  '  ' 

.  ....  ?  ..  r  .!  ,  •  -  ’£  ?  i 

■  r  .  .  .  .  L  '  '  . '  .1  . 1  i'1  '  £■  •  '■■■' 

c.:.  .  :  •  ,  ,  •!  J  K  I  ■  ID  ..  c  SJ. 

,  .  '  ■  •'  ■  • 

> 

•  o  oi  i;  lx 

J-  ■;  ■'  ■  •'  »'  ..  'C.:  i'::  1  .  .c  •  ■  • 


X  - 


87 


factor  present  and  supported  the  idea  that  morale  was  not  a  scalable  area. 

A  quasi  scale  with  a  coefficient  of  reproducibility  of  .73  was  found 
to  exist  in  the  form  of  a  psychoneurotic  inventory  (32,  p.  I64).  Here  the 
scale  pattern  did  not  indicate  that  this  might  be  broken  down  into  scalable 
subareas.  Lack  of  scalability  did  not  prove  to  be  a  great  disadvantage;  it 
just  required  that  more  items  be  retained  in  the  scale  than  would  otherwise 
be  necessary. 

Checking  Existing  Scales 

It  has  been  suggested  by  Edwards  that  one  purpose  for  which  scalogram 
analysis  might  be  used  would  be  to  check  existing  scales  which  have  been 
developed  by  other  means  in  order  to  determine  whether  they  approach  the 
Guttman  model  (8,  p.  177)*  Case  (5)  was  interested  in  whether  the 
radicalism-conservatism  battery  developed  by  Centers  was  actually  working 
with  a  single  dimension  when  defined  in  this  way.  To  explore  this  he  used 
a-  samiple  of  forty-four  individuals  out  of  441  who  had  completed  the  six 
items  in  the  battery.  When  three  response  categories  were  retained  in  each 
item,  the  coefficient  of  reproducibility  was  .81,  and,  in  addition  to  this 
low  reproducibility,  four  of  the  six  items  had  more  error  than  non-error. 
When  some  categories  were  combined  so  that  there  were  four  dichotomies  and 
two  trichotomies  the  coefficient  rose  to  .88,  but  now  there  was  evidence  of 
some  non-scale  types.  Although  Case  does  not  draw  attention  to  it  in  his 
article,  further  doubt  i3  cast  on  scalability  by  the  existence  of  extreme 
modal  categories  in  two  of  the  six  items.  As  a  result  of  his  analysis,  Case 
concludes  that  the  scale  looks  at  an  attitude  of  dominant  proportions  plus 


one  or  more  of  lesser  proportions. 


'  1  :  • 

'  ■  ' 

' 

: 

.  r:  ;<  •  ’  ;  .1 

\ 

.  -  ,  ;  .  ■  3  •  ■  •  '  '•  7  .  •  '  gtf  7  -  ■■ 

' 

,  ■  .  y  ■  7  7  l  '  -r  ■  '0  :■  7  "  ■  M  '  :■  l 

. 

- 

.  7  '  - 

,  :T  ,  .  ■ 

(,  t 

„  ••  ’  >  ■  ■  ■  : 

•  •  >•; 

.  . 

;  7  7  • ) 

• 

■•■■■<  v  • 

.  ' 

.  7 

•  ■  V  .  > 

V  V  .<  -  '  i 

.  ■  .  • 

88 

Case  does  recognize  that  scales  are  relative  to  time  and  sample  and 
agrees  that  it  might  have  been  better  to  have  used  Centers'  own  data  rather 
than  the  data  which  he  collected  in  order  to  test  his  hypothesis.  It  is 
unfortunate  that  "research”  such  as  this  is  even  reported  for  it  shows  a 
lack  of  concern  for  the  most  basic  features  of  the  scalogram  technique. 
Guttman  is  cautious  of  using  as  few  as  one  hundred  individuals  in  his 
sample  let  alone  the  forty-four  which  Case  used.  In  addition  to  this 
violation  of  procedures,  Case  overlooks  some  of  the  other  criteria  for 
scalability  which  the  theory  includes.  It  would  seem  that  whether  the 
conservatism-radicalism  battery  is  scalable  is  still  a  good  question. 

Data  From  Interviews 

An  investigation  carried  out  by  Wallin  (38)  tapped  an  area  different 
from  that  mentioned  in  the  other  studies  reported  here  and  also  employed  a 
different  data-gathering  method.  The  problem  was  to  investigate  the 
neighborliness  of  women  by  interviewing  to  obtain  information  concerning 
the  type  and  the  frequency  of  interaction  with  their  neighbors.  A  total 
of  481  interviews  were  carried  out  and  the  data  analyzed  using  the  Cornell 
technique.  The  analysis  of  thirteen  items  yielded  a  coefficient  of 
reproducibility  of  .9175  there  was  one  non-scale  type  which  was  eliminated 
when  one  item  was  removed  from  the  scale.  In  the  final  scale  twelve  items 
had  a  reproducibility  of  .92,  and  Wallin  concluded  on  the  basis  of  the 
criteria  which  had  been  applied  that  the  twelve  items  did  form  a  scale. 

In  studying  the  social  psychological  and  family  structure  of 
fertility  control  behavior  among  lower  class  families  in  Peurto  Rico,  Back 


■  ■  '  •>  :  ; 

1 

;  ’  v  j  J:  ;<.■:!  o  •  •  /:  r  orTj 

, 

. 

. 

« 

, 

- 

,  .  /  •  .  •  ■  -•  ' 

.  r"  '  ‘  "  oJi 

. 

T  '■  '•  Li, 

. 

- 


89 


and  his  associates  (2)  made  use  of  four  scales  for  which  information  was 
also  obtained  by  interviews.  The  scales  are  of  interest  here  in  terms  of 
the  number  of  items  and  the  reproducibility  of  each  scale.  This  information 
is  given  following  the  name  of  the  scale:  activities  prohibited  to  wife 
(5  -  .952),  communication  with  spouse  (5  -  *904)?  communication  with 
friends  (4  -  .895)?  and  modesty  (6  -  .896).  It  is  not  too  clear  from  the 
report  what  method  of  analysis  was  used,  but  it  seems  certain  that  the 
number  of  subjects  was  at  least  one  hundred.  In  considering  the  scales 
themselves,  it  would  appear  that  the  reproducibilities  are  satisfactory  but 
that  the  number  of  items  in  each  scale  is  fairly  low. 

Church  Orthodoxy 

Vernon  (37)  was  interested  in  studying  church  orthodoxy  as  defined 
by  the  extent  to  which  institutional  norms  find  expression  in  either  overt 
or  covert  behavior.  His  study,  which  was  based  on  a  sample  of  194  indi¬ 
viduals,  was  designed  to  determine  whether  church  orthodoxy  so  defined  was 
scalable.  Although  he  started  with  thirty-six  items,  he  reduced  this  to 
twelve  after  a  pretest.  He  found  that  it  was  necessary  to  dichotomize  each 
of  the  items  in  order  to  obtain  a  reproducibility  of  .89  and  concluded  that 
there  was  a  good  possibility  that  such  scales  could  be  used  in  further 
studies.  Vernon  used  different  response  categories  for  different  items; 
it  would  be  interesting  to  note  what  affect  uniform  response  categories 
would  have  had  on  the  reproducibility. 

Urban  Structures 


In  a  study  which  involved  the  assessment  of  the  utility  of  aerial 


'  ■  •  V. 

, 


■ 


' 

' 

. 

. 

!  ,1  • 

*  • 

1  . 

' 

■ 

.  •  > 

■ 

: 

1  •:  ;  J-  ■ 

' 

[ 

.  * 

'  '  '•  : 

'  •"* 

.  .  :  -  '  ;  -  .  ;  '  ’  '  ;  '' r  *'•  •  '  " 
,  ,  ■  ■  /  .  ■  '  ■  :  •  ;  •  •  ■  '  ' 

C  • 

, 


90 


photography  in  collecting  data  for  sociological  and  ecological  research, 

Green  (12)  attempted  to  rank  twenty-eight  census  tracts  on  a  residential 
desirability  continuum.  The  four  items  which  he  used,  namely,  zonal 
location,  prevalence  of  single  family  homes,  dwelling-unit  density  per 
block,  and  general  land  use,  together  contained  eleven  categories.  Using 
the  Cornell  technique  he  carried  out  a  scale  analysis  which  produced  a 
coefficient  of  reproducibility  of  .93  for  twenty-eight  census  tracts. 

Green  also  constructed  a  scale  of  five  trichotomous  items  dealing  with 
socio-economic  status  and  again  carried  out  a  scale  analysis  which  this  time 
resulted  in  a  reproducibility  of  .92. 

The  significance  of  this  study  lies  in  the  fact  that  this  use  of  scale 
analysis  is  somewhat  novel;  however,  it  also  has  serious  limitations  in  that 
the  number  of  subjects  in  the  sample  was  far  below  that  for  which  the 
coefficients  are  usually  calculated  in  the  scale  analysis.  Since  the 
sampling  distribution  for  the  coefficient  is  unknown,  using  fewer  than  one 
hundred  cases  makes  comparison  with  other  results  very  difficult  as  has  been 
pointed  out  previously.  This  study  does  suggest  that  scale  analysis  has 
applications  even  where  it  is  not  a  matter  of  sampling  items  from  a  universe; 
in  this  study  the  items  themselves  define  the  universe  of  interest. 

Student  Attitudes 

McDill  ( 18 )  investigated  the  relationship  between  the  content  of  an 
attitude  and  the  intensity  of  that  attitude,  but  his  study  is  also  of 
interest  here  because  of  the  scales  which  he  used.  These  were  (l)  Attitiides 
Toward  Quitting  School.,  and.  (2.  Attitudes  Toward  the  Honesty  of  Policemen 


•- 

, 

.  ■:  ■  •  .  r.  1 -« 

.  ;■ 

•» 

»  .  ‘  " 

■  '  •  '  "  ' 

' 

.  ' 


.  .  . 


. 


-•  '■ 

•  r  ■  ■ 

' 

'  '• 


•  ■  ■  . 


91 

for  which  he  obtained,  reproducibilities  of  .9 6  and  .95  respectively.  Even 
though  these  coefficients  were  high,  the  scaies  are  not  nearly  as  impressive 
as  it  might  seem  because  each  scaie  contained  only  four  questions  and  the 
analysis  revealed  some  non-scale  types.  The  evidence  that  more  than  one 
dimension  was  present  did  not  seem  to  enter  into  his  considerations,  and 
this  is  further  evidence  that  in  actual  use  some  of  the  criteria  set  forth 
in  scale  theory  are  violated  or  ignored  without  arousing  much  concern  in 
some  in ve s t ig at or s . 

Voting  Behavior 

The  study  discussed  above  used  scalogram  analysis  in  the  more 
traditional  area,  of  attitude  measurement,  but  Schubert  (27)  offers  a  more 
unique  application  in  the  study  of  the  voting  behavior  of  Supreme  Court 
justices.  He  states  that  scalogram  analysis  provides  a  technique  for 
arranging  those  ca.ses  which  have  the  same  content,  and  the  justices 
themselves,  in  a  uniquely  determined  rank  order.  In  the  study  fourteen 
cases  of  a  particular  kind  and  the  justices  involved,  which  also  happened 
to  be  fourteen,  were  found  to  scale  perfectly.  This  would  seem  to  contra¬ 
dict  the  popular  notion  that  perfect  scalability  is  una.ttainable  in  actual 
practice.  Before  coming  to  this  conclusion,  one  must  bear  in  mind  that 
only  fourteen  subjects  were  involved  and  not  all  of  these  voted  on  all  of 
the  cases;  that  is,  those  justices  who  were  appointed  only  recently  voted 
only  on  recent  cases.  The  high  scalability  can  be  accounted  for  by  the 
fact  that  the  justices  do  make  a  deliberate  attempt  to  be  consistent. 
However,  the  liberties  which  Echubert  takes  with  the  method  makes  it 
doubtful  whether  one  should  consider  this  to  be  an  example  of  the  use  of 


■  •  i  <  '  =  ' 

:  .  .  .  v-  .  .  '  •  ’• 

j  .!  . ..  .  ■  ■  ■'<  '  .  : 


•  i  -ii ■  or,  X  • 1  -  :  > ■ 

■  ;  '  .  .  ’  ■ ;  . 

.  -  ;  v  > 

' 

'  ■  ■  -  c  .  .  •  •:  • 

•  .  ?.  '  .  ’  i  .  .u  ‘  ‘u.i 

'  t  ■  j  i:  c  T  .  ■  '■ .  if 

,  .0  '  c .  ' 

/  :  c  ::  '  •  v  I:  ■  J"  -  •*:’  ,  ' •  c >'  • 

v  i  A  V  -  W  '  J  . 

;  ■  0  '  .  i.1'  /  •  '  k 

.  ;  >i  :■ 

.  :  .  = 

t  v  X 

.  v.'  c.':.'  .  j  •  * 

;  j. 1  i  • '  •  •  ':•> 

-  ;■  c  •/  "O'... 

?  r ; J" 

;  •  j\'  .  ■  '  J;  .  c • .  ;h  cl* "•  ;•  no. 

CitOi  ■  ■  .  1 

•  -  .  .  -  Oil  ; 

lo  II  .  .  $  vXovnx  ©lew  s; 

'  g  1  .  c  j 

i  ;i‘  .*..■•••  ■  c  ■  .  ■  ,  . 

<  .  1  ■  "  • 

.  .  ■  >  c  ,  c  if-  J:  r 

.  ■;  J’  G  '  ..  ■ 

o  ■  .)•  ■  I'  •  Is.  : 

:  •  : 

'  ;  g  :  c 

f:  ’ •  .  '  --.i.'.'  < 

r  <  ■  i  •  j'  1  c  '  •  • 

.  ;  0  .  v  •  > . 

92 


scale  analysis. 

Delinquent  Behavior 

The  scalability  of  delinquent  behavior  among  institutionalized 
delinquents  and  also  among  high  school  students  was  examined  by  Nye  and 
Short  (20).  They  began  with  twenty-one  items  which  were  reduced  to  nine 
during  the  scale  analysis.  The  scale  analysis  resulted  in  a  coefficient 
of  reproducibility  of  .78  when  some  items  were  left  with  trichotomous 
categories.  The  number  of  items  was  reduced  to  seven  because  of  extreme 
marginal  frequencies,  and  these  were  found  to  scale  satisfactorily  for  the 
non-institutionalized  students  but  not  for  the  institutionalized  delinquents 
for  whom  a  scale  of  more  serious  offences  would  have  been  necessary.  This 
offers  evidence  that  a  set  of  items  may  be  scalable  for  one  sample  but  not 
for  another.  The  investigators  felt  that  they  had  been  successful  in 
demonstrating  unidimensionality  in  this  series  of  behaviors. 

Using  the  above  study  as  a  point  of  departure,  Scott  (29)  further 
investigated  the  dimensionality  of  delinquent  behavior.  He  hypothesized 
that  there  were  differences  between  individuals  whose  delinquent  acts  were 
directed  against  anonymous  persons  or  corporate  property  and  those  whose 
activities  were  directed  against  their  own  primary  group.  A  fifteen  item 
scale  was  administered  to  a  group  of  undergraduate  sociology  students  and 
scale  analysis  using  the  Cornell  technique  was  carried  out  on  a  sample  of 
one  hundred.  The  results  showed  eight  of  the  items  to  scale  with  a 
reproducibility  of  .90  while  four  of  the  remaining  items  definitely  did  not 
seem  to  fit  in  with  those  which  were  scalable.  The  manifest  difference  in 
content  3eemed  to  be  whether  or  not  the  behaviors  were  related  to  the 


, 

•  ,  ;  .  •  .  .X  •"  -  "  '  '  - 

. 

;  ■:  '  ■  i 

-  '  '  :r  : '  '  '  ' 

, 

■ 

>i ' 

. 

;  ' 

• 

1  1 

‘  ■  ’ :  ! 

,  0  +  •  "•  * 

93 


individual's  interpersonal  relations.  To  investigate  the  relationship 
between  the  two  scales,  the  total  scores  were  correlated  and  were  found  to 
yield  a  Pearsonian  r  of  .16  which  would  indicate  that  the  relationship 
between  the  two  scales  was  not  significant. 

Cone ept  Development 

Strauss  (35)  investigated  the  application  of  scale  analysis  to  concept 
development  in  monetary  meanings  where  he  felt  that  there  might  be  a  fairly 
regular  pattern  of  advance.  In  his  study,  seventy-one  questions  were 
administered  by  interview  to  sixty-six  children.  The  items  were  scored 
using  an  arbitrary  weighting,  and  scale  analysis  of  the  Cornell  technique 
was  used.  Strauss  concluded  that  with  few  exceptions  the  items  were  scalable, 
and  that  it  was  meaningful  to  place  the  children  in  a  rank  order  on  the  basis 
of  their  total  scores. 

Strauss  used  a  different  basis  for  establishing  cutting  points  in  view 
of  the  large  number  of  items;  cutting  points  were  placed  closest  to  the  point 
where  there  was  a  shift  in  the  responses  to  a  number  of  items.  He  admits 
that  this  is  the  same  as  discarding  some  of  the  items  which  have  similar 
cutting  points,  but  that  for  his  purposes  he  wished  to  retain  all  of  the  items. 
The  investigator  suggests  that  scalogram  analysis  is  useful  to  show  how 
monetary  meanings  at  a  given  level  are  built  upon  previous  levels  of  develop¬ 
ment.  His  study  would  have  had  more  methodological  value  in  the  field  if  he 
would  have  attempted  to  determine  the  scalability  of  only  a  few  items  rather 
than  all  of  those  which  he  had  available. 

Aggressive  Fantasy 

One  of  the  few  psychological  applications  of  scale  analysis  is 


,  ■  ;  [  •  ■  ■  '• 
. i  ■  -  '  -  ir 


•  ‘ '  •  '  \  :  i* 


''  '  -  :  .  • 7  ff.<  ;t 7  r 


. 


.  .  -  -  • 

" 

-  ■  •  .  "  '  -7:  <  7-  ‘  ■ 


'  '7 

'  ■■ ■  ■  ■ . 

.  :  7‘  '  •  <  :  •  '  '  '  j  • : 


- 


.  .  :  '  v  .  '■  '■  1 

,  '■  ■  7-7  :  ■  ■  •  '•  .  7  c  ■ 

.  '  v  .  ),  * .  ■  7  :  * 


94 


reported,  by  Lesser  (17)  who  attempted,  to  apply  the  criteria  of 
scalability  to  aggression  responses  of  a  sample  of  pre-adolescent  boys. 
The  stimulus  presented,  to  the  boys  was  a  series  of  ten  pictures;  the 
responses  were  coded  according  to  the  degree  to  which  indications  of 
aggressive  tendencies  were  present.  As  a  result  of  the  study,  Lesser 
found  seven  of  the  pictures  to  stimulate  scalable  responses „  The 
investigator  felt  that  he  had  been  successful  in  ranking  not  only  the 
boys  according  to  the  degree  to  which  they  gave  aggressive  responses  but 
also  the  pictures  according  to  the  extent  to  which  they  evoked  such 
responses. 

Object  Scales 

Riley  and  her  associates  (24,  25,  26)  have  extended  the  use  of 
scalogram  analysis  to  object  scales  in  which  the  items  are  not  verbal 
responses  about  attitudes  but  are  some  characteristics  of  people,  for 
example  their  influence  in  a  group  or  their  popularity  in  a  group. 
Similar  procedures  have  been  used  to  scale  the  characteristics  of 
groups  (24).  The  same  studies  have  investigated  the  group  phenomenon  of 
consensus  as  being  analagous  to  the  individual  characteristic  of  in¬ 
tensity  (25).  There  is  not  much  evidence  in  the  literature  that  any 
extensive  development  has  taken  place  in  any  of  these  areas. 

IV.  SCALE  ANALYSIS  IN  THE  STUDY  OF  ROLE  CONFLICT 

At  least  two  studies  have  been  carried  out  in  which  the  method  of 


scalogram  analysis  was  used  to  investigate  the  problem  of  role  conflict. 


:  '  '  •  ■  '  '  '  '  '  '  ‘  :  '  " 

■ 

,  ' 

. 

' 

'■  ■  ■  •  ■: 

■ 

. 

;  '  ■  ' 

■ ;  1  "  ' 

•  '•  ■  •  ■  '  ■ 

•  :  •  :  . 

.  ' 


95 

The  first  of  these  is  reported  by  Stouffer  and  Toby  (35)  who  sought  to 
provide  a  link  between  the  study  of  social  norms  and  the  study  of  person¬ 
ality.  The  specific  problem  which  they  investigated  concerned  the 
question  of  whether  some  individuals  had  a  personality  bias  toward  one 
type  of  conflict  resolution  while  others  were  biased  toward  another  type 
of  resolution  in  similar  situations  where  there  was  a  lack  of  consensus 
as  to  the  proper  thing  to  do.  The  conflict  which  was  investigated  was 
that  arising  from  one's  obligations  to  a  friend  and  one's  obligation 
to  general  society  in  a  specific  situation;  that  is,  the  particularistic - 
universalistic  dilemma  was  studied.  A  number  of  hypothetical  role  conflict 
situations  were  constructed  and  responses  were  obtained  which  indicated 
the  extent  to  which  individuals  tended  to  be  particularistic  or  uni¬ 
versalistic  in  each  of  these.  Four  items  were  found  to  scale  with  a 
coefficient  of  reproducibility  of  .91?  but  there  wa.s  evidence  of  some 
non-scale  types.  Further  analysis  was  also  carried  out  in  which  the 
general  approaches  of  the  latent  structure  model  were  employed.  Although 
the  investigators  recognized  the  limitations  of  their  study,  they  did 
conclude  that  the  study  suggested  that  it  was  possible  "to  classify  people 
according  to  a  predisposition  to  select  one  or  the  other  horn  of  a  dilemma 
in  a  role  conflict."  (33?  P*  404) 

The  application  of  the  scalogram  analysis  technique  to  the  study 
of  role  conflict  was  also  explored  by  Laulicht  (16)  who  used  the  pattern 
variable  theory  as  a  source  of  possible  dimensions  of  role  conflict  for 
studying  the  degree  to  which  individuals  were  consistent  in  resolving  role 


■  ■  '  1  ■'  '  of  .r /  1  ' 

;  :  ’  .  -  -  ..I  .  •  • 


o': S'  l  s  ■.  : 

< .  S  ■ ;  '  ! 


•  ; 


r 

V  v 


..'j 

Q 

•  -  c 


.  \  '  k  "  .»  S.C  ' '  ,  .  ‘  ’  .  -L  .1: 

1  oiJ  .  •  ' 

>  1 

u  ;  ■  .• :c  :>  J.  ' 

’ 

ifr  ..  y  j  i  Jt  i  i 

■  S  ■  :  ■ . '  .  i  s  .  :  iJ  5 

■  ^-Y- 

^ ■  S  '  •'  : 

■  '  /. '  Jj  •  • :  •  ■  ~  - 

•  .  ■  :  J  :  •:!  '  .  ’.i  V  •  '■  '  ?  X  '  ■  : 

.  j  f  '  ■  '  >  :  r  i 


'  \  '  '  ; 


*. 


■  ■  ■  J  '  I  'If' '  C  '  -  '  5  0 ! : 

.  .  y  I  of: I  ;  "  ■  •'  ''  '  •'  '  oo  ■  >!• 


.  i f.1 ; ,  ..  .  .  •  ■  ■ 


i-  '  S. 


96 


conflict.  Two  of  these,  the  universalistic-particularistic  and  the 

specific -diffuse  choice  dilemmas,  were  used  to  provide  a  basis  for  the 
construction  of  role  conflict  situations.  Laulicht  hypothesized  that 
although  the  dimensions  as  proposed  by  Parsons  and  Shils  (2l)  would 
not  be  scalable  there  would  be  scalable  sub-areas  within  the  dimension  . 
Scalability  would  be  dependent  upon  holding  either  the  situation  constant 
for  different  alters  or  holding  alters  constant  for  different  situations. 

When  responses  were  obtained  for  hypothetical  situations  it  was 
found  that  these  were  not  scalable  when  alters  we re  held  constant  but  the 
situation  was  changed.  This  is  contrary  to  the  findings  of  Stouffer  and 
Toby  reported  above  who  concluded  that  very  similar  items  were  scalable 
(15 »  33) •  Laulicht  did  find  that  when  situational  contexts  were  held 
constant  and  alters  were  allowed  to  vary,  the  responses  satisfied  the 
criteria  for  the  presence  of  a  scale.  He  concluded  that  he  had  been 
successful  in  identifying  dimensions  of  role  conflict  even  though  the 
tests  of  the  scale  hypothesis  were  not  always  rigorous.  Furthermore,  he 

felt  that  the  scale  analysis  approach  had  been  a  useful  and  economical 
means  for  detecting  offending  questions  and  seems  to  be  satisfied  with  its 
suitability  for  this  particular  purpose. 

The  two  studies  cited  above  differ  to  some  extent  from  the  study 
which  is  reported  in  this  thesis  in  that  while  the  former  were  more  con¬ 
cerned  with  the  resolution  of  the  conflicts,  this  study  was  more  concerned 
with  the  dimensionality  of  the  expectations.  The  difference  in  final 

outcomes  may  be  more  apparent  than  real  for  the  basic  purpose  is  the 


' 


•  ‘  .  ■ 


•  * 


.1.  t  ,  -  *  * 


.  •  ■'  v  •  ■:  ' 

. 


- 


•  v  .•  ■  .1-  ' 


r  "  M 


97 


same,  namely,  to  study  the  structure  of  the  variables  which  are  involved. 
In  studying  role  expectations  in  formal  organizations  it  seems  more 
appropriate  to  study  the  structure  of  the  expectations  before  specific 
attention  is  given  to  the  manner  in  which  different  individuals  resolve 
the  conflicts. 

V.  CONCLUSION 

The  criticisms  of  the  technique  and  the  studies  discussed  in  this 
chapter  lead  one  to  conclude  that  even  though  many  people  have  been  chary 
of  the  model,  scabe  analysis  has  proven  itself  to  be  of  empirical  and 
heuristic  value  in  a  variety  of  circumstances.  Many  of  the  quanti¬ 
fications  would  have  been  difficult  to  achieve  without  scale  analysis; 
for  others  it  served  to  clarify  the  nature  of  the  phenomena  under  investi¬ 
gation.  Of  main  concern  is  the  fact  that  in  some  instances  the  basic 
criteria  and  cautions  which  Guttman  emphasized  were  either  misinterpreted 
or  ignored.  This  is  unfortunate  from  the  point  of  view  of  the  effect  tha.t 
it  has  on  the  acceptance  of  the  technique  generally.  No  doubt  the  problem 
arises  because  many  investigators  are  so  anxious  to  arrive  at  the  stage  of 
hypothesis  testing  that  they  tend  to  ignore  some  of  the  more  basic  steps 
during  the  process  of  quantification. 

The  fact  that  scale  analysis  has  been  found  to  be  suitable  in  a 
variety  of  circumstances  including  the  study  of  role  conflict  suggests 
that  its  use  in  the  particular  problem  investigated  in  this  study  is 


justified. 


.  I  •  '  '  :  .  -  --  -  '  -  i 


'it  ,':J‘  ©t  " .  ■  i. "  ■  .  -  - 

'■  •  J:-  ;  •• 

.  .  ::  '  , 

v ’•  '■  .  .  ■ 

C-j '  ■■ 

'  >.  V 

:  .  ■  o'  Yi 

' 

•.  .  '  '  /  '  ■ '  i  '■ 

o  X'  ,•  c  ;.::o 

:  .  ;  .  '  ' 

x  .  .  '  '  V:  i 

-j  ...  .  :  :  .  '  '  ■' 

.  v  ■  ' y'  " 

•  -  ■  .  i  Hx  ’  ' 

•  i;j  .•  ■  '  .  0  011 

'.'  Y  /i 

■'  .  J-  ■'  :  ■  '.  .  r  •  ;r;  •>;  J  t: 

■;  ■  '  ,  •' 

( 

.  /.  ;  ■  f  •  ■  ' 

_ ._/ 

:  • u  .  i. 

.  •  ;■  '  V:  i  •  /:  ■"  'y •  o'  1"  ■  •  ’J'  1  y  ''o'  '  ^.0:;:.: 

. 

,•  i  '  XJ  QB&  ■  ■- 

v  "■  ■  "... 

' 

•  ■  ■  "  .'  .■  .  i  X  '  o  'o  J'  Y  o  ' 

.  Y  ...  . 

.i;J  '  ■ 

■  i  "  '  -■  j  :  o 

.  ■  Y  .  .  .  t :  o' 

, 

- 


REFERENCES  FOR  CHAPTER  V 


Ake,  James  N.  "Rapid.  Machine  Procedure  for  Determining 
Scalability  of  any  Number  of  Questions,"  Public  Opinion 
Quarterly,  26:121-125,  Spring,  1962. 

Back,  Kurt,  Rueben  Hill,  and  J.  Mayone  Stycos.  “Interviewer 
Effect  on  Scale  Reproducibility,”  American  Sociological 
Review,  20:445-446.  August,  1955* 

Borgatta,  Edgar  F.  “An  Error  Ratio  for  Scalogram  Analysis," 
Public  Opinion  Quarterly ,  19:96-100,  Spring,  1955* 

Campbell,  Ernest  Q. ,  and  Alan  C.  Kerckhoff.  “A  Critique  of 
the  Concept  'Universe  of  Attributes',"  Public  Opinion 
Quarterly,  21:295-505?  Summer,  1957* 

Case,  Herman  M.  “Guttman  Scaling  Applied  to  Centers' 
Conservatism-Radicalism  Battery,"  American  Journal  of 
Sociology,  58:556-565?  May,  1955* 

Danielson,  Wayne  A.  "A  Data  Reduction  Method  for  Scaling 
Dichotomous  Items Public  Opinion  Quarterly,  21:577-579? 

Fall,  1957. 

Davis,  James  A.  "On  Criteria  for  Scale  Relationships," 

American  Journal  of  Sociology,  65:571-580,  January,  1958. 

Edwards,  Allen  L.  Techniques  of  Attitude  Scale  Construction. 

New  York:  Apple ton-Century-Crofts ,  1957* 

Ford,  Robert  N.  "A  Rapid  Scoring  Procedure  for  Scaling 

Attitude  Questions,"  Public  Opinion  Quarterly,  14:507-552, 
Fall,  1950. 

Green,  Bert  F.  "A  Method  of  Scalogram  Analysis  Using  Summary 
Statistics,"  Psychometrika,  21:  79-88,  March,  1956. 

Green,  Bert  F.  "Attitude  Measurement,"  in  Gardner  Lindzey  (ed.). 
Handbook  of  Social  Psychology  Vol.  I_.  Reading,  Mass.:  Addison- 
Wesley  Publishing  Company,  1954*  Pp*  555-569* 

Green,  Norman  E.  "Scale  Analysis  of  Urban  Structures:  A  Study 
of  Birmingham,  Alabama,"  American  Sociological  Review, 

21:8-15,  February,  1956. 

Guilford,  J.P.  Psychometric  Methods.  Toronto:  McGraw-Hill 
Book  Company,  1954* 

Henry,  Andrew  F.,  and  Edgar  F.  Borgatta.  "A  Consideration  of  Some 
Problems  of  Content  Identification  in  Scaling."  Public  Opinion 
Quarterly,  20:457-469?  Summer,  1956. 


' 


99 


(15)  Korber,  George.  "Comments  on  'Role  Conflict  and  Personality'," 

American  Journal  of  Sociology,  52:48-49*  July,  1951* 

(16)  Laulicht,  Jerome.  "Role  Conflict,  the  Pattern  Variable  Theory, 

and  Scalogram  Analysis,"  Social  Forces,  33:250-254*  March,  1955* 

(17)  Lesser,  Gerald  S.  "Applications  of  Guttman's  Scaling  Method  to 

Aggressive  Fantasy  in  Children,"  Educational  and  Psychological 
Measurement .  18:543-561,  August,  1958. 

(18)  McDill ,  Edward  L.  "A  Comparison  of  Three  Measures  of  Attitude 

Intensity,"  Social  Forces ,  38:95-99*  December,  1959* 

(19)  Menzel,  Herbert.  'A  New  Coefficient  for  Scalogram  Analysis," 

Public  Opinion  Quarterly,  17:  268-280,  Summer,  1953* 

(20)  Nye,  Ivan  F. ,  and  James  F.  Short,  Jr.  "Scaling  Delinquent  Behavior," 

American  Sociological  Review,  22:326-331*  June,  1957* 

(21)  Parsons,  Talcott,  and  Edward  A.  Shils  (eds.).  Toward.  A  General 

Theory  of  Action.  Cambridge:  Harvard  University  Press,  1951* 

(22)  Peaks,  Helen.  "Problems  of  Objective  Observation,"  in  Leon 

Festinger  and  Daniel  Katz  (eds.).  Research  Methods  in  the 
Behavioral  Sciences.  New  York:  Holt,  Rinehart  and  Winston, 

1953. 

(23)  Pearson,  Richard  G.  "Plus  Percentage  Ratio  and  the  Coefficient 

of  Scalability,"  Public  Opinion  Quarterly,  21:379-380, 

Fall,  1957. 

(24)  Riley,  Matilda  White  and  Jackson  Toby.  "Subject  and  Object 

Scales:  A  Sociological  Application,"  American  Sociological 
Review,  17:287-296,  June,  1952. 

(25)  Riley,  Matilda  White,  John  W.  Riley,  Jr.,  and  Marcia  L.  Toby. 

"The  Measurement  of  Consensus,"  Social  Forces,  31:97-106, 

December,  1952. 

(26)  Riley,  Matilda  White,  John  W.  Riley,  Jr.,  and  Jackson  Toby. 

Sociological  Studies  in  Scale  Analysis .  New  Brunswick,  New 
Jersey:  Rutgers  University  Press,  1954* 

(27)  Schubert,  Glendon  A.  "The  Study  of  Judicial  Decision-Making  as 

an  Aspect  of  Political  Behavior;  Scalogram  Analysis,"  American 
Political  Science  Review ,  52:1014-1017,  December,  1958. 

(28)  Schuessler,  Karl  F.  "Item  Selection  in  Scale  Analysis,"  American 

Sociological  Review,  17:183-192,  April,  1952. 


(29)  Scott,  John  Finley.  "Two  Dimensions  of  Delinquent  Behavior," 
American  Sociological  Review,  24: 240-243 ,  April,  1959* 

(50)  Steiner,  Ivan  D.  "Scalogram  Analysis  as  a  Tool  for  Selecting 
Poll  Questions,".  Public  Opinion  Quarterly,  19:415-424? 

Winter,  1955* 

(31)  Stone,  Carol  Larson.  "A  Machine  Method  for  Scaling  as  Many  as 

Twelve  Dichotomies,"  Station  Circular  #29,  Pullman,  Washington: 
Washington  Agricultural  Experiment  Station,  Institute  of 
Agricultural  Sciences,  State  College  of  Washington,  August,  1958 
(Mimeographed) . 

(32)  Stouffer,  Samuel  A.,  et  al.  Measurement  and  Prediction.  Vol.  IV 

of  Studies  in  Social  Psychology  in  World  War  II.  4  Vols. 
Princeton,  New  Jersey:  Princeton  University  Press,  1950* 

(33)  Stouffer,  Samuel  A.  and  Jackson  Toby.  "Role  Conflict  and 

Personality,"  American  Journal  of  Sociology,  58:395-406, 

March,  1951* 

(34)  Stover,  Robert  E.  "The  Measurement  of  Change  in  a  Uni- 

dimensional  Attitude  by  Guttman  Scale  Analysis  Techniques," 
Public  Opinion  Quarterly,  22:116-122,  Summer,  1958. 

(35)  Strauss,  Anselm  L.  "The  Development  and  Transformation  of 

Monetary  Meanings  in  the  Child,"  American  Sociological 
Review,  17:275-286,  June,  1952. 

(36)  Torgerson,  Warren  S.  Theory  and  Methods  of  Scaling. 

New  York:  John  Wiley  and  Sons,  1958. 

(37)  Vernon,  Glenn  M.  "Inquiry  into  the  Scalability  of  Church 

Orthodoxy,"  Sociology  and  Social  Research,  39:324-327, 

May,  1955* 

(38)  Wallin,  Paul.  "A  Guttman  Scale  for  Measuring  V/ omen' s 

Neighborliness,"  American  Journal  of  Sociology,  59:243-246, 
November,  1953* 


«  '  t 

.  ■  .  . 


,  '  ’  r  r  , 


, 


' 


- 


- 


, 


.  '  '  •’ 


. 


. 

. 

, 


i 


' 


.  '■ 


'  • 


CHAPTER  VI 


COLLECTION  OP  DATA 

Various  aspects  of  the  task  of  obtaining  data  for  the  study  are  out¬ 
lined  in  this  chapter.  These  include  designing  the  questionnaires,  obtain¬ 
ing  the  sample  of  respondents,  and  distributing  the  questionnaires  to 
schools.  Relevant  information  concerning  the  returns  and  the  characteristics 
of  the  respondents  is  also  discussed. 

I.  TEACHERS’  AND  PRINCIPALS’  QUESTIONNAIRES 

Since  only  a  fraction  of  the  necessary  data  could  be  obtained  through 
the  use  of  instruments  which  were  already  available,  it  was  necessary  to 
design  instruments  for  collecting  the  major  portion  of  the  data. 

Selection  of  Items 

In  order  to  select  the  items  which  would  be  used  to  test  the  hypothesis 
that  there  are  scalable  areas  in  the  expectations  which  teachers  hold  for  the 
behavior  of  principals,  the  first  task  was  to  determine  what  dimensions  would 
be  considered.  As  has  been  indicated  in  the  statement  of  the  problem,  it  was 
decided  to  test  for  scalability  the  dimensions  of  role  conflict  which  were 
proposed  by  Seeman  (7).  After  each  of  the  universes  had  been  defined  clearly, 
a  search  was  begun  for  items  whose  content  would  make  them  valid  for  particular 
"dimensions .” 

Some  of  the  items  which  had  been  used  in  the  Seeman  study  (7)  appeared 
suitable  for  the  hypothesized  dimensions.  In  addition  to  this  source,  other 


.  • 

- 


' 


! 


. 


. 

....  •  *  • 


102 

studies  of  the  role  of  the  principal  and  other  administrators  were  reviewed 
in  a  search  for  appropriate  items  ($,  4»  5>  6).  One  of  the  criteria  which 
was  applied,  in  addition  to  that  of  validity  for  a  hypothesized  dimension, 
was  that  there  should  be  a  lack  of  consensus  on  the  particular  item  of 
expectations.  The  aim  of  the  investigator  was  to  identify  at  least  twenty- 
five  items  for  each  of  the  four  hypothesized  dimensions.  In  addition  to 
making  use  of  existing  items  reviews  of  literature  such  as  (l)  and  (2)  and 
discussions  with  fellow  graduate  students  were  also  used  as  sources  for  items. 
Through  these  approaches  the  desired  number  of  items  was  obtained. 

Pretests ,  Validity,  and  Revisions 

A  pretest  of  the  items  was  considered  essential.  To  facilitate  the 
obtaining  of  responses,  it  was  decided  to  solicit  the  cooperation  of  under¬ 
graduate  education  students  and  their  instructors.  In  order  to  make  fewer 
demands  on  these  students,  the  questionnaire  was  divided  into  its  four 
component  parts,  and  any  one  student  was  asked  to  respond  to  only  two  of  the 
parts  or  to  not  more  than  fifty  items.  Although  approximately  ninety  sets 
of  items  for  each  dimension  were  distributed  to  about  180  students,  only 
about  forty-five  completed  sets  for  each  hypothesized  dimension  were  returned. 
It  was  felt  that  this  was  a  satisfactory  response  for  the  purposes  of  the 
pretest. 

The  items  were  examined  for  any  indications  of  ambiguity  which  the 
respondents  had  been  asked  to  note.  The  frequencies  for  each  response  category 
were  also  computed  to  determine  how  much  agreement  there  was  on  each  of  the 
items.  This  revealed  that  the  pretest  had  not  been  of  much  value  in  this 
respect  for  there  was  a  large  measure  of  agreement  on  all  of  the  items.  This 


-  '  ■  :  '  i 


" 

*  ■ 


• .  '  ■ 


ai 


•  /  '  ; 

J  ■'  '  ,  '• 

•  .  .  ■ , 

...  .,  •  ••  ••  .  ;.  ,  •••  v. ;  'v~' 

...  ,  ■  ..  ■  .  y  .  •  '/•.:■  >  -•  v-  :o  r' 

'  '•  ■  ''  * 


r  •  . 


•  .  .  ;; 

.....  .  -  ' 

•;  •  ,  '  ■  ’■  ’ 

. 


103 


may  have  been  due  to  the  fact  that  few  of  the  students  had  been  in  actual 
teaching  situations  and  all  were  responding  from  a  common  frame  of  reference 
as  to  what  they  thought  ought  to  be;  it  was  evident  that  they  all  knew  what 
the  "correct"'  answers  were.  The  questionnaire  did  reveal  the  presence  of 
items  which  would  probably  not  be  suitable  even  with  a  different  sample  and 
these  were  either  reworded  or  discarded. 

Eight  graduate  students  also  completed  the  questionnaire  and  criticized 
the  items.  This  proved  to  be  more  useful  for  purposes  of  revising  the  items 
than  were  the  responses  of  the  undergraduates  particularly  in  relation  to 
identifying  items  on  which  there  was  some  measure  of  disagreement. 

Other  graduate  students  we re  asked  to  sort  cards  each  of  which  contained 
one  item  of  expectations  into  the  four  hypothesized  dimensions  on  the  basis 
of  definitions  provided  for  each  of  the  categories.  No  directions  were  given 
other  than  that  the  cards  were  to  be  placed  in  the  category  for  which  the 
content  of  the  items  best  suited  them.  This  validation  process  indicated 
that  there  were  items  which  clearly  belonged  in  certain  categories  while  others 
were  placed  in  different  categories  by  different  judges.  Discussion  with 
some  of  the  judges  after  the  sorting  process  served  to  clarify  why  ambiguous 
items  we re  placed  in  certain  categories  by  particular  judges;  this  served  as 
a  basis  for  making  revisions  in  the  content  or  emphasis  of  items. 

On  the  basis  of  the  steps  which  were  outlined  above,  changes  were  made 
in  many  items  and  some  items  were  discarded  and  new  ones  constructed.  A 
second  questionnaire  consisting  of  one  hundred  items  was  then  prepared; 
this  time  the  items  were  not  grouped  by  categories  but  were  distributed 


■  ■  ' 

■  ; ••  -I  ,'r  or:  G-tmvb.rri  '  ’or: 

•  •.  .  '  • ;  '  •  :  •' 

•  ■  •  -  -  ■  •  . 

. 

;  '  '  , 

■  i.Tr  •  > 

•  •  •  •  •  .  ;  :  :  •  ■  ■’  ■  ■  •  "  '  - 

■ 

■ 

>ti-  i 


' 


104 


randomly  throughout  the  questionnaire.  Furthermore,  five  response 
categories  ranging  from  "Strongly  Agree"  to  "Strongly  Disagree"  we re 
provided  instead  of  the  three  which  had  been  used  in  the  pretests.  The 
revised  questionnaire  was  distributed  to  approximately  seventy-five 
elementary,  junior  high,  and  senior  high  school  teachers  in  four  schools 
located  near  the  city.  Approximately  forty-five  usable  questionnaires 
were  returned  in  time  for  analysis  before  the  final  form  of  the  question¬ 
naire  was  prepared. 

The  items  were  grouped  according  to  the  hypothesized  dimensions  and 
were  examined  to  determine  the  distribution  of  responses  for  each  item.  It 
was  found  that  there  was  greater  variation  in  the  responses  to  items  that 
had  been  left  unchanged  from  the  other  pretests  than  there  had  been  for 
either  the  graduate  or  undergraduate  students.  This  suggested  that  many  of 
the  items  which  had  been  considered  as  marginal  might  be  suitable  for  the 
study.  Items  on  which  there  was  a  large  measure  of  agreement  were  again 
rejected  in  favor  of  those  on  which  there  was  less  agreement. 

As  a  result  of  these  pretests,  it  was  decided  to  select  twenty  of  the 
more  promising  items  for  each  of  the  hypothesized  dimensions  for  the  final 
form  of  the  questionnaire.  On  the  final  form  of  the  questionnaire  (see 
Appendix  A) ,  the  items  were  so  arranged  that  every  fifth  item  was  for  the 
same  hypothesized  dimension.  There  was  no  indication  on  the  questionnaire 
or  in  the  instructions  provided  the  teachers  that  the  questionnaire  contained 
any  subtests.  This,  plus  the  fact  that  there  were  a  large  number  of  items 
on  the  questionnaire,  served  to  make  it  difficult  for  respondents  to  be 


.  ' 


i: 


. 


' 

'■  . 

■ 

« 

■  ' 


■  ' 

■ 

, 


,  • 

.  '■ 


105 

deliberately  consistent  in  responding.  It  was  reasoned  that  if  the 
items  were  scalable  under  these  conditions,  there  would  be  good  grounds 
for  assuming  that  scalability  did  not  result  from  factors  inherent  in  the 
design  of  the  questionnaire. 

LBDQ 

Each  teacher  was  also  required  to  describe  the  behavior  of  the 
principal  on  the  Leader  Behavior  Description  Questionnaire  (see  Appendix 
B).  The  instrument  is  available  commercially. 

Principal  Ambivalence 

Principals  were  requested  to  complete  the  same  expectations  question¬ 
naire  as  did  the  teachers.  This  served  a  two-fold  purpose  in  that  it  was 
later  used  to  derive  the  index  of  teacher- principal  agreement  and  also 
served  as  the  content  questions  on  which  the  ambivalence  measure  was  based. 
For  this  second  purpose  an  additional  questionnaire  was  designed  (see 
Appendix  C)  which  principals  used  in  conjunction  with  the  expectations 
questionnaire.  After  responding  to  each  expectation  item,  a  principal  was 
directed  to  use  the  second  questionnaire,  or  more  correctly  answer  sheet,  to 
indicate  how  difficult  it  had  been  for  him  to  decide  on  his  response.  The 
same  procedure  was  to  be  followed  for  all  items  on  the  questionnaire. 

II.  SELECTING  THE  SAMPLE  AND  OBTAINING  DATA 


The  Sample 

The  study  was  carried  out  in.  schools  which  included  all  grades  from 


one  to  twelve,  which  had  no  fewer  than  nine  nor  more  than  twenty-four 


- 


'  ' 


•  :  •  •  •  ! 


. 

..  ;  ■ 

■  ■  ■ 

•  •  :  •  '■  '  - 

.  : ;;  ■  ’ '  ■  •  \'"  •  '•  ■  "/  sty. 


.  ' 

.  ■  '  ’  "  '  ■■  ' 


... 


...  •. 


' 

'  . 


106 


teachers,  and  which  were  located  in  non-city  areas  of  central  Alherta. 

Schools  were  selected  on  the  "basis  of  the  above  criteria  and  also  on  their 
proximity  to  Edmonton  from  which  the  investigation  was  carried  out. 

Principals  of  selected  schools  were  mailed  a  form  letter  early  in  April 
in  which  the  general  nature  of  the  study  was  outlined  and  which  requested 
the  cooperation  of  the  principal  and  his  staff  in  the  collection  of  the 
necessary  data.  In  all  sixty-seven  schools  were  contacted  in  this  way  on 
the  assumption  that  at  least  sixty  of  these  would  be  willing  to  take  part 
in  the  study.  The  form  letter  also  advised  the  principals  that  the  investi¬ 
gator  would  personally  visit  the  schools  during  the  first  half  of  May  to 
explain  the  study  in  more  detail  and  to  deliver  questionnaires  for  them  to 
distribute  to  the  teachers  in  each  school.  As  a  result  of  this  and  later 
contacts  three  principals  indicated  that  they  did  not  wish  to  participate 
in  the  study.  These  schools  were  eliminated  from  the  list  and  no  further 
effort  was  made  to  encourage  them  to  participate. 

Visiting  Schools 

A  schedule  was  prepared  for  visiting  the  sixty-four  schools,  and  the 
principal  of  each  school  was  notified  of  the  exact  day  on  which  the  visit 
would  be  made  to  his  school.  A  procedure  similar  to  that  described  below 
was  carried  out  in  each  of  the  schools.  The  investigator  attempted  to  time 
his  arrival  so  that  it  would  not  interfere  greatly  with  the  normal  operation 
of  the  school;  an  attempt  was  made  to  arrive  at  some  regular  break  in  the 
school  day,  and  if  the  principal  happened  to  be  in  class  he  was  not  disturbed 
until  the  end  of  the  period.  The  investigator  first  asked  the  principal 


whether  there  were  any  further  questions  and  reviewed  the  general  nature  of 


- 

.  '  ,  ■  ■:  -  '  i'  '  •"  "  •  •'  ' 


■ 

■ 

■  •  "  • 

■ 

: 

. 

■ 

. 

■  -  ■  ,  •  -  'y  '•  i'j'  ‘..'."v 

'  :  ■  ■  -  '  '  "  ''  "•  ’■  :-4  '  ■  "  '  '  '  '  ‘ 


- 


fj 

J"  9 

-  ■ 

...  ..•••.  •  ,  '  '  •  ‘  : 

•I  '  ■ 

. 


loy 

the  study  with  him  while  examining  the  questionnaires  that  the  teachers 
and  principal  were  being  asked  to  complete.  In  only  one  case  did  the 
principal  refuse  to  participate  in  the  study  after  examining  the  question¬ 
naires. 

The  principal  was  asked  to  use  his  judgment  in  selecting  a  suitable 
time  for  distributing  the  questionnaires  to  the  teachers  in  his  school. 

Each  teacher  was  provided  with  an  envelope  containing  the  questionnaires 
and  the  necessary  instructions.  Teachers  were  instructed  to  complete  the 
questionnaires  independently  of  other  staff  members  and  after  having  done 
so  to  place  completed  forms  in  the  envelope,  to  seal  it,  and  to  hand  it  to 
the  staff  member  who  had  been  designated  to  hold  the  envelopes.  Each  school 
was  provided  with  a  large  envelope  in  which  the  questionnaires  were  to  be 
returned  to  the  investigator  and  sufficient  postage  was  provided. 

Response 

In  all  sixty-two  schools  were  visited  in  this  way  and  principals  were 
instructed  in  the  details  of  carrying  out  the  data  collection.  As  has  been 
indicated  only  one  principal  refused,  to  participate  after  discussing  the 
study  with  the  investigator.  In  one  other  school  the  principal  was  absent 
due  to  an  extended  illness,  and  this  school  was  eliminated  from  the  study. 

Two  of  the  schools  which  were  initially  contacted  were  some  distance  away 
from  any  of  the  others,  and  due  to  the  excellent  cooperation  from  the  schools 
which  had  been  visited,  it  was  decided  to  mail  the  questionnaires  to  these 
schools  and  to  attempt  to  complete  all  aspects  of  the  study  by  ma.il.  The 
response  from  these  two  schools  was  again  highly  satisfactory,  and  this 
suggests  that  perhaps  more  of  the  study  could  have  been  carried  out  in  this 


' 

•  }  ■  .  "  ;• 

1  •  •  •  r 

, 

r  '  •  i 

' 

.  r.  ■  '  :  OT  1 

•  • :  •  ' 

' 

'■ .  £  -  ■■  i 

'  ' 

■ 

- 

' 

,  '  '■  '  '■  . 

■  ■  :  •  -  '  '• 

'■  ~  ■  if  . ' 

- 

. 

■  •  '■  "  » 

- 

•  ■  -  •  ■  ■  . '•  -  • 

* 

‘  '  '•  • 

■  ■ 

■  ■  j  ■•••; 

->  • 

•  -  r  •.  nlooti  •" 

•  '  .  .  "I  V  ; 

•  .  .  -  ,  -■  T  (  r r  1  .  ' •:  r '•  ')  O  •<  •  ■ 

108 


way.  However,  it  was  the  opinion  of  the  investigator  that  the  personal 
contact  with  the  principals  probably  was  a  factor  in  the  excellent  response. 
Table  I  presents  some  information  about  the  schools  which  were  contacted  and 
the  response  which  was  obtained. 


TABLE  I 

HUMBER  OP  QUESTIONNAIRE)  DISTRIBUTED  AND  COMPLETED 
BY  SIZE  OP  SCHOOL 


Number  of 

Teachers 

Questionnaires 

Distributed 

N 

Questionnaires 

Returned 

N 

9 

3 

27 

24 

10 

9 

90 

80 

11 

12 

132 

126 

12 

5 

60 

58 

13 

3 

39 

21 

14 

5 

70 

64 

15 

5 

75 

61 

16 

3 

48 

44 

17 

3 

51 

22 

18 

3 

54 

52 

19 

3 

57 

55 

20 

3 

60 

52 

21 

2 

42 

42 

22 

2 

44 

40 

23 

24 

1 

24 

24 

Totals 

62 

873 

765 

Two  of  the  schools  which  were  visited  did  not  return  any  completed 
questionnaires „  Correspondence  with  the  principal  of  one  of  the  schools 
resulted  in  a  reply  to  the  effect  that  the  teachers  did  not  wish  to 
participate  in  the  study  and  that  the  questionnaires  would  be  returned; 
the  principal  of  the  other  school  did  not  reply  to  the  investigator's 


....  '  '■  ■  r  ■ "  ■ 

i  cx  '  i\b$i 

. 


'  ' 


'  • 


•  . . 


' 


109 

inquiry.  No  further  attempts  were  made  to  obtain  completed  questionnaires 
from  either  of  these  schools.  The  relatively  poor  response  from  four  schools 
led  eventually  to  their  elimination  from  the  study;  it  was  felt  that  the 
nature  of  the  study  required  a  high  proportion  of  usable  responses  from 
each  school.  In  view  of  the  usual  percentage  of  responses  in  questionnaire 
studies,  the  investigator  felt  that  he  had  been  highly  successful  in  his 
attempts  to  obtain  an  adequate  response  from  each  of  the  schools  and  from 
the  total  schools. 

Table  I  shows  that  of  the  873  questionnaires  which  were  distributed 
765  were  returned  completed  in  sufficient  detail  so  that  they  could  be 
included  in  the  study.  This  represents  a  return  of  over  When  six 

schools  were  eliminated  from  the  study  due  to  insufficient  returns  from 
each  school,  the  completed  returns  rose  to  93^  of  the  possible  response 
from  the  fifty-six  schools.  The  questionnaire  returns  by  schools  is  given 
in  Appendix  D,  and  particular  schools  eliminated,  from  the  final  stages  of 
the  study  are  indicated. 

III.  CHARACTERISTICS  OF  RESPONDENTS 

The  general  characteristics  of  the  schools  in  which  the  study  was 
carried  out  were  discussed  in  relation  to  the  selection  of  the  sample.  Some 
consideration  will  now  be  given  to  characteristics  of  the  respondents  even 
though  it  is  not  intended  to  make  any  comparisons  to  show  that  the  respondents 
are  representative  of  any  larger  group.  The  information  which  is  provided 
does  serve  to  indicate  that  the  respondents  did  not  differ  to  any  great 


•  •  '■ 


' 

1  • 

a 

.  ;•  •  "  ■  ' 


■ 

■  ■■  > , 

, 

. 

1 

r 

■t  ■ 

, 

' 

■ 

■  ;  r  '  -  ■  ■  "  ■ 

' 

)■•  ■;  '  ■ 


' 


110 


extent  from  what  one  might  have  expected  on  the  basis  of  rather  general 
knowledge  about  the  teaching  population. 

Table  II  provides  data  concerning  sex,  age,  education,  and  teaching 
experience  of  both  teachers  and  principals  as  well  as  data  about  the  teaching 
responsibilities  of  the  teachers.  It  is  evident  that  only  slightly  over  one- 
third  of  the  teachers  were  males  while  nearly  all  of  the  principals  were 
male.  The  modal  age  group  for  the  teachers  was  the  21  to  30  category  with 
almost  two -thirds  of  the  teachers  being  under  forty  as  compared  to  less  than 
one-half  of  the  principals  who  were  forty  years  of  age  or  younger. 

When  principals  and  teachers  are  compared  on  years  of  training  it  is 
evident  that  the  principals  have  the  edge;  only  eight  principals  had  fewer 
than  four  complete  years  of  training  while  almost  one-half  of  the  teachers 
who  responded  had  only  one  complete  year  of  teacher  education.  The  distri¬ 
bution  of  teachers  by  years  of  experience  is  almost  rectangular,  but  only 
one-third  of  the  principals  who  responded  had  fifteen  or  fewer  years  of 
experience  in  teaching.  The  distribution  of  teachers  by  the  grades  in  which 
they  were  spending  most  teaching  time  is  also  very  close  to  being  rectangular. 

Since  the  teachers  were  asked  to  describe  the  leader  behavior  of 
principals,  it  is  of  interest  to  note  on  how  much  knowledge  these  descriptions 
were  based  in  terms  of  length  of  association  with  the  present  principal. 

Table  III  reveals  that  over  5 of  the  teachers  had  been  in  the  present  school 
with  the  present  principal  for  less  than  two  complete  years  and  two-thirds 
had  been  in  the  present  school  for  five  years  or  fewer.  The  principals  also 
appeared  to  be  fairly  mobile  for  although  two-thirds  of  them  had  more  than 


.  - 

. 

♦ 

•  ;  •  '  i'  Bi  ■'  ' 

.  - 

■ 

;  •  -  ’ 

,  ■  ■  •  ■  :  '  '  '  ' 

. 

" 

i  7  ■  -  7  :>-v  $('  •  ■  ■ 


■ 

■ 

i  £ 

'  •  •  '  •  v 

' 

0 

• 

' 

•  1  '  i  * 

• 

■ 

■  -  .  r 

• 

‘ 

' 

•  • 

■  >. ' 

'  ■ 

Ill 


TABLE  II 

SELECTED  CHARACTERISTICS  OF  RESPONDENTS 


Teachers 

N=765 

Principals 

N=60 

SEX: 

Male 

262 

56 

Female 

492 

4 

Not  Stated 

11 

0 

AGE: 

Years 

Under  20 

55 

0 

21-30 

250 

2 

31-40 

184 

24 

41-50 

156 

22 

Over  50 

127 

12 

Not  Stated 

13 

0 

TRAINING: 

Years 

1 

373 

0 

2 

156 

3 

5 

57 

5 

4 

104 

23 

5 

35 

16 

6 

13 

10 

7 

6 

2 

7+ 

5 

1 

Not  Stated 

16 

0 

EXPERIENCE: 

Years 

2  or  less 

131 

1 

5  -  5 

129 

2 

6-10 

159 

6 

11  -  15 

135 

13 

16—  20 

83 

16 

Over  20 

115 

22 

Not  Stated 

13 

0 

GRADES  TAUGHT: 

Grades 

1  -  5 

4-6 

7-9 

10  -12 

Not  Stated 

191 

188 

197 

175 

44 

- 

'  •• 

'  _  "  ' 


'  * 

- 


112 


TABLE  III 

DISTRIBUTION  OP  PRINCIPALS  BY  EXPERIENCE 
AND  TEACHERS  BY  TENURE 


Years 

No.  of  Principals 

No.  of  Teachers 

Total 

Experience 

In  Present 
School 

In  Present 
School 

With  present 
Principal 

2  or  less 

9 

15 

501 

407 

3  -  5 

11 

19 

211 

251 

6-10 

17 

14 

157 

93 

11  -  15 

7 

2 

54 

13 

16  -  20 

6 

5 

15 

7 

Over  20 

9 

4 

11 

0 

Not  Stated 

1 

1 

16 

14 

Totals 

60 

60 

765 

765 

five  years  of  experience  as  a  principal,  only  slightly  more  than  5 Nad. 
held  the  present  position  for  less  than  five  complete  years  at  the  time 
of  the  study. 

One  implication  of  the  data  concerning  the  tenure  of  teachers  and 
principals  is  that  within  any  one  school  both  of  these  are  faced  with 
adjusting  to  many  changing  relationships.  Because  of  the  mobility  of 
both  the  teachers  and  principals  there  is  further  need  for  knowing  more 
about  the  role  expectations  involved  and  the  means  for  assessing  them.  It 
may  be  that  if  too  much  is  left  to  be  developed  through  on-the-job  relation¬ 
ships  this  is  achieved  at  just  about  the  time  that  one  or  the  other  role 
incumbent  is  about  to  move  to  a  new  position. 


. 


■  ■  ■  •  - 

> 


*  ' 


, 


■  ■  •  v 


113 


REFERENCES  FOR  CHAPTER  VI 

(1)  Campbell,  Roald  F.,  John  E.  Corbally,  Jr.  and  John  A.  Ramseyer. 

Introduction  to  Educational  Admini s trat i on .  Second  Edition. 
Boston:  Allyn  and  Bacon,  Inc.,  1962. 

(2)  Campbell,  Roald  F.  and  Russell  T.  Gregg  (eds.).  Administrative 

Behavior  in  Education.  New  York:  Harper  and  Brothers,  1957* 

(3)  Cheal,  John  E.  "Role  Conflict  in  the  Principalship  of  the 

Composite  High  School.”  Unpublished  Master's  thesis. 

The  University  of  Alberta,  Edmonton,  1958* 

(4)  Gross,  Neal,  Ward  S.  Mason,  and  Alexander  W.  McEachem. 

Explorations  in  Role  Analysis:  S tudi es  of  the  School 
Superintendency  Role.  New  York:  John  Wiley  and  Sons, 

Inc.,  1958. 

(5)  Hhlpin,  Andrew  W.  The  Leadership  Behavior  of  School  Super¬ 

intendents.  Second  Edition.  Chicago:  Midwest  Administration 
Center,  1959* 

(6)  Warren,  Philip  J.  "Leadership  Expectations  of  the  Principal 

in  Newfoundland's  Regional  and  Central  High  Schools  as 
Perceived  by  Principals  and  Staffs."  Unpublished  Master's 
thesis,  University  of  Alberta,  Edmonton,  1959* 

(7)  Seeman,  Melvin.  Social  S tatus  and.  Leadership.  Columbus, 

Ohio:  The  Ohio  State  University,  I9S0. 


■ 


,  ‘  ..  '■  .  ■ 1  ■ 
■'  '  „  ■ 


'  •  •  '  • 

i 

> 

, 


CHAPTER  YII 


SCALOGRAM  ANALYSIS  OF  DATA 

In  the  scale  analysis  a  hoard  similar  in  construction  to  that 
described  in  Chapter  IV  was  used.  The  scalogram  hoard  technique  was 
selected  for  a  number  of  reasons.  The  first  of  these  was  that  the 
investigator  supposed  that  a  number  of  items  from  each  "dimension" 
might  have  to  be  rejected  because  of  inappropriate  content;  these  items 
could  be  detected  quite  readily  through  use  of  this  method.  Furthermore, 
it  was  quite  certain  that  the  response  categories  would  have  to  be 
combined,  and  this  seems  more  logical  when  based  on  a  sample  than  when 
it  is  done  for  all  items  on  an  a  priori  basis  as  is  necessary  when 
machine  procedures  are  used.  The  other  great  advantage  of  the  board 
is  that  it  provides  a  visual  picture  which  is  of  great  value  in  gaining 
insights  into  the  technique  and  the  apparent  structure  of  the  items. 

I.  GENERAL  PROCEDURES 

For  the  purposes  of  the  scale  analysis,  a  sample  of  one  hundred 
questionnaires  was  selected  at  random  from  among  those  which  had  been 
completed  and  returned.  The  items  were  grouped  by  the  hypothesized 
dimensions  and  the  responses  were  tabulated  to  facilitate  the  scalogram 
analysis.  For  each  of  the  items  the  range  of  marginal  frequencies  was 
calculated,  and  those  items  which  had  extreme  marginals  were  temporarily 
set  aside  in  favor  of  those  which  had  a  more  even  distribution  of  marginal 

frequencies.  This  process  was  anticipated  when  the  items  were  being 


. 


' 


'  '  '  i  ■  v 

' 


■ 

■ 

•  '  ■ : 

. 

e 

' 

•  • 

- 

••  '  b  ,  '  • 


115 


constructed.,  and  for  this  reason  more  items  than  were  actually  necessary 
were  included  in  the  questionnaire. 

After  the  process  of  setting  aside  items  on  the  basis  of  their 
marginal  frequencies,  a  further  criterion  was  applied  in  that  items  were 
ranked  according  to  the  extent  to  which  the  judges  had  been  able  to  agree 
on  the  suitability  of  an  item  for  a  particular  category.  Those  items  on 
which  there  was  a  high  degree  of  agreement  were  rejected  in  favor  of  those 
on  which  there  was  less  agreement.  Thus  the  items  which  were  selected  for 
scale  analysis  had  to  meet  two  criteria:  one  requiring  an  acceptable  dis¬ 
tribution  of  marginal  frequencies  and  the  other  requiring  a  high  degree 
of  validity  for  a  particular  category.  This  process  left  about  twelve 
items  in  each  of  the  four  hypothesized  dimensions,  and  it  is  these  items 
which  were  subjected  to  the  scale  analysis. 

The  scale  analysis  presented  some  procedural  problems  in  tha.t 
there  is  a  degree  of  subjectivity  in  determining  initial  rank  order  of 
subjects  and  also  in  the  combining  of  response  categories.  The  problem 
was  complicated  by  the  fact  that  only  one  board  was  available  rather  than 
the  two  which  are  suggested  for  maximum  efficiency.  Eventually  all  of  the 
items  were  dichotomized,  and  a  rank  order  v/hich  approached  a  parallelogram 
pattern  of  responses  was  established  for  each  of  the  four  hypothesized 
dimensions.  These  results  were  highly  disappointing  for  the  patterns 
revealed  an  extreme  amount  of  error  as  well  as  many  grouped  errors  which 
were  indicative  of  non-scale  types.  There  was  no  doubt  but  that  the  four 
sets  of  items  did  not  come  anywhere  near  satisfying  the  criteria  required 
for  a  scale  or  even  a  quasi  scale. 

In  the  next  step  an  attempt  was  made  to  improve  the  sc al ability  by 


* 


• ' : 


[j: 


.  ;  .1 


:  •  a  -  'X. 


V.O 


c  r 


Ji 


')• 


•  ■  i  :.i  !/./  j.  ■ 


■ 


rj 


V.  . . 


-t. 

0 


•!. 


••  ; 


'  :  . 


o'.;  o  .  •  >.  C  • 


V.o  :  s.  f.  ;.c'o  ia 


X 


. 


■ 


I 


-V 


>xJ 


; 


r. 


(  :: 


■ 


c 


)  v 


■  n 


*x  ‘  ; 


■ 


116 


removing  items  which  appeared  to  have  a  high  proportion  of  errors  and 
non-scale  grouped  errors.  As  long  as  this  approach  was  used  without  any 
special  consideration  being  given  to  the  content  of  the  items  the  approach 
was  fruitless;  the  elimination  of  one  item  with  a  high  degree  of  error  for 
the  most  part  did  not  improve  the  scale  as  a  whole.  It  was  then  hypothe¬ 
sized  that  if  individual  items  were  not  at  fault,  there  might  be  subsets 
of  items  which  might  approach  scalability.  An  attempt  was  made  to  identify 
such  subsets  of  items  by  studying  carefully  the  items  within  each  category 
for  possible  similarities  in  content  above  that  established  by  the  general 
content  of  the  area.  Thus  a  random  approach  to  discarding  items  on  the 
basis  of  their  error  was  replaced  by  a  more  logical  and  rational  approach 
which  concerned  itself  with  the  content  of  the  items.  Some  of  the  items 
which  had  been  set  aside  in  the  initial  stages  of  the  scale  analysis  were 
again  examined  in  the  light  of  the  redefinition  of  content.  Since  the 
inves tigator  now  became  the  sole  judge  of  the  content  of  items,  even  those 
on  which  judges  had  not  been  able  to  agree  were  accepted  as  suitable  if 
they  were  similar  to  items  on  which  there  had  been  some  agreement  and  which 
seemed  to  be  related  in  content. 

II.  APPROXIMATELY  SCALABLE  AREAS 

Since  the  sets  of  items  did  not  approach  scalability  when  grouped 
as  originally  hypothesized,  this  aspect  of  the  analysis  is  not  discussed 
in  detail.  However,  more  detailed  attention,  will  be  given  to  the  various 
subareas  which  proved  to  be  approximately  scalable.  For  these  subareas, 


-  '■ 

• 

't 

•  '  '■ 

■  ' 

.  .v  •  • 

* 

. 

, 

• 

■ 

■  .  r-  ‘  •-  t  '  •  ■ 

'  V  :  ' 


117 

the  sets  of  items  involved  and  scalogram  results  axe  outlined  in  the 
following  sections. 


Status  Dimension 


The  content  of  items  in  this  dimension  concerned  the  expectations 
which  teachers  held  about  the  extent  to  which  a  principal  should  emphasize 
his  rank  in  the  formal  structure  as  opposed  to  emphasizing  equality  with 
teachers.  A  close  examination  of  the  items  which  had  been  planed  in  this 
category  suggested  the  possibility  of  two  scalable  subareas.  The  largest 
of  these  consisted  of  nine  items  which  were  closely  concerned  with  the  extent 
to  which  a  principal  should  emphasize  his  position  in  the  hierarchy  during 
his  job  relations  around  the  school.  The  other  set  of  eight  items  was  more 
closely  related  to  the  extent  to  which  a  principal  should  emphasize  his  social 
distance  although  the  distinction  is  not  as  clear  cut  as  might  be  desired. 

When  these  were  tested  for  scalability,  each  yielded  a  coefficient 
of  reproducibility  of  .85;  however,  there  was  evidence  of  the  presence  of 
non-scale  types  and  in  addition  some  of  the  item  categories  contained  more 
error  than  non-error.  In  each  case  the  scales  were  examined  for  unacceptable 
items  an d  these  were  eliminated.  By  chance  seven  items  remained  in  each  of 
the  scales  when  the  error  had  been  reduced  to  an  acceptable  level.  One  set 
(Si)  consisted  of  the  following  items: 

(1)  A  principal  should  think  of  himself  as  being  one 
of  the  teachers  in  the  school.  (lf,D,SD)l 

(2)  If  a  principal  has  lunch  at  school,  he  should  associate 
with  those  teachers  who  are  also  at  school  during  the 
lunch  hour  as  one  of  their  equals.  (U,D,SD) 

(5)  A  principal  should  keep  his  relations  with  the  staff 
on  an  informal  level.  (u,D,8D) 


Answer  categories  in  brackets  were  weighted  1  in  the  scoring 
process;  other  categories  were  weighted  0.  Items  arc  arranged  in  ascending 
order  of  total  responses  in  the  indicated  categories. 


.  ' 


■  r 


j;  '  ;; 

■  1 

' 


/J:  fwroJ  .i:  '.L^  /.  •.  J'lit 

......  ;1, 

i  . . 

I  ■  ■ 

:  >• 


•  ■  :  ■  .  •;  T  i.  '  , 

:  '  -  :  .  r  •  : 


• 

-  '  .’.  .’.  ;  ■ 


■  -  ,r  •  j  o 

.'. ' :  ■’  j  •  ■  ;•  c 

. >  i 

"  <  '  '  '  ' .  '.  ■  ’  ■  •  ■  £>■  :  < 

■  "■  .'.  ■  •  ■  •  u  ■  ,f  •’ 1  '* .' r 


'  '  '  ■  ;  .  '■  '.  "  ■  *  ’■  •  ••••.' 

- .  .  •'  ;  'Co  ■  ’j.  :  ..  •  r  :  •  • 


118 


(4)  A  principal  should  be  more  like  a  superintendent 
than  like  a  teacher.  (SA,A,U) 

(5)  A  principal  should  not  strive  to  achieve  a  higher 
social  position  in  the  community  than  that  of 
the  teachers  in  his  school.  (D,SD) 

(6)  A  principal  should  keep  a  certain  professional 
distance  between  himself  and  the  teachers  in 
his  school.  (SA,A,u) 

(7)  Within  the  school  a  principal  should  try  to  be 
just  another  member  of  the  group  in  his  job 
relations.  (U,D,SD) 

It  is  evident  that  the  items  in  this  set  were  not  nearly  as  specific 
as  were  those  in  the  second  set  (S2)  which  consisted  of  the  following 
items : 

(1)  If  a  teacher  and  a  principal  disagree  over  a. 
matter  of  classroom  procedure,  it  is  reasonable 
for  the  principal  to  expect  the  teacher  to  yield 
to  his  point  of  view.  (SA,A) 

(2)  A  principal  should  receive  a,  salary  considerably 
above  that  of  the  highest  paid  teacher  in  his 
school.  (SA,A) 

(3)  A  principal  should  submit  to  the  superintendent 
written  reports  on  each  teacher's  effectiveness.  (SA,A,U) 

(4)  A  principal  should  spend  as  much  time  supervising 
playgrounds,  hallways,  and  extra-curricular  activities 
as  any  of  the  teachers  in  the  school.  (U,D,SD) 

(5)  Teachers  should  not  call  the  principal  by  his 
first  name.  (SA,A,U) 

(6)  In  a  fairly  large  school  it  is  reasonable  for  a. 
principal  to  ask  teachers  to  make  appointments  to 
see  him  rather  than  to  come  to  his  office  whenever 
they  wish.  (SA,A,U) 

(7)  Even  though  he  may  have  less  training  and/or 
experience  than  some  teachers,  the  opinions  of 
a  principal  should  carry  more  weight  than  those 
of  the  teachers  in  the  school.  (SA,A,U) 

The  scalograms  for  both  sets  of  items  are  presented  in  Figure  4*  When  the 

various  criteria  for  scalability  were  applied  it  was  evident  that  the 

reproducibility  wan  not  high  enough  in  order  for  either  of  these  to  qualify 

on  a.  scale,  since  these  were  only  .88  and  .85  respectively.  However,  due 

to  the  fact  that  each  item  contains  more  non -error  than  error,  that  the 


c  ;  - 

v  '■>:  ./  :  . 

::  .  /  J;:.  - 

.  .  J. 

J'  '  j’<  ■  ■  '  ~ 

.c  '•  u  ■  J  "'J  j ;  ■  '  , 

;  '  5  , 

.  V  {  ' 

t  t 


j 


: 


■  y  (  ~~  ^ 

' 

<  • 

J:  : 

. 

' 

: 

■■  . 

.  ; 

.  i:  i' 

'  ; 

'■  ■  ' 

'  ■■ 

i: 

5  •  .  c 

1 

.  ■  ■ 

^  .  / 

’ 

•  :  •  J  rlo. 

■ 

. 

■ 

•  r 

•  •  •  ■  • 

...  ; 

- 

’ 

r 

t  t  u 

' 

;  : 

■  ’  .  '  r,.  ' 

. 

•  •  .. 

- 

119 


pattern  of  errors  is  random,  and  that  each  set  contains  a  number  of 
items  whose  modal  frequencies  lie  in  the  30  -  7&/°  range,  there  is  good 
justification  for  considering  the  sets  of  items  as  forming  quasi  scales. 

In  order  to  meet  all  the  necessary  criteria,  it  would  have  been 
desirable  to  have  more  items  than  those  which  were  available.  Unfortunately, 
it  was  impossible  to  add  items  at  this  point  in  the  study. 


Authority  Dimension 

The  items  which  were  originally  included  in  this  category  were 
designed  around  content  concerning  the  extent  to  which  a  principal  should 
permit  teachers  to  make  their  own  decisions  in  relation  to  their  work  in 
the  school  and  in  their  classrooms.  When  the  items  in  this  area  were 
examined  for  the  existence  of  possible  sub-areas,  it  was  found  that  seven 
of  the  twenty  items  dealt  with  the  extent  to  which  teachers  should  look 
to  the  principal  for  help  with  their  discipline  problems.  The  others 
were  concerned  with  the  authority  relationships  between  teachers  and 
principals  generally. 

The  seven  items  related  to  discipline  (A]_)  were  as  follows: 

(1)  A  principal  should  not  do  anything  to  help  a  teacher 
who  is  having  teaching  difficulties  until  the  teacher 
comes  to  him  for  help.  (SA,A) 

(2)  A  principal  should  visit  regularly  the  classes  of 
those  teachers  who  are  weak  in  discipline  and  exert 
his  personal  influence  in  an  attempt  to  keep  the 
classes  under  control.  (D,SD) 

(3)  A  principal  should  intervene  immediately  if  he 
believes  that  a  teacher  is  having  discipline 
problems.  (D,SD) 

(4)  A  principal  should  not  encourage  teachers  to  look  to 
him  for  help  in  controlling  their  classes.  (SA,A,U) 

(5)  A  principal  should  allow  teachers  to  work  out  their 
classroom  problems  by  themselves.  (SA,A,U) 


.  '  5 


• 

' 

'  ■'  t  r  '  ‘  r  : 


.  •  V : 


•.( 

, 

: 

. 


. 

. 


120 


STATUS  NO.  1  ITEMS  STATUS  NO.  2  ITEMS 


fRen.  = 

.88) 

(Hep 

.85) 

1 

2 

3 

4 

5 

6 

7 

8 

9 

10 

1 1 

12 

13 

14 

1 

2 

3 

4 

5 

6 

7 

8 

9 

10 

11 

12 

13 

14 

1 

25 

X 

X 

X 

X 

X 

X 

X 

1 

15 

X 

X 

X 

X 

X 

X 

X 

2 

27 

X 

X 

X 

X 

X 

X 

X 

2 

25 

X 

X 

X 

X 

X 

X 

X 

38 

X 

X 

X 

X 

X 

X 

X 

3 

9 

X 

X 

X 

X 

X 

X 

X 

4 

6 

X 

X 

X 

X 

X 

X 

X 

4 

51 

X 

X 

X 

X 

X 

X 

X 

5 

88 

X 

X 

X 

X 

X 

X 

X 

5 

5 

X 

X 

X 

X 

X 

X 

X 

6 

79 

X 

X 

X 

X 

X 

X 

X 

6 

45 

X 

X 

X 

X 

X 

X 

X 

7 

5 

X 

X 

X 

X 

X 

X 

X 

7 

52 

X 

X 

X 

X 

X 

X 

X 

p 

80 

X 

X 

X 

X 

X 

X 

X 

8 

31 

X 

X 

X 

X 

X 

X 

X 

9 

15 

X 

X 

X 

X 

X 

X 

X 

9 

24 

X 

X 

X 

X 

X 

X 

X 

10 

45 

X 

X 

X 

X 

X 

X 

X 

10 

67 

X 

X 

X 

X 

X 

X 

X 

11 

21 

X 

X 

X 

X 

X 

X 

X 

11 

39 

X 

X 

X 

X 

X 

X 

X 

1? 

46 

X 

X 

X 

X 

X 

X 

X 

12 

34 

X 

X 

X 

X 

X 

X 

X 

n 

58 

X 

X 

X 

X 

X 

X 

X 

13 

36 

X 

X 

X 

X 

X 

X 

X 

14 

55 

X 

X 

X 

X 

X 

X 

X 

14 

82 

X 

X 

X 

X 

X 

X 

X 

15 

36 

X 

X 

X 

X 

X 

X 

X 

15 

11 

X 

X 

X 

X 

X 

X 

X 

16 

32 

X 

X 

X 

X 

X 

X 

X 

16 

27 

X 

X 

X 

X 

X 

X 

X 

1  7 

1 1 

X 

X 

X 

X 

X 

X 

X 

17 

43 

X 

X 

X 

X 

X 

X 

X 

18 

31 

X 

X 

X 

X 

X 

X 

X 

18. 

49 

X 

X 

X 

X 

X 

X 

X 

19 

22 

X 

X 

X 

X 

X 

X 

X 

19 

97 

X 

'  X 

X 

X 

X 

X 

X 

20 

9 

X 

X 

X 

X 

X 

X 

X 

20 

89 

X 

X 

X 

X 

X 

X 

X 

21 

84 

X 

X 

X 

X 

X 

X 

X 

21 

62 

X 

X 

X 

X 

X 

X 

X 

22 

24 

X 

X 

X 

X 

X 

X 

X 

22 

85 

X 

X 

X 

X 

X 

X 

X 

22 

4 

X 

X 

X 

X 

X 

X 

X 

23 

22 

X 

X 

X 

X 

X 

X 

X 

24 

89 

X 

X 

X 

X 

X 

X 

X 

24 

44 

X 

X 

X 

X 

X 

X 

X 

25 

30 

X 

X 

X 

X 

X 

X 

X 

25 

54 

X 

X 

X 

X 

X 

X 

X 

26 

39 

X 

X 

X 

X 

X 

X 

X 

26 

76 

X 

X 

X 

X 

X 

X 

X 

27 

51 

X 

X 

X 

X 

X 

X 

X 

27 

3 

X 

X 

X 

X 

X 

X 

X 

28 

23 

X 

X 

X 

X 

X 

X 

X 

28 

96 

X 

X 

X 

X 

y 

X 

X 

29 

56 

X 

X 

X 

X 

X 

X 

X 

29 

88 

X 

X 

X 

X 

X 

X 

X 

30 

82 

X 

X 

X 

X 

X 

X 

X 

30 

12 

X 

X 

X 

X 

X 

X 

X 

31 

47 

X 

X 

X 

X 

X 

X 

X 

31 

84 

X 

X 

♦ 

X 

X 

X 

X 

X 

32 

67 

X 

X 

X 

X 

x 

X 

X 

32 

75 

X 

X 

X 

X 

X 

X 

X 

33 

98 

X 

X 

X 

X 

X 

X 

X 

33 

81 

X 

X 

X 

X 

X 

X 

X 

34 

61 

X 

X 

X 

X 

X 

X 

X 

34 

48 

X 

X 

X 

X 

X 

X 

X 

35 

41 

X 

X 

X 

X 

X 

X 

X 

35 

21 

X 

X 

X 

X 

X 

X 

X 

36 

65 

X 

X 

X 

X 

X 

X 

X 

36 

91 

X 

X 

X 

X 

X 

X 

X 

37 

96 

X 

X 

X 

X 

X 

X 

X 

37 

33 

X 

X 

X 

X 

X 

X 

X 

38 

97 

X 

X 

X 

X 

X 

X 

X 

38 

4° 

X 

X 

X 

X 

X 

X 

X 

39 

50 

X 

X 

X 

X 

X 

X 

X 

39 

60 

X 

X 

X 

X 

y 

X 

X 

40 

94 

X 

X 

X 

X 

X 

X 

X 

40 

65 

X 

X 

X 

X 

X 

X 

X 

41 

59 

X 

X 

X 

X 

X 

X 

X 

41 

68 

X 

X 

X 

X 

X 

X 

X 

42 

35 

X 

X 

X 

X 

X 

X 

X 

42 

35 

X 

X 

X 

V 

X 

X 

X 

43 

60 

X 

X 

X 

X 

X 

X 

X 

43 

26 

X 

X 

X 

X 

X 

X 

X 

44 

20 

X 

X 

X 

X 

X 

X 

X 

44 

100 

X 

X 

X 

X 

X 

X 

X 

45 

4? 

X 

X 

X 

X 

X 

X 

X 

45 

16 

X 

X 

X 

X 

X 

X 

X 

46 

75 

X 

X 

X 

X 

X 

X 

X 

46 

47 

X 

X 

X 

X 

X 

X 

X 

47 

2 

X 

X 

X 

X 

X 

X 

X 

47 

80 

X 

X 

X 

X 

X 

X 

X 

48 

74 

X 

X 

X 

X 

X 

X 

X 

48 

55 

X 

X 

X 

X 

X 

X 

X 

49 

8] 

X 

X 

X 

X 

X 

X 

X 

49 

32 

X 

X 

X 

X 

X 

X 

X 

50 

26 

X 

X 

X 

X 

X 

X 

X 

50 

61 

X 

X 

X 

X 

X 

X 

X 

51 

69 

X 

X 

X 

X 

X 

X 

X 

51 

72 

X 

X 

X 

X 

X 

X 

X 

52 

87 

X 

X 

X 

X 

X 

X 

X 

52 

93 

X. 

X 

X 

X 

X 

X 

Y 

53 

93 

X 

X 

X 

X 

X 

X 

X 

53 

46 

x' 

X 

X 

X 

X 

X 

X 

54 

28 

X 

X 

X 

X 

X 

X 

X 

54 

78 

X 

X 

X 

X 

X 

X 

X 

55 

13 

X 

X 

X 

X 

X 

X 

X 

55 

98 

X 

X 

X 

X 

X 

X 

X 

56 

99 

X 

X 

X 

X 

X 

X 

X 

56 

94 

X 

X 

X 

X 

X 

X 

X 

5  7 

48 

X 

X 

X 

X 

X 

X 

X 

57 

57 

X 

X 

X 

X 

X 

X 

X 

58 

14 

X 

X 

X 

X 

X 

X 

X 

58 

38 

X 

X 

X 

X 

X 

X 

X 

59 

68 

X 

X 

X 

X 

X 

X 

X 

59 

17 

X 

X 

X 

X 

X 

X 

X 

•  60 

43 

X 

X 

X 

X 

X 

X 

X 

60 

53 

X 

X 

X 

X 

X 

X 

X 

61 

10 

X 

X 

X 

X 

X 

X 

X 

61 

1 

X 

X 

X 

X 

X 

X 

X 

62 

.3 

X 

X 

X 

X 

X 

X 

X 

62 

73 

X 

X 

X 

X 

X 

X 

X 

63 

76 

X 

X 

X 

X 

X 

X 

X 

63 

18 

X 

X 

X 

X 

X 

X 

X 

64 

37 

X 

X 

X 

X 

X 

X 

X 

64 

92 

X 

X 

X 

X 

X 

X 

X 

65 

16 

X 

X 

X 

X 

X 

X 

X 

65 

4 

X 

X 

X 

X 

X 

X 

X 

66 

40 

X 

X 

X 

X 

X 

X 

X 

66 

74 

X 

X 

X 

X 

X 

X 

X 

67 

33 

X 

X 

X 

X 

X 

X 

X 

67 

6 

X 

X 

X 

X 

X 

X 

X 

68 

49 

X 

X 

X 

X 

X 

X 

X 

68 

14 

X 

X 

X 

X 

X 

X 

X 

69 

90 

X 

X 

X 

X 

X 

X 

X 

69 

83 

X 

X 

X 

X 

X 

X 

X 

70 

54 

X 

X 

X 

X 

X 

X 

X 

70 

13 

X 

X 

X 

X 

X 

X 

X 

71 

1  7 

X 

X 

X 

X 

X 

X 

X 

71 

30 

X 

X 

X 

X 

X 

X 

X 

72 

92 

X 

X 

X 

X 

X 

X 

X 

72 

40 

X 

X 

X 

X 

X 

X 

X 

73 

53 

X 

X 

X 

X 

X 

X 

X 

73 

41 

X 

X 

X 

X 

X 

X 

X 

74 

83 

X 

X 

X 

X 

X 

X 

X 

74 

58 

X 

X 

X 

X 

X 

X 

X 

75 

29 

X 

X 

X 

X 

X 

X 

X 

75 

8 

X 

X 

X 

X 

X 

X 

X 

76 

7 

X 

X 

X 

X 

X 

X 

X 

76 

20 

X 

X 

X 

X 

X 

X 

X 

77 

1 

X 

X 

X 

X 

X 

X 

X 

77 

23 

X 

X 

X 

X 

X 

X 

X 

78  100 

X 

X 

X 

X 

X 

X 

X 

78 

90 

X 

X 

X 

X 

X 

X 

X 

70 

16 

X 

X 

X 

X 

X 

X 

X 

79 

56 

X 

X 

X 

X 

X 

X 

X 

80 

91 

X 

X 

X 

X 

X 

X 

X 

.80 

29 

X 

X 

X 

X 

X 

X 

X 

81 

77 

X 

X 

X 

X 

X 

X 

X 

81 

37 

X 

X 

X 

X 

X 

X 

X 

82 

52 

X 

X 

X 

X 

X 

X 

X 

82 

10 

X 

X 

X 

X 

X 

X 

X 

d3 

57 

X 

X 

X 

X 

X 

X 

X 

83 

87 

X 

X 

X 

X 

X 

X 

X 

84 

44 

X 

X 

X 

X 

X 

X 

X 

84 

70 

X 

X 

X 

X 

X 

X 

X 

85 

34 

X 

X 

X 

X 

X 

X 

X 

85 

99 

X 

X 

X 

X 

X 

X 

X 

86 

78 

X 

X 

X 

X 

X 

X 

X 

86 

2 

X 

X 

X 

X 

X 

X 

X 

87 

85 

X 

X 

X 

X 

X 

X 

X 

87 

50 

X 

X 

X 

X 

X 

X 

X 

88 

95 

X 

X 

X 

X 

X 

X 

X 

88 

77 

X 

X 

X 

X 

X 

X 

X 

89 

70 

X 

X 

X 

X 

X 

X 

X 

89 

63 

X 

X 

X 

X 

X 

X 

X 

90 

71 

X 

X 

X 

X 

X 

X 

X 

90 

69 

X 

X 

X 

X 

X 

X 

X 

91 

72 

X 

X 

X 

X 

X 

X 

X 

91 

86 

X 

X 

X 

X 

X 

X 

X 

92 

73 

X 

X 

X 

X 

X 

X 

X 

92 

95 

X 

X 

X 

X 

X 

X 

X 

93 

8 

X 

X 

X 

X 

X 

X 

X 

93 

66 

X 

X 

X 

X 

X 

X 

X 

94 

12 

X 

X 

X 

X 

X 

X 

X 

94 

28 

X 

X 

X 

X 

X 

X 

X 

95 

19 

X 

X 

X 

X 

X 

X 

X 

95 

79 

X 

X 

X 

X 

X 

X 

X 

96 

66 

X 

X 

X 

X 

X 

X 

X 

96 

59 

X 

X 

X 

X 

X 

X 

X 

‘,7 

62 

X 

X 

X 

X 

X 

X 

X 

97 

7 

X 

X 

X 

X 

X 

X 

X 

98 

6.3 

X 

X 

X 

X 

X 

X 

X 

98 

64 

X 

X 

X 

X 

X 

X 

X 

99 

64 

X 

X 

X 

X 

X 

X 

X 

99 

71 

X 

X 

X 

X 

X 

X 

X 

100 

86 

X 

X 

X 

X 

X 

X 

X 

100 

19 

X 

X 

X 

X 

X 

X 

X 

f 

19 

24 

31 

32 

36 

49 

58 

81 

76 

69 

68 

64 

51 

42 

f 

26 

36 

46 

48 

62 

67 

74 

74 

64 

54 

52 

38 

33 

26 

1  rrorb 

7 

10 

15 

10 

10 

9 

0 

u 

2 

3 

5 

5 

4 

7 

Hrrors 

10 

9 

11 

7 

10 

6 

n 

0 

6 

10 

13 

8 

7 

10 

FIGURE  4 

SCALOGRAMS  FOR  STATUS  ITEMS 


121 


(6)  A  teacher  should  not  he  expected  to  consult  the  principal 
before  administering  corporal  punishment  to  a  pupil.  (SA,A,U) 

(7)  A  principal  should  encourage  teachers  to  refer  serious 

•  behavioral  problems  to  him  only  as  a  last  resort.  (SA,A,U) 

When  these  items  were  tested  for  scalability,  they  were  found  to  have  a  low 
coefficiency  of  reproducibility  of  .83  but  a  generally  random  pattern  of 
errors  as  indicated  by  Figure  5»  In  addition  to  this,  the  modal  frequencies 
of  each  of  six  items  contained  between  3C $  and  "JO/o  of  the  total  responses 
for  that  particular  item.  This  would  probably  justify  treating  the  items 
as  a  quasi  scale  which  is  what  was  done  in  fact,  but  it  also  suggests  that 
teachers  may  not  be  too  consistent  in  what  they  expect  of  the  principal  in 
this  particular  area. 

Since  three  quasi  scales  containing  seven  items  each  had  now  been 
developed,  it  was  decided  to  select  the  most  appropriate  seven  of  the 
remaining  items  in  this  category  to  test  for  scalability.  It  was  felt  that 
there  would  be  some  value  in  having  the  same  range  of  scores  for  each  of  the 
scales  for  purposes  of  future  analysis.  However,  it  must  be  made  clear  that 
the  seven  items  were  selected  by  reducing  larger  lists  and  not  by  forcing 
unacceptable  items  into  a  particular  series. 

The  seven  items  which  were  selected  for  analysis  yielded  a  coefficient 
of  reproducibility  of  .90  and  might  be  classified  as  a  scale  on  the  basis  of 
this  criterion.  However,  when  the  range  of  marginal  frequencies  is  examined, 
it  is  evident  that  the  reproducibility  may  be  spuriously  high  due  to  some 
extreme  marginals.  In  fact  only  two  of  the  seven  items  have  between  thirty 
and.  seventy  per  cent  of  their  responses  in  one  of  the  two  response  categories. 
For  this  reason  the  set  of  items  can  be  considered  as  approaching  a  scale, 
but  caution  suggests  that  they  be  classified  as  forming  a  quasi  scale  as 
do  the  other  sets  of  items  considered  so  far.  The  items  in  question  (Ao) 


' 

■ 

'■ 

-  " v ■  -  j- 


.  ^  -  .  ,  ■ 

■ 

'  '•  •  •  ,  '  .  :  ' 

' 

.  '  . 

,  •.  .  .  .  . 


. 


122 


AUIHOKITY  NO.  1  ITEMS  AUTHORITY  NO.  2  ITEMS 

(Rep.  =  ,b3)  (Rep.  =  .VO) 


1 

2 

3 

4 

5 

6 

7 

8 

9 

10 

11 

12 

13 

14 

1 

2 

3 

4 

5 

6 

7 

8 

9 

10 

11 

12 

13 

1 

70 

X 

X 

X 

X 

X 

X 

X 

1  63 

X 

X 

X 

X 

X 

X 

X 

2 

57 

X 

X 

X 

X 

X 

X 

X 

2  46 

X 

X 

X 

X 

X 

X 

X 

3 

4 

X 

X 

X 

X 

X 

X 

X 

3  4 

X 

X 

X 

X 

X 

X 

X 

4 

55 

X 

X 

X 

X 

X 

X 

X 

4  1 

X 

X 

X 

X 

X 

X 

X 

5 

37 

X 

X 

X 

X 

X 

X 

X 

5  20 

X 

X 

X 

X 

X 

X 

X 

6 

41 

X 

X 

X 

X 

X 

X 

X 

6  30 

X 

X 

X 

X 

X 

X 

X 

7 

35 

X 

X 

X 

X 

X 

X 

X 

7  16 

X 

X 

X 

X 

X 

X 

X 

b 

4b 

X 

X 

X 

X 

X 

X 

X 

b  61 

X 

X 

X 

X 

X 

X 

X 

9 

63 

X 

X 

X 

X 

X 

X 

X 

9  19 

X 

X 

X 

X 

X 

X 

X 

10 

89 

X 

X 

X 

X 

X 

X 

X 

10  64 

X 

X 

X 

X 

X 

X 

X 

11 

59 

X 

X 

X 

X 

X 

X 

X 

11  37 

X 

X 

X 

X 

X 

X 

X 

12 

b6 

X 

X 

X 

X 

X 

X 

X 

12  7 

X 

X 

X 

X 

X 

X 

X 

13 

7b 

X 

X 

X 

X 

X 

X 

X 

13  40 

X 

X 

X 

X 

X 

X 

X 

14 

71 

X 

X 

X 

X 

X 

X 

X 

14  62 

X 

X 

X 

X 

X 

X 

X 

15 

74 

X 

X 

X 

X 

X 

X 

X 

15  27 

X 

X 

X 

X 

X 

X 

X 

16 

27 

X 

X 

X 

X 

X 

X 

X 

16  45 

X 

X 

X 

X 

X 

X 

X 

1  7 

16 

X 

X 

X 

X 

X 

X 

X 

17  10 

X 

X 

X 

X 

X 

X 

X 

lb 

14 

X 

X 

X 

X 

X 

X 

X 

18  86 

X 

X 

X 

X 

X 

X 

X 

19 

18 

X 

X 

X 

X 

X 

X 

X 

19  51' 

X 

X 

X 

X 

X 

X 

X 

20 

53 

X 

X 

X 

X 

X 

X 

X 

20  5  7 

X 

X 

X 

X 

* 

X 

X 

21 

82 

X 

X 

X 

X 

X 

X 

X 

21  33 

X 

X 

X 

X 

X 

X 

X 

22 

96 

X 

X 

X 

X 

X 

X 

X 

22  75 

X 

X 

X 

X 

X 

X 

X 

23 

64 

X 

X 

X 

X 

X 

X 

X 

23  50 

X 

X 

X 

X 

X 

X 

X 

24 

50 

X 

X 

X 

X 

X 

X 

X 

24  48 

X 

X 

X 

X 

X 

X 

X 

25 

67 

X 

X 

X 

X 

X 

X 

X 

25  96 

X 

X 

X 

X 

X 

X 

X 

26 

19 

X 

X 

X 

X 

X 

X 

X 

26  88 

X 

X 

X 

X 

X 

X 

X 

27 

21 

X 

X 

X 

X 

X 

X 

X 

27  69 

X 

X 

X 

X 

X 

X 

X 

2b 

32 

X 

X 

X 

X 

X 

X 

X 

28  66 

X 

X 

X 

X 

X 

X 

X 

29 

3 

X 

X 

X 

X 

X 

X 

X 

29  49 

X 

X 

X 

X 

X 

X 

X 

30 

84 

X 

X 

X 

X 

X 

X 

X 

30  87 

X 

X 

X 

X 

X 

X 

X 

31 

11 

X 

X 

X 

X 

X 

X 

X 

31  56 

X 

X 

X 

X 

X 

X 

32 

66 

X 

X 

X 

X 

X 

X 

X 

32  71 

X 

X 

X 

X 

X 

X 

X 

33 

25 

X 

X 

X 

X 

X 

X 

X 

33  44 

X 

X 

X 

X 

X 

X 

X 

34 

30 

X 

X 

X 

X 

X 

X 

X 

34  17 

X 

X 

X 

X 

X 

X 

X 

35 

61 

X 

X 

X 

X 

X 

X 

X 

35  2 

X 

X 

X 

X 

X 

X 

X 

36 

76 

X 

X 

X 

X 

X 

X 

X 

36  55 

X 

X 

X 

X 

X 

X 

X 

37 

8 

X 

X 

X 

X 

X 

X 

X 

37  93 

X 

X 

X 

X 

X 

X 

X 

3b 

95 

X 

X 

X 

X 

X 

X 

X 

38  6 

X 

X 

X 

X 

X 

X 

X 

39 

23 

X 

X 

X 

X 

X 

X 

X 

39  8 

X 

X 

X 

X 

X 

X 

X 

40 

94 

X 

X 

X 

X 

X 

X 

X 

40  11 

X 

X 

X 

X 

X 

X 

X 

41 

39 

X 

X 

X 

X 

X 

X 

X 

41  12 

X 

X 

X 

X 

X 

3* 

X 

42 

92 

X 

X 

X 

X 

X 

X 

X 

42  14 

X 

X 

X 

X 

X 

X 

X 

43 

29 

X 

X 

X 

X 

X 

X 

X 

43  28 

X 

X 

X 

X 

X 

X 

X 

44 

85 

X 

X 

X 

X 

X 

X 

X 

44  74 

X 

X 

X 

X 

X 

X 

X 

45 

68 

X 

X 

X 

X 

X 

X 

X 

45  77 

X 

X 

X 

X 

X 

X 

X 

46 

22 

X 

X 

X 

X 

X 

X 

X 

46  60 

X 

X 

X 

X 

X 

X 

X 

47 

73 

X 

X 

X 

X 

X 

X 

X 

4  7  72 

X 

X 

X 

X 

X 

X 

X 

48 

15 

X 

X 

X 

X 

X 

X 

X 

48  91 

X 

X 

X 

X 

X 

X 

X 

49 

28 

X 

X 

X 

X 

X 

X 

X 

49  90 

X 

X 

X 

X 

X 

X 

X 

50 

42 

X 

X 

X 

X 

X 

X 

X 

50  98 

X 

X 

X 

X 

X 

X 

X 

51 

1 

X 

X 

X 

X 

X 

X 

X 

51  80 

X 

X 

X 

X 

X 

X 

X 

52 

93 

X 

X 

X 

X 

X 

X 

X 

52  94 

X 

X 

X 

X 

X 

X 

X 

53 

51 

X 

X 

X 

X 

X 

X 

X 

53  13 

X 

X 

X 

X 

X 

X 

X 

54 

62 

X 

X 

X 

X 

X 

X 

X 

54  3 

X 

X 

X 

X 

X 

X 

X 

55 

83 

X 

X 

X 

X 

X 

X 

X 

55  58 

X 

X 

X 

X 

X 

X 

56 

75 

X 

X 

X 

X 

X 

X 

X 

56  76 

X 

X 

X 

X 

X 

X 

57 

34 

X 

X 

X 

X 

X 

X 

X 

57  18 

X 

X 

X 

X 

X 

X 

58 

7 

X 

X 

X 

X 

X 

X 

X 

58  24 

X 

X 

X 

X 

X 

X 

X 

59 

31 

X 

X 

X 

X 

X 

X 

X 

59  73 

X 

X 

X 

X 

X 

X 

X 

60 

45 

X 

X 

X 

X 

X 

X 

X 

60  100 

X 

X 

X 

X 

X 

X 

X 

61 

26 

X 

X 

X 

X 

X 

X 

X 

61  9 

X 

X 

X 

X 

X 

X 

X 

62 

88 

X 

X 

X 

X 

X 

X 

X 

62  26 

X 

X 

X 

'  X 

X 

X 

X 

63 

13 

X 

X 

X 

X 

X 

X 

X 

63  25 

X 

X 

X 

X 

X 

X 

X 

64 

98 

X 

X 

X 

X 

X 

X 

X 

64  29 

X 

X 

X 

X 

X 

X 

X 

65 

20 

X 

X 

X 

X 

X 

X 

X 

65  21 

X 

X 

X 

X 

X 

X 

X 

66 

80 

X 

X 

X 

X 

X 

X 

X 

66  39 

X 

X 

X 

X 

X 

X 

X 

67 

6 

X 

X 

X 

X 

X 

X 

X 

6  7  35 

X 

X 

X 

X 

X 

X 

X 

6b 

43 

X 

X 

X 

X 

X 

X 

X 

6b  84 

X 

X 

X 

X 

X 

X 

X 

69 

40 

X 

X 

X 

X 

X 

X 

X 

69  43 

X 

X 

X 

X 

X 

X 

X 

70 

91 

X 

X 

X 

X 

X 

X 

X 

70  89 

X 

X 

X 

X 

X 

X 

X 

71 

69 

X 

X 

X 

X 

X 

X 

X 

71  92 

X 

X 

X 

X 

X 

X 

X 

72 

2 

X 

X 

X 

X 

X 

X 

X 

72  99 

X 

X 

X 

X 

X 

X 

X 

73 

99 

X 

X 

X 

X 

X 

X 

X 

73  59 

X 

X 

X 

X 

X 

X 

X 

74 

97 

X 

X 

X 

X 

X 

X 

X 

74  5 

X 

X 

X 

X 

X 

X 

X 

75 

79 

X 

X 

X 

X 

X 

X 

X 

75  82 

X 

X 

X 

X 

X 

X 

X 

76 

5 

X 

X 

X 

X 

X 

X 

X 

76  32 

X 

X 

X 

X 

X 

X 

X 

77 

77 

X 

X 

X 

X 

X 

X 

X 

77  85 

X 

X 

X 

X 

X 

X 

X 

7b 

24 

X 

X 

X 

X 

X 

X 

X 

78  52 

X 

X 

X 

X 

X 

X 

X 

79 

33 

X 

X 

X 

X 

X 

X 

X 

79  78 

X 

X 

X 

X 

X 

x 

X 

bO 

1  7 

X 

X 

X 

X 

X 

X 

X 

80  41 

X 

X 

X 

X 

X 

X 

X 

81 

44 

X 

X 

X 

X 

X 

X 

X 

81  53 

X 

X 

X 

X 

X 

X 

X 

82 

52 

X 

X 

X 

X 

X 

X 

X 

82  70 

X 

X 

X 

X 

X 

X 

X 

83 

9 

X 

X 

X 

X 

X 

X 

X 

83  42 

X 

X 

X 

X 

X 

X 

X 

84 

90 

X 

X 

X 

X 

X 

X 

X 

84  22 

X 

X 

X 

X 

X 

X 

X 

85 

54 

X 

X 

X 

X 

X 

X 

•X 

85  6  7 

X 

X 

X 

X 

X 

X 

X 

86 

49 

X 

X 

X 

X 

X 

X 

X 

86  65 

X 

X 

X 

X 

X 

X 

X 

87 

81 

X 

X 

X 

X 

X 

X 

X 

87  6b 

X 

X 

X 

X 

X 

X 

X 

8b 

58 

X 

X 

X 

X 

X 

X 

X 

88  81 

X 

X 

X 

X 

X 

X 

X 

89 

47 

X 

X 

X 

X 

X 

X 

X 

89  97 

X 

X 

X 

X 

X 

X 

X 

90 

12 

X 

X 

X 

X 

X 

X 

X 

90  34 

X 

X 

X 

X 

X 

X 

X 

91 

56 

X 

X 

X 

X 

X 

X 

X 

91  54 

X 

X 

X 

X 

X 

X 

X 

92 

10 

X 

X 

X 

X 

X 

X 

X 

92  38 

X 

X 

X 

X 

X 

X 

93 

38 

X 

X 

X 

X 

X 

X 

X 

93  4  7 

X 

X 

X 

X 

X 

X 

94 

72 

X 

X 

X 

X 

X 

X 

X 

94  23 

X 

X 

X 

X 

X 

X 

95 

60 

X 

X 

X 

X 

X 

X 

X 

95  36 

X 

X 

X 

X 

X 

X 

96 

65 

X 

X 

X 

X 

X 

X 

X 

96  31 

X 

X 

X 

X 

X 

X 

97 

87 

X 

X 

X 

X 

X 

X 

X 

97  83 

X 

X 

X 

X 

X 

X 

98 

46 

X 

X 

X 

X 

X 

X 

X 

98  95 

X 

X 

X 

X 

X 

X 

99  . 

100 

X 

X 

X 

X 

X 

X 

X 

99  15 

X 

X 

X 

X 

X 

X 

100 

36 

X 

X 

X 

X 

X 

X 

X 

100  79 

X 

X 

X 

X 

X 

X 

f 

26 

38 

39 

55 

67 

68 

70 

74 

62 

61 

45 

33 

32 

30 

f 

15 

22 

33 

66 

74 

84 

87' 

85 

78 

67 

34 

26 

16 

Lrrorf. 

JO 

18 

8 

12 

6 

8 

0 

0 

5 

5 

13 

3 

14 

15 

Errors 

7 

5 

8 

11 

5 

3 

0 

0 

3 

9 

7 

7 

3 

FIGURE  5 

5CAL0GRAMS  FOR  AUTHORITY  ITEMS 


were  the  following: 


125 


(1)  The  principal  should  expect  teachers  to  consult 
with  him  before  making  any  final  decisions  about 
promotions.  (D,SD) 

(2)  Even  the  decision  of  whether  or  not  a  teacher 
participates  in  policy  development  should  be 
left  to  the  teacher.  (SA,A) 

(3)  It  is  reasonable  for  a  principal  to  expect  new 
teachers  to  fit  themselves  into  the  traditional 
ways  of  the  school.  (U,D,SD) 

(4)  A  principal  should  encourage  teachers  to  look 

to  him  for  guidance  concerning  appropriate  modes 
of  dress  and  conduct  in  the  community.  (D,SD) 

(5)  Prior  to  issuing  reports  to  parents,  teachers 
should  submit  all  comments  and  gradings  to  the 
principal  for  his  approval.  (D,SD) 

(6)  A  principal  should  expect  teachers  to  submit  to 
him  for  his  approval  copies  of  all  examinations 
which  they  plan  to  administer.  (U,D,SD) 

(7)  In  delegating  some  definite  non-teaching  respon¬ 
sibilities  to  teachers,  a  principal  should  give 
them  full  authority  to  act  as  they  see  fit.  (SA,AU  ) 

The  high  coefficient  of  reproducibility  for  these  items  does  suggest  that 

if  an  area  is  well-defined,  the  expectations  which  teachers  hold  for  the 

behavior  of  a  principal  may  approach  scalability  much  more  closely  than 

the  other  sets  of  items  would  have  led  one  to  believe. 


Personal  Dimension 

The  personal  dimension  concerns  the  extent  to  which  a  principal 
should  engage  in  particularistic  as  opposed,  to  universalis  tic  behavior 
in  his  relations  with  the  staff.  The  problem  with  this  dimension  was  the 
same  as  it  was  with  all  of  the  others  in  that  the  sets  of  items  were  not 
scalable  in  the  way  that  was  hypothesized.  A  closer  examination  of  the 
items  suggested  that  there  were  some  which  concerned  the  extent  to  which 
a  principal  should  engage  in  closer  social  relationships  with  some 
teachers  than  with  others  while  other  items  were  more  concerned  with  the 


extent  to  which  a  principal  should  take  individual  differences  into  account 


•  ■  '  '■  ;i .  .  %  '  • : 

.  . 

. 

. 

•  r 

' 

■■  • .  ■ 

■  ’  • 

.  /  .  •  :  '  . 

■'  '■■•Vi.  i  '  • 

-  ;  •  . 


, 


124 


when  assigning  duties  or  granting  special  requests. 

In  the  first  subarea  ten  items  were  tested  for  scalability,  and  a 
coefficient  of  reproducibility  of  .84  was  obtained.  The  scalogram  pattern 
did  reveal  some  non-scale  types  indicating  that  more  than  one  dominant 
variable  was  involved.  It  was  decided  to  eliminate  the  three  least 
satisfactory  items  in  order  to  attempt  to  increase  the  reproducibility 
and  to  eliminate  the  non-scale  types.  The  seven  items  (Pp)  which  were 
retained  consisted  of  the  following: 

(1)  A  principal  should  not  visit  the  homes  of  some  teachers 
any  more  often  than  he  visits  the  homes  of  other  teachers. 
(SA,A) 

(2)  A  principal  should  invite  teachers  to  his  home  only  as 
a  group  if  he  invites  them  at  all.  (SA,A,U) 

(3)  A  principal  cannot  carry  out  his  duties  effectively 
if  he  has  close  friends  on  the  staff.  (SA,A,U) 

(4)  A  principal  should  restrict  his  relationships 
with  staff  members  to  the  formal  requirements 

in  order  to  avoid  preferential  treatment  of  some 
teachers.  (SA,A,U) 

(5)  A  principal  should  not  have  close  relatives  on 
his  teaching  staff.  (SA,A,U) 

(6)  A  principal  should  not  discuss  his  personal 
problems  with  any  of  his  teachers.  (SA,A,U) 

(7)  A  principal  should  do  personal  favors  for  staff 
members.  (U,D,SD) 

The  coefficient  of  reproducibility  for  these  items  was  .88  and  Figure  6, 
which  contains  the  scalogram,  shows  that  the  pattern  of  errors  was  random. 
Four  of  the  items  also  had.  acceptable  marginal  frequencies  in  each  cate¬ 
gory,  and  on  the  basis  of  this  and  the  other  criteria  these  seven  items 
could  be  considered  as  forming  a  quasi  scale. 

The  following  items  (P2)  were  considered  to  have  common  content 
related  to  the  extent  to  which  a  principal  should  take  individual 


, 


, 


■ 


■ 


125 


PERSONAL  NO.  ;  ITLMS 
(Rep.  =  .88) 


PERSONAL  NO.  2  ITEMS 
(Rep.  -  .87) 


Hank 

No. 

1 

2 

3 

4 

5 

6 

7 

8 

9 

10 

11 

12 

13 

14 

Rank 

No. 

1  2 

3 

4 

5 

6 

7 

8 

9 

10 

11 

12 

13 

14 

I 

5 

X 

X 

X 

X 

X 

X 

X 

1 

3 

X  X 

X 

X 

X 

X 

X 

2 

22 

X 

X 

X 

X 

X 

X 

X 

2 

11 

X  X 

X 

X 

X 

X 

X 

3 

36 

X 

X 

X 

X 

X 

X 

X 

3 

57 

X  X 

X 

X 

X 

X 

X 

4 

66 

X 

X 

X 

X 

X 

X 

X 

4 

80 

X  X 

X 

X 

X 

X 

X 

5 

7 

X 

X 

X 

X 

X 

X 

X 

5 

94 

X  X 

X 

X 

X 

X 

X 

6 

21 

X 

X 

X 

X 

X 

X 

X 

6 

39 

X  X 

X 

X 

X 

X 

X 

7 

27 

X 

X 

X 

X 

X 

X 

X 

7 

9 

X  X 

X 

X 

X 

X 

X 

8 

89 

X 

X 

X 

X 

X 

X 

X 

8 

21 

X  X 

X 

X 

X 

X 

X 

9 

15 

X 

X 

X 

X 

X 

X 

X 

9 

36 

X  X 

X 

X 

X 

X 

X 

10 

8 

X 

X 

X 

X 

X 

X 

X 

10 

15 

X  X 

X 

X 

A 

X 

X 

11 

61 

X 

X 

X 

X 

X 

X 

X 

11 

17 

X  X 

X 

X 

X 

X 

X 

i2 

1  7 

X 

X 

X 

X 

X 

X 

X 

12 

69 

X  X 

X 

X 

X 

X 

X 

13 

55 

X 

X 

X 

X 

X 

X 

X 

13 

81 

X  X 

X 

X 

X 

X 

X 

14 

57 

X 

X 

X 

X 

X 

X 

X 

14 

77 

X  X 

X 

X 

X 

X 

X 

15 

48 

X 

X 

X 

X 

X 

X 

X 

15 

7 

X 

X 

X 

X 

X 

X 

X 

16 

84 

X 

X 

X 

X 

X 

X 

X 

16 

82 

X 

X 

X 

X 

X 

X 

X 

17 

81 

X 

X 

X 

X 

X 

X 

X 

17 

55 

X 

X 

X 

X 

X 

X 

X 

18 

65 

X 

X 

X 

X 

X 

X 

X 

18 

87 

X 

X 

X 

X 

X 

X 

X 

19 

20 

X 

X 

X 

X 

X 

X 

X 

19 

31 

X 

X 

X 

X 

X 

X 

X 

20 

38 

X 

X 

X 

X 

X 

X 

X 

20 

74 

X 

.  X 

X 

X 

X 

X 

X 

21 

52 

X 

X 

X 

X 

X 

X 

X 

21 

41 

X 

X 

X 

X 

X 

X 

X 

22 

43 

X 

X 

X 

X 

X 

X 

X 

22 

72 

X 

X 

X 

X 

X 

X 

X 

23 

78 

X 

X 

X 

X 

X 

X 

X 

23 

47 

X 

X 

X 

X 

X 

X 

X 

24 

40 

X 

X 

X 

X 

X 

X 

X 

24 

12 

X 

X 

X 

X 

X 

X 

X 

25 

32 

X 

X 

X 

X 

X 

X 

X 

25 

89 

X 

X 

X 

X 

X 

X 

X 

26 

16 

X 

X 

X 

X 

X 

X 

X 

26 

86 

X 

X 

X 

X 

X 

X 

X 

27 

14 

X 

X 

X 

X 

X 

X 

X 

27 

29 

X 

X 

X 

X 

X 

X 

X 

28 

42 

X 

X 

X 

X 

X 

X 

X 

28 

32 

X 

X 

X 

X 

X 

X 

X 

29 

95 

X 

X 

X 

X 

X 

X 

X 

29 

28 

X 

X 

X 

X 

X 

X 

X 

30 

98 

X 

X 

X 

X 

X 

X 

X 

30 

59 

X 

X 

X 

X 

X 

X 

X 

31 

31 

X 

X 

X 

X 

X 

X 

X 

31 

56 

X 

X 

X 

X 

X 

X 

X 

32 

3 

X 

X 

X 

X 

X 

X 

X 

32 

43 

X 

X 

X 

X 

X 

X 

X 

33 

90 

X 

X 

X 

X 

X 

X 

X 

33 

38 

X 

X 

X 

X 

X 

X 

X 

34 

93 

X 

X 

X 

X 

X 

X 

X 

34 

76 

X 

X 

X 

X 

X 

X 

X 

35 

47 

X 

X 

X 

X 

X 

X 

X 

35 

25 

X 

X 

X 

X 

X 

X 

X 

36 

37 

X 

X 

X 

X 

X 

X 

X 

36 

58 

X 

X 

X 

X 

X 

X 

X 

37 

53 

X 

X 

X 

X 

X 

X 

X 

37 

14 

X 

X 

X 

X 

X 

X 

X 

38 

41 

X 

X 

X 

X 

X 

X 

X 

38 

33 

X 

X 

X 

X 

X 

X 

X 

39 

9 

X 

X 

X 

X 

X 

X 

X 

39 

88 

X 

X 

X 

X 

X 

X 

X 

40 

26 

X 

X 

X 

X 

X 

X 

X 

40 

4 

X 

X 

X 

X 

X 

X 

X 

41 

35 

X 

X 

X 

X 

X 

X 

X 

41 

95 

X 

X 

X 

X 

X 

X 

X 

42 

97 

X 

X 

X 

X 

X 

X 

X 

42 

23 

X 

X 

X 

X 

X 

X 

X 

43 

11 

X 

X 

X 

X 

X 

X 

X 

43 

8 

X 

X 

X 

X 

X 

X 

X 

44 

24 

X 

X 

X 

X 

X 

X 

X 

44 

93 

X 

X 

X 

X 

X 

X 

X 

45 

59 

X 

X 

X 

X 

X 

X 

X 

45 

63  ; 

X 

X 

X 

X 

X 

X 

X 

46 

94 

X 

X 

X 

X 

X 

X 

X 

46 

54 

X 

X 

X 

X 

X 

X 

X 

47 

28 

X 

X 

X 

X 

X 

X 

X 

47 

85 

X 

X 

X 

X 

X 

X 

X 

48 

99 

X 

X 

X 

X 

X 

X 

X 

48 

60 

X 

X 

X 

X 

X 

X 

X 

49 

74 

X 

X 

X 

X 

X 

X 

X 

49 

10 

X 

X 

X 

X 

X 

X 

X 

50 

39 

X 

X 

X 

X 

X 

X 

X 

50 

91 

X 

X 

X 

X 

X 

X 

X 

51 

71 

X 

X 

X 

X 

X 

X 

X 

51 

75 

X 

X 

X 

X 

X 

X 

X 

52 

64 

X 

X 

X 

X 

X 

X 

X 

52 

22 

X 

X 

X 

X 

X 

X 

X 

53 

68 

X 

X 

X 

X 

X 

X 

X 

53 

90 

X 

X 

X 

X 

X 

X 

X 

54 

46 

X 

X 

X 

X 

X 

X 

X 

54 

13 

X 

X 

X 

X 

X 

X 

X 

55 

25 

X 

X 

X 

X 

X 

X 

X 

55 

5 

X 

X 

X 

X 

X 

X 

X 

56 

45 

X 

X 

X 

X 

X 

X 

X 

56 

49 

X 

X 

X 

X 

X 

X 

X 

57 

88 

X 

X 

X 

X 

X 

X 

X 

57 

84 

X 

X 

X 

X 

X 

X 

X 

58 

18 

X 

X 

X 

X 

X 

X 

X 

58 

78 

X 

X 

X 

X 

X 

X 

X 

59 

49 

X 

X 

X 

X 

X 

X 

X 

59 

79 

X 

X 

X 

X 

X 

X 

X 

60 

30 

X 

X 

X 

X 

X 

X 

X 

60 

70 

X 

X 

X 

X 

X 

X 

X 

61 

86 

X 

X 

X 

X 

X 

X 

X 

61 

100 

X 

X 

X 

X 

X 

X 

X 

62 

82 

X 

X 

X 

X 

X 

X 

X 

62 

35 

X 

X 

X 

X 

X 

X 

X 

63 

54 

X 

X 

X 

X 

X 

X 

X 

63 

50 

X 

X 

X 

X 

X 

X 

X 

64 

75 

X 

X 

X 

X 

X 

X 

X 

64 

27 

X 

X 

X 

X 

X 

X 

X 

65 

92 

X 

X 

X 

X 

X 

X 

X 

65 

62 

X 

X 

X 

X 

X 

X 

X 

66 

70 

X 

X 

X 

X 

X 

X 

X 

66 

40 

X 

X 

X 

X 

X 

X 

X 

67 

1 

X 

X 

X 

X 

X 

X 

X 

67 

19 

X 

X 

X 

X 

X 

X 

X 

68 

60 

X 

X 

X 

X 

X 

X 

X 

68 

6 

X 

X 

X 

X 

X 

X 

X 

69 

91 

X 

X 

X 

X 

X 

X 

X 

69 

48 

X 

X 

X 

X 

X 

X 

X 

70 

58 

X 

X 

X 

X 

X 

X 

X 

70 

64 

X 

X 

X 

X 

X 

X 

X 

71 

33 

X 

X 

X 

X 

X 

X 

X 

71 

99 

X 

X 

X 

X 

X 

X 

X 

72 

4 

X 

X 

X 

X 

X 

X 

X 

72 

34 

X 

X 

X 

X 

X 

X 

X 

73 

6 

X 

X 

X 

X 

X 

X 

X 

73 

44 

X 

X 

X 

X 

X 

X 

X 

74 

10 

X 

X 

X 

X 

X 

X 

X 

74 

67 

X 

X 

X 

X 

X 

X 

X 

75 

51 

X 

X 

X 

X 

X 

X 

X 

75 

1 

X 

X 

X 

X 

X 

X 

X 

76 

80 

X 

X 

X 

X 

X 

X 

X 

76 

61 

X 

X 

X 

X 

X 

X 

X 

77 

23 

X 

X 

X 

X 

X 

X 

X 

77 

2 

X 

X 

X 

X 

X 

X 

X 

78 

83 

X 

X 

X 

X 

X 

X 

X 

78 

96 

X 

X 

X 

X 

X 

X 

X 

79 

44 

X 

X 

X 

X 

X 

X 

X 

79 

66 

X 

X 

X 

X 

X 

X 

X 

80 

76 

X 

X 

X 

X 

X 

X 

X 

80 

71 

X 

X 

X 

X 

X 

X 

X 

81 

79 

X 

X 

X 

X 

X 

X 

X 

81 

51 

X 

X 

X 

X 

X 

X 

X 

82 

62 

X 

X 

X 

X 

X 

X 

X 

82 

26 

X 

X 

X 

X 

X 

X 

X 

83 

12 

X 

X 

X 

X 

X 

X 

X 

83 

98 

X 

X 

X 

X 

X 

X 

X 

84 

87 

X 

X 

X 

X 

X 

X 

X 

84 

45 

X 

X 

X 

X 

X 

X 

X 

85 

72 

X 

X 

X 

X 

X 

X 

X 

85 

52 

X 

X 

X 

X 

X 

X 

X 

86 

96 

X 

X 

X 

X 

X 

X 

X 

86 

68 

X 

X 

X 

X 

X 

X 

X 

87 

63 

X 

X 

X 

X 

X 

X 

X 

87 

30 

X 

X 

X 

X 

X 

X 

X 

88 

50 

X 

X 

X 

X 

X 

X 

X 

88 

65 

X 

* 

X 

X 

X 

X 

X 

89  100 

X 

X 

X 

X 

X 

X 

X 

89 

42 

X 

X 

X 

X 

X 

X 

X 

90 

34 

X 

X 

X 

X 

X 

X 

X 

90 

20 

X 

X 

X 

X 

X 

X 

X 

91 

69 

X 

X 

X 

X 

X 

X 

X 

91 

16 

X 

X 

X 

X 

X 

X 

X 

92 

73 

X 

X 

X 

X 

X 

X 

X 

92 

37 

X 

X 

X 

X 

X 

X 

X 

93 

29 

X 

X 

X 

X 

X 

X 

X 

93 

46 

X 

X 

X 

X 

X 

X 

X 

94 

85 

X 

X 

X 

X 

X 

X 

X 

94 

97 

X 

X 

X 

X 

X 

X 

X 

95 

56 

X 

X 

X 

X 

X 

X 

X 

95 

24 

X 

X 

X 

X 

X 

X 

X 

96 

2 

X 

X 

X 

X 

X 

X 

X 

96 

73 

X 

X 

X 

X 

X 

X 

X 

9/ 

19 

X 

X 

X 

X 

X 

X 

X 

97 

83 

X 

X 

X 

X 

X 

X 

X 

9H 

67 

X 

X 

X 

X 

X 

X 

X 

98 

53 

X 

X 

X 

X 

X 

X 

X 

99 

77 

X 

X 

X 

X 

X 

X 

X 

99 

18 

X 

X 

X 

X 

X 

X 

X 

mo 

13 

X 

X 

X 

X 

X 

X 

X 

100 

92 

X 

X 

X 

X 

X 

X 

X 

f 

18 

28 

.38 

46 

52 

66 

75 

82 

72 

62 

54 

48 

34 

25 

f  19 

30  1 

53 

61 

66  1 

67  i 

83 

81 

70 

47 

39 

34 

33 

17 

Errors 

7 

4 

6 

6 

4 

4 

0 

0 

4 

6 

6 

11 

13 

11 

Errors  2 

10 

14 

12 

10 

3 

0 

0 

T 

5 

3 

1 1 

8 

9 

FIGURE  6 

SCALOCRAMS  FOR  PERSONAL  ITEMS 


126 


differences  into  account  when  assigning  duties: 

(1)  In  assigning  duties  a  principal  should  he  cognizant 

of  the  out-of-school  responsibilities  of  a  teacher.  (D,SD) 

(2)  A  principal  should  take  into  account  such  personal 
obligations  of  teachers  as  family  responsibilities 
when  assigning  extra-curricular  duties.  (u,D,SD) 

(3)  A  principal  should  refuse  to  assist  teachers  who 
apply  to  the  board  for  some  special  concession 
such  as  permission  to  leave  school  for  a  few 
days  before  the  end  of  the  term  so  tnat  they  can 
have  a  longer  holiday.  (SA,A) 

(4)  A  principal  should  be  more  concerned  with  qualifications 
than  with  teacher  preferences  when  assigning  teaching 
duties.  (SA,A) 

(5)  A  principal  should  make  it  possible  for  those  teachers 
who  have  been  in  school  longer  to  have  a  greater 
voice  in  school  affairs.  (D,SD) 

(6)  A  principal  should  be  willing  to  give  special 
privileges  to  teachers  who  are  very  effective 
in  the  classroom.  (D,SD) 

(7)  A  principal  should  not  expect  all  teachers  to 
take  part  in  supervising  hallways,  playgrounds, 
and  extra-curricular  activities.  (U,D,SD) 

The  coefficient  of  reproducibility  for  these  items  was  .87,  and  the  other 

criteria  for  the  existence  of  a  quasi  scale  were  also  fulfilled.  These 

included  a  satisfactory  range  of  marginal  frequencies  and  a  generally 

random  distribution  of  errors. 


Means -Ends  Dimension 

The  judges  who  were  asked  to  categorize  items  during  the  initial 
stages  of  the  study  showed  a  decided  lack  of  consensus  on  items  whose 
content  would  justify  placing  them  in  this  category.  This  should  have  been 
taken  as  a  sign  that  the  area  was  probably  not  scalable,  but  it  was  never¬ 
theless  retained.  By  a  process  of  trial  and  error  and  a  close  examination 
of  the  extent  to  which  judges  were  able  to  agree  on  content,  these  seven 
items  (m)  were  found  to  come  closest  to  being  scalable  of  all  the 


' 


'•  '  }■  ■ 

■ 

'  ’•  i  :  v‘ 

. 

■  . 


;  •'  •  •  ■■■  ■  i  . •: 


r  .-O  '- 


‘ 


...  ,  '  "  '  ’  .  '  i-  •  - 

1  .  -■ 

' 


' 


.  ■ 

■ 

■ 


combinations  which  were  tried: 


127 


(1)  A  principal  should  not  hesitate  to  depart  from 
official  procedures  if  it  means  that  certain  tasks 
will  he  carried  out  more  effectively.  (D,SD) 

(2)  A  principal  should  not  hesitate  to  go  against  the 
wishes  of  his  teachers  on  a  matter  of  school  policy 
if  he  considers  it  necessary.  (D,SD) 

(3)  A  principal  should  call  regular  staff  meetings  even 
though  he  may  not  have  anything  of  importance  to 
discuss  with  the  staff.  (SA,A,U) 

(4)  A  principal  should  not  feel  obligated  to  consult 
all  teachers  on  a  school  matter  if  he  can  gain 

a  general  impression  of  the  views  of  the  staff 
by  consulting  only  a  few  teachers.  (D,SD) 

(5)  A  principal  should  recognize  that  there  is  a 
greater  value  in  running  a  school  democratically 
than  in  doing  tasks  more  quickly  by  less  democratic 
means.  (SA,A) 

(6)  A  principal  need  not  involve  the  staff  in  formulating 
school  policies  in  order  to  run  a  school  effectively. 

(D,SD) 

(7)  A  principal  should  delay  action  until  there  is  staff 
consensus  before  proceeding  with  a  staff  project.  (SA,A) 

The  seal ogram  in  Figure  7  reveals  that  the  coefficient  of  reproducibility 
for  these  items  was  .87,  but  it  also  reveals  some  tendencies  toward  non¬ 
scale  types  which  suggests  that  more  than  one  variable  was  being  tapped 
by  these  questions.  In  order  to  have  some  indication  of  the  expectations 
of  teachers  in  this  area,  it  was  decided  to  retain  the  items  in  the  study 
as  forming  a  quasi  scale. 


Scoring 

After  the  scale  analysis  had  been  completed,  a  key  was  constructed 
according  to  the  way  in  which  response  categories  had  been  combined  for 
each  of  the  items.  Since  each  of  the  3even  items  in  each  quasi  scale  had 
been  dichotomized  this  meant  that  each  item  was  scored  as  either  one  or 
zero  and  the  possible  range  of  scale  scores  for  each  quasi  scale  was  from 
zero  to  seven.  This  simplified  scoring  procedure  was  the  only  one  that 


-■  i  ■  .  .  [ 

« 

' 

- 

:: 

.. 

.  . 

.  ''  '  9 

. 

' 

, 

,  . 

.  . 

' 


. 

’ 

s  J: 


' 

V  J:  j 


' 


128 


Mb ANS-F.NDS  ITEMS 
(Rep.  =  .b7) 


1 

2 

3 

4 

5 

6 

7 

8 

9 

10 

11 

12 

13 

14 

1 

4 

X 

X 

X 

X 

X 

X 

X 

2 

46 

X 

X 

X 

X 

X 

X 

X 

3 

54 

X 

X 

X 

X 

X 

X 

X 

4 

8 

X 

X 

X 

X 

X 

X 

X 

5 

13 

X 

X 

X 

X 

X 

X 

X 

6 

90 

X 

X 

X 

X 

X 

X 

X 

7 

19 

X 

X 

X 

X 

X 

X 

X 

8 

10 

X 

X 

X 

X 

X 

X 

X 

9 

39 

X 

X 

X 

X 

X 

X 

X 

10 

82 

X 

X 

X 

X 

X 

X 

X 

11 

87 

X 

X 

X 

X 

X 

X 

X 

12 

12 

X 

X 

X 

X 

X 

X 

X 

13 

14 

X 

X 

X 

X 

X 

X 

X 

14 

24 

X 

X 

X 

X 

X 

X 

X 

15 

29 

X 

X 

X 

X 

X 

X 

X 

16 

76 

X 

X 

X 

X 

X 

X 

X 

17 

66 

X 

X 

X 

X 

X 

X 

X 

18 

56 

X 

X 

X 

X 

X 

X 

X 

19 

60 

X 

X 

X 

X 

X 

X 

X 

20 

96 

X 

X 

X 

X 

X 

X 

X 

21 

.71 

X 

X 

X 

X 

X 

X 

X 

22 

70 

X 

X 

X 

X 

X 

X 

X 

23 

7 

X 

X 

X 

X 

X 

X 

X 

24 

75 

X 

X 

X 

X 

X 

X 

X 

25 

98 

X 

X 

X 

X 

X 

X 

X 

26 

34 

X 

X 

X 

X 

X 

X 

X 

27 

3 

X 

X 

X 

X 

X 

X 

X 

28 

89 

X 

X 

X 

X 

X 

X 

X 

29 

41 

X 

X 

X 

X 

X 

X 

X 

30 

94 

X 

X 

X 

X 

X 

X 

X 

31 

72 

X 

X 

X 

X 

X 

X 

X 

32 

100 

•X 

X 

X 

X 

X 

X 

X 

33 

1 

X 

X 

X 

X 

X 

X 

X 

34 

22 

X 

X 

X 

X 

X 

X 

X 

35 

91 

X 

X 

X 

X 

X 

X 

X 

36 

97 

X 

X 

X 

X 

X 

X 

X 

37 

95 

X 

X 

X 

X 

X 

X 

X 

38 

65 

X 

X 

X 

X 

X 

X 

X 

39 

38 

X 

X 

X 

X 

X 

X 

X 

40 

2 

X 

X 

X 

X 

X 

X 

X 

41 

50 

X 

X 

X 

X 

X 

X 

X 

42 

84 

X 

X 

X 

X 

X 

X 

X 

43 

6 

X 

X 

X 

X 

X 

X 

X 

44 

77 

X 

X 

X 

X 

X 

X 

X 

45 

79 

X 

X 

X 

X 

X 

X 

X 

46 

27 

X 

X 

X 

X 

X 

X 

X 

47 

9 

X 

X 

X 

X 

X 

X 

X 

48 

73 

X 

X 

X 

X 

X 

X 

X 

49 

28 

X 

X 

X 

X 

X 

X 

X 

50 

40 

X 

X 

X 

X 

X 

X 

X 

51 

58 

X 

X 

X 

X 

X 

X 

X 

52 

74 

X 

X 

X 

X 

X 

X 

X 

53 

i5 

X 

X 

X 

X 

X 

X 

X 

54 

61 

X 

X 

X 

X 

X 

X 

X 

55 

5 

X 

X 

X 

X 

X 

X 

X 

56 

78 

X 

X 

X 

X 

X 

X 

X 

57 

52 

X 

X 

X 

X 

X 

X 

X 

58 

44 

X 

X 

X 

X 

X 

X 

X 

59 

57 

X 

X 

X 

X 

X 

X 

X 

60 

45 

X 

X 

X 

X 

X 

X 

X 

61 

23 

X 

X 

X 

X 

X 

X 

X 

62 

20 

X 

X 

X 

X 

X 

X 

X 

63 

25 

X 

X 

X 

X 

X 

X 

X 

64 

64 

X 

X 

X 

X 

X 

X 

X 

65 

59 

X 

X 

X 

X 

X 

X 

X 

66 

33 

X 

X 

X 

X 

X 

X 

X 

67 

80 

X 

X 

X 

X 

X 

X 

X 

68 

11 

X 

X 

X 

X 

X 

X 

X 

69 

67 

X 

X 

X 

X 

X 

X 

X 

70 

35 

X 

X 

X 

X 

X 

X 

X 

71 

16 

X 

X 

X 

X 

X 

X 

X 

72 

32 

X 

X 

X 

X 

X 

X 

X 

73 

31 

X 

X 

X 

X 

X 

X 

X 

74 

1  7 

X 

X 

X 

X 

X 

X 

X 

75 

63 

X 

X 

X 

X 

X 

X 

X 

76 

47 

X 

X 

X 

X 

X 

X 

X 

77 

86 

X 

X 

X 

X 

X 

X 

X 

78 

18 

X 

X 

X 

X 

X 

X 

X 

79 

83 

X 

X 

X 

X 

X 

X 

X 

80 

48 

X 

X 

X 

X 

X 

X 

X 

81 

37 

X 

X 

X 

X 

X 

X 

X 

82 

99 

X 

X 

X 

X 

X 

X 

X 

83 

55 

X 

X 

X 

X 

X 

X 

X 

84 

30 

X 

X 

X 

X 

X 

X 

X 

85 

43 

X 

X 

X 

X 

X 

X 

X 

86 

21 

X 

X 

X 

X 

X 

X 

X 

87 

85 

X 

X 

X 

X 

X 

X 

X 

88 

68 

X 

X 

X 

X 

X 

X 

X 

89 

93 

X 

X 

X 

X 

X 

X 

X 

90 

88 

X 

X 

X 

X 

X 

X 

X 

91 

26 

X 

X 

X 

X 

X 

X 

X 

92 

81 

X 

X 

X 

X 

X 

X 

X 

93 

42 

X 

X 

X 

X 

X 

X 

X 

94 

53 

X 

X 

X 

X 

X 

X 

X 

95 

51 

X 

X 

X 

X 

X 

X 

X 

96 

36 

X 

X 

X 

X 

X 

X 

X 

97 

62 

X 

X 

X 

X 

X 

X 

X 

98 

69 

X 

X 

X 

X 

X 

X 

X 

99 

92 

X 

X 

X 

X 

X 

X 

X 

100 

49 

X 

X 

X 

X 

X 

X 

X 

f 

1  7 

27 

64 

73 

76 

79 

81 

83 

73 

36 

27 

24 

21 

19 

f-rrOrs 

5 

4 

12 

1 1 

12 

5 

V 

0 

2 

8 

7 

9 

8 

8 

FIGURE  7 

SCALOGRAM  FOR  MEANS -ENDS  ITEMS 


129 


could  be  adopted  in  view  of  the  fact  that  the  sets  of  items  did  not 
satisfy  all  of  the  criteria  to  permit  them  to  he  classified  as  scales. 
Table  IV  shows  the  distribution  of  scale  scores  for  each  of  the  seven 
quasi  scales. 


TABLE  IV 

FREQUENCY  DISTRIBUTION  OF  TEACHERS'  SCALE  SCORES  (N=765) 


Scale 

Score 

Sets  of 

Items 

Quasi  Scales 

Pi 

P2 

A1 

a2 

si 

S2 

M 

0 

28 

4 

14 

2 

73 

10 

1 

1 

120 

40 

52 

26 

155 

45 

16 

2 

171 

115 

105 

96 

185 

128 

53 

3 

175 

188 

175 

17? 

141 

177 

162 

4 

121 

.184 

185 

208 

117 

193 

221 

5 

81 

146 

136 

196 

63 

120 

210 

6 

45 

61 

80 

53 

23 

73 

86 

7 

14 

27 

18 

12 

8 

19 

16 

III.  SCALE  ANALYSIS  OF  PRINCIPALS'  RESPONSES 

The  same  items  as  were  included  in  each  of  the  teachers'  scales  were 
also  tested  for  scalability  for  the  sixty  principals  using  the  scalogram 
board  technique.  The  only  difference  in  the  analysis  resulted  from  the 
fact  that  the  marginal  frequencies  of  some  of  the  items  were  such  that  the 
items  were  not  in  the  same  order  as  they  had  been  for  the  teachers.  Further¬ 
more,  when  item  categories  were  combined,  the  "undecided"  responses  did  not 


' 

.  • r  l :  ;  ->  ' 


. 

■  : 

• 

: 

' 

'  • 

•  .  •  •.  ■■  ■  ‘  ,  ■  1  '■  ■" 1  ' 

. 


150 


always  fit  in  with  the  same  direction  as  they  had  in  the  first  analysis. 
The  purpose  of  this  analysis  was  not  to  establish  a  score  for  the  prin¬ 
cipals  but  only  to  determine  how  closely  the  data  approximated  a  scale 
for  them.  It  was  found  that  in  general  the  items  also  approximated  quasi 
scales  for  the  principals  as  well.  A  comparison  of  coefficients  of 
reproducibility  is  given  in  Table  V. 

TABLE  V 

COEFFICIENTS  OP  REPRODUC IBILITY  FOR  TEACHERS'  AND 
PRINCIPALS'  RESPONSES  TO  SEVEN  SETS  OP  ITEMS 


Sets 

of  Items 

Op  as 

i  Scales 

Group 

*1 

CM 

fU 

Ar 

S1 

S2  M 

Teachers 

.88 

.87  .85 

.90 

.88 

.85  .87 

Principals 

.88 

.90  .86 

.87 

.85 

.86  .88 

In  addition  to  having  very  similar  coefficients  of  reproducibility, 
the  items  also  resembled  the  teachers'  scalograms  with  respect  to  the 
other  criteria..  In  view  of  the  fact  that  there  were  only  sixty  principals 
in  the  sample  at  this  stage,  it  was  decided  to  score  their  responses  using 
the  same  key  as  had  been  constructed  for  the  teachers'  questionnaires. 

It  was  felt  that  this  was  more  justifiable  than  using  the  scale  analysis 
based  on  such  a.  relatively  small  sample  as  the  principals  group.  The 
distribution  of  scale  scores  for  the  principals  are  presented  in  Table  VI. 


' 


!.  .  . 


.  ..  . 


( 


. ; 


£ 


■  -  .  ■  ■ 


t 


! 


I 


! 


131 


TABLE  VI 

FREQUENCY  DISTRIBUTION  OF  PRINCIPALS'  SCALE  SCORES  (N=60) 


Scale 

Score 

Sets  of  Items 

Quad  Scales 

P1 

P2 

% 

A 2 

sl 

s2 

M 

0 

5 

0 

0 

0 

4 

i 

0 

1 

15 

4 

0 

2 

8 

6 

0 

2 

10 

4 

6 

6 

14 

7 

6 

3 

15 

7 

19 

20 

18 

13 

12 

4 

6 

17 

12 

14 

7 

18 

19 

5 

7 

14 

18 

13 

5 

9 

16 

6 

2 

8 

4 

4 

3 

3 

6 

7 

2 

6 

1 

1 

l 

3 

1 

Ambivalence  Scores 

Since  the  approximate  scalability  of  the  seven  sets  of  items  had 
been  demonstrated,  it  became  more  meaningful  to  determine  ambivalence 
scores  on  the  basis  of  the  intensity  function  for  each  of  these  areas. 

After  responding  to  each  content  item  on  the  questionnaire,  each 
principal  ha,d  indicated  how  difficult  it  had  been  for  him  to  respond  to 
that  item.  The  response  alternatives  provided  consisted  of  five  categories 
ranging  from  "Very  Difficult"  to  "Not  at  all  Difficult."  For  the  purpose 
of  the  scale  analysis  these  were  each  dichotomized  in  such  a  way  as  to 
provide  suitable  ranges  of  marginal  frequencies.  The  main  reason  for 
carrying  out  the  scale  analysis  was  to  determine  whether  these  items 


vt 


; 


- 


. 


' 


. 


:  '  ■ 


- j. 


;  .... 


'  -  >'  ’  ..  JO  ::  '  0  . 

/  .  . f: 


'  :  ;V  •  ..  :  '  : : 


•  • . 


0  . 


132 

would  form  a  quasi  scale  in  the  same  way  as  has  heen  found  in  other 
research.  In  general  the  results  of  previous  studies  were  borne  out 
and  a  low  coefficient  of  reproducibility  plus  a  random  pattern  of  error 
suggests  a  quasi  scale.  A  comparison  of  the  coefficients  of  repro¬ 
ducibility  for  the  ambivalence  scores  and  the  scale  scores  is  provided 
in  Table  VII. 


TABLE  VII 

COEFFICIENTS  OF  REPRODUCIBILITY  FOR  PRINCIPALS' 
CONTENT  AND  AMBIVALENCE  SCALES 


Quasi -Sc ales 

Analysis 

of 

Pi 

P2 

Al 

a2 

Si 

s2 

M 

Content 

.88 

•  90 

.86 

.87 

.85 

.86 

.88 

Ambi val enc  e 

.86 

.84 

00 

• 

.84 

.84 

.83 

.87 

In  each  case  the  coefficient  of  reproducibility  is  higher  for  the 
content  scale  than  it  is  for  the  ambivalence  scale.  Since  the  sampling 
distribution  of  the  statistic  is  not  known,  it  is  difficult  to  generalize 
about  this  difference,  but  it  does  suggest  that  the  content  scales  did 
contain  something  which  tended  to  make  principals  more  consistent  in  their 
responses  from  set  to  set  than  did  the  intensity  scales.  It  would  be 
encouraging  if  one  could  assume  that  this  difference  was  due  to  the 
common  content  of  the  sets  of  questions. 

The  ambivalence  questionnaire  was  scored  on  the  basis  of  the 
dichotomizations  which  had  been  established  by  the  scaling  process. 


' 

'  , 

'  '  '•  •  •  •  9.'  :  ■  \->r  -■  '  :  '  ■■ 

' 


'•  r  ,  ’ 

’  v 

....... 

■  '  ;  '  !•'  :  h  :  ''."I  ' 

r  t  '  1  '  '  ’.IB'  ’  1 


133 


Since  all  questions  had  "been  dichotomized  this  again  resulted  in  seven 
sets  of  scores  ranging  from  zero  to  seven.  The  distribution  of  these 
scores  by  various  quasi  scales  is  presented  in  Table  VIII. 


TABLE  VIII 

FREQUENCY  DISTRIBUTION  OF  PRINCIPALS’  AMBIVALENCE  SCORES  (N=60) 


Scale 

Score 

Quasi -Sc ales 

pl 

p2 

% 

A2 

S1 

s2 

M 

0 

4 

1 

2 

4 

3 

4 

6 

1 

5 

5 

5 

5 

5 

7 

9 

2 

9 

7 

10 

6 

7 

9 

6 

3 

8 

11 

10 

12 

12 

9 

9 

4 

8 

16 

11 

13 

11 

11 

8 

5 

13 

14 

14 

14 

14 

13 

11 

6 

7 

2 

7 

4 

7 

6 

3 

7 

6 

4 

1 

2 

1 

1 

8 

Through  the  scale  analysis  procedures  and  the  scoring  system  based 
on  this,  it  was  possible  to  quantify  the  expectations  which  teachers  held 
for  the  behavior  of  principals  and  also  the  expectations  which  the  principals 
held  for  themselves.  In  addition  the  procedure  proved  useful  for  arriving  at 
an  index  of  leader  ambivalence.  Since  this  study  was  not  concerned  with  the 
expectations  per  se,  these  will  not  be  discussed  any  further  than  has  been 
done  in  this  chapter.  The  quantifications  of  expectations  were  used  for 
deriving  other  necessary  indices;  this  process  is  outlined  in  the  next  chapter. 


. 

, 


I 


.  .  -  .  .  : 


' 


' 


■ 

■' 

r  ■  *  ■  '■ 

. 


CHAPTER  VIII 


TREATMENT  OF  DATA 

This  chapter  further  outlines  the  treatment  to  which  the  data  were 
subjected.  It  includes  distributions  of  the  raw  and  converted  scores  for 
some  of  the  variables  as  well  as  details  of  the  manner  in  which  indices 
for  consensus  and  teacher- principal  agreement  on  expectations  were  derived. 
As  a  forward  look  to  the  following  chapter  in  which  the  testing  of  the 
hypotheses  is  discussed,  a  section  of  this  chapter  is  devoted  to  a 
consideration  of  the  various  statistical  techniques  which  were  employed. 

I.  LEADER  BEHAVIOR  SCORES 

In  order  to  obtain  measures  of  the  leader  behavior  of  each  of  the 
principals,  seven  Leader  Behavior  Description  Questionnaires  were  selected 
at  random  from  among  those  which  had  been  completed  by  the  teachers  in 
each  school.  The  key  which  is  provided  scores  each  item  on  a  scale 
ranging  from  zero  to  four.  Since  there  are  fifteen  items  on  each  of  the 
two  dimensions,  the  possible  range  of  scores  is  from  zero  to  sixty  on  each 
of  these  dimensions.  The  arithmetic  mean  of  the  seven  descriptions  was 
computed  in  order  to  arrive  at  a  score  for  each  principal.  Although  there 
is  considerable  variance  in  the  descriptions  of  one  leader  by  the  several 
group  members,  studies  have  shown  that  the  means  computed  for  each  leader 
do  have  some  meaning  (4,  p.  47 )« 

The  distribution  of  the  raw  scores  for  each  of  the  two  dimensions 
is  presented  in  Table  IX.  The  means  and  standard  deviations  do  not  differ 
to  any  extent  from,  the  means  and  standard  deviations  obtained  in  other 


-  ■ 

' 

. 

- 

,  •  ■ 

'  •  , 

'•  t  .  '  '  ' 

.  - 

- 

«  .  ■  o; 

, 


135 


studies.  An  examination  of  the  distributions  reveals  that  the  Consideration 
scores  show  a  negative  skew  while  the  Structure  scores  are  definitely 
leptokurtic.  Since  it  was  essential  to  use  statistical  techniques  which 
assume  normality  of  distribution,  it  was  decided  to  convert  these  scores 
to  T-scores  (3 »P»  314)*  The  converted  scores  are  presented  in  the  same 
table  and  show  means  and  standard  deviations  close  to  those  which  would  be 


TABLE  IX 

FREQUENCY  DISTRIBUTION  OF  RAW  AND  T-SCORES  FOR  CONSIDERATION 
AND  INITIATING  STRUCTURE  LBDQ,  SCORES  (N=56) 


Interval 

Raw  Scores 

T-Scores 

Consideration  Structure 

Consideration  Structure 

72-74 

1 

69-71 

2 

0 

66-6  8 

2 

1 

63-65 

2 

4 

60-62 

4 

4 

57-59 

3 

4 

54-56 

1 

8 

6 

51-53 

6 

0 

8 

7 

48-50 

7 

1 

6 

6 

45-47 

16 

8 

3 

9 

42-44 

9 

11 

6 

2 

39-41 

8 

15 

5 

4 

36-38 

4 

10 

3 

4 

33-35 

2 

6 

2 

3 

30-32 

2 

4 

0 

0 

27-29 

2 

2 

0 

24-26 

1 

Mean 

S.D. 

43.4 

6.0 

40.1 

4.9 

50.0 

9.9 

49-9 

9.7 

expected  to  result  from  the  normalization  process.  The  investigator  felt 
that  normalizing  the  distribution  could  be  justified  on  the  grounds  that 


.  ... 

. 


.  .. 


‘ 


; 

'  -  • 


- 

5 

. 


„  V 


the  leader  behavior  characteristics  were  either  normally  distributed 
or  close  to  normally  distributed  in  the  population  from  which  the 


136 


principals  in  the  study  were  selected. 

A  quadrant  approach  is  often  used  to  depict  the  leader  behavior 
styles  of  a  group  of  individuals.  The  sample  is  divided  into  two  groups 
using  each  dimension  by  classifying  those  above  the  mean  or  median  as  being 
high  in  that  particular  dimension  and  those  which  fall  below  the  mean  or 
the  median  as  being  low.  In  this  analysis  the  mean  was  used  as  the  point 
of  separation  and  the  sample  was  divided  into  four  groups  on  this  basis: 
those  who  were  high  in  both,  those  who  were  low  in  both,  and  two  groups 
consisting  of  those  who  were  high  on  one  dimension  but  low  on  the  other 
one.  The  distribution  of  the  principals  in  the  four  quadrants  both  on 
the  basis  of  the  raw  and  of  the  T-scores  is  given  in  Figure  8. 


C+S- 


C+S+ 


17* 


18* 


(13) 


(16) 


12* 


9* 


(16) 


(11) 


C-S- 


C-S+ 

*original  scores 
(T-Scores) 


FIGURE  8 


DISTRIBUTION  OF  PRINCIPALS  IN  FOUR  QUADRANTS 
DERIVED  FROM  LBDQ  DIMENSIONS  (N=56) 


The  existence  of  the  negative  skew  in  the  Consideration  scores  is 


evidenced  by  the  distribution  of  the  principals  in  the  four  quadrants 


■ 

, 

, ' 

' 

■ 

-  - 

, 

' 

■ 


~ . . . . 

• 

1 

4 

137 


When  the  distributions  were  normalized  the  extreme  variation  is  reduced, 
but  there  is  still  evidence  that  there  is  some  correlation  between  scores 
on  the  two  dimensions  when  the  quadrants  are  considered  as  comprising  a 
fourfold  table. 


II.  AMBIVALENCE,  CONSENSUS,  AND  AGREEMENT 

Ambival enc e  Scores 

The  method  by  which  ambivalence  scores  were  derived  for  the 
principals  on  each  of  the  seven  content  areas  has  already  been  outlined 
in  the  previous  chapter.  The  results  of  this  portion  of  the  analysis  were 
presented  in  Table  VIII  on  page  133 •  At  first  it  was  anticipated  that 
each  of  the  content  areas  might  be  used  separately  when  testing  hypothe¬ 
ses  concerning  ambivalence,  and  for  this  reason  the  scale  scores  on 
ambivalence  for  each  content  area  were  converted  to  normalized  scores 
according  to  the  method  given  by  Garrett  for  sigma  scaling  (3,  pp.  323- 
327).  An  overall  ambivalence  score  was  then  obtained  for  each  principal 
by  summing  the  separate  scores  for  the  seven  content  areas.  As  the 
analysis  developed  it  became  apparent  that  the  ambivalence  scores  for 
separate  content  areas  would  not  be  used;  however,  it  is  doubtful  whether 
normalizing  scores  for  separate  areas  would  have  any  harmful  effect. 

Due  to  the  conversion  to  normalized  scores  the  sum  of  the  seven 
separate  ambivalence  measures  resulted  in  numerically  large  scores.  The 
distribution  of  these  composite  ambivalence  scores  is  given  in  Table  X. 

For  purposes  of  later  analysis,  it  was  decided  to  convert  these  scores 
to  T-scores  as  well,  and  these  are  shown  in  the  right  half  of  Table  X. 


The  conversion  of  all  measures  and  indices  to  normalized  standard  scores 


, 

» 

’ 

.... 

. 

>  V 

„ 

o  '  '  ®  ■  a  v  • r  m 

, 

. 

■  ; 


138 

has  the  great  advantage  of  making  all  of  these  measures  for  a  particular 


school  comparable. 


TABLE  X 

FREQUENCY  DISTRIBUTION  OF  RAW  AND  T-SC0RE3  FOR  PRINCIPALS » 

AMBIVALENCE  SCORES 


Interval 

Raw  Scores 

Interval 

T-Score 

f 

f 

285  -  299 

2 

69  -  71 

2 

270  -  284 

5 

66  -  68 

1 

255  -  269 

6 

63  -  65 

4 

240  -  254 

2 

60  -  62 

3 

225  -  239 

6 

57  -  59 

5 

210  -  224 

10 

54  -  56 

5 

195  -  209 

4 

51  -  53 

6 

180  -  194 

5 

48  -  50 

6 

165  -  179 

4 

45  -  47 

8 

150  -  I64 

3 

42  -  44 

5 

135  -  149 

4 

39  -  41 

5 

120  -  I34 

1 

36  -  38 

2 

105  -  119 

1 

33  -  35 

1 

90  -  104 

2 

30  -  32 

2 

75  -  89 

1 

27  -  29 

0 

24  -  26 

1 

Mean 

206.8 

49.9 

S.D. 

52.1 

9.9 

Consensus 

In  order  to  test  the  hypotheses  involving  the  degree  of  consensus 
which  existed  among  the  teachers  in  a  particular  school  it  was  necessary 
to  have  a  measure  of  consensus.  A  measure  which  has  been  used  previously 
is  that  of  the  variance  of  a  set  of  scores.  So  far  in  the  analysis  of  data 
there  had  been  derived  measures  of  the  expectations  of  each  of  the  teachers 
in  each  of  the  schools  on  seven  areas  of  principal  leader  behavior.  For 

each  of  these  seven  areas  of  expectations  the  variance  of  the  scores  for 


■  '  :  :  '  -I  '■  ■  ■  -  'r 


i  v:.: ' 


•  ■  • 

.  ■ 

. 

•X 

:  •  : 

i  •• 

--  < 

•:  ... 

' ' 

“ 

:  - 

. 

.  , 


, 


- 

' 

■  : k-.: 

159 


the  teachers  in  the  school  was  calculated.  This  was  carried  out  using 
the  formula: 


1 


(5,  p.  89) 


which  yields  an  unbiased  estimate  of  the  population  variance.  That  is, 
using  this  formula  rules  out  the  effects  of  unequal  Ns  in  the  different 
schools. 


However,  what  was  desired  was  not  a  number  of  measures  of  variance, 
but  one  index  of  the  amount  of  consensus  which  existed  in  a  particular 
school.  This  was  obtained  by  summing  the  variances  for  the  seven  content 
areas  for  each  of  the  schools.  The  index  thus  obtained  ranged  from  9*01 
to  25.55  with  a  definite  positive  skew.  These  numbers  were  also  inversely 
related  to  consensus  in  that  a  high  degree  of  consensus  was  indicated  by 
a  low  numerical  value  of  the  index.  These  scores  were  also  normalized 
and  the  direction  reversed  so  that  consensus  would  vary  directly  with 
the  numerical  size  of  the  index  and  would  thus  make  the  interpretation 
of  various  tests  less  confusing.  Both  the  raw  and  the  T-scores  for  this 
variable  are  given  in  Table  XI. 


Teacher- Pr i nc i pal  Agreement 

It  was  also  necessary  to  have  some  index  of  the  extent  to  which 
there  was  agreement  between  the  principal  in  a  school  and  his  teachers  on 
the  role  of  a  principal.  What  was  required  was  a  means  of  comparing  the 
profile  of  the  principal's  scores  on  the  seven  areas  of  expectations  with 
the  scores  of  the  teachers  on  his  staff.  In  order  to  simplify  calculations 
the  mean  score  of  the  teachers  on  each  dimension  was  determined,  and  the 


,  '  '  ' .  •  •  .  V-  '  •  ;  ...  :• 

4 

■ 

, 

.... 

.  E  ■ 

' 

. 

.  '  '  ■  .'  : 


•  • 


■ 

.  . 

! 

•  ’  , 


.  ■  ■  •'  •  : 

' 


TABLE  XI 


140 


FREQUENCY  DISTRIBUTION  OF  RAW  AND  NORMALIZED  CONSENSUS  SCORES 


Raw  Scores 

Normalized 

Scores 

Interval  f 

Interval 

f 

9.0 

-  9.9 

2 

72 

-  74 

1 

10.0 

-  10.9 

2 

69 

-  71 

1 

11.0 

-  11.9 

4 

66 

-  68 

1 

12.0 

-  12.9 

9 

63 

-  65 

3 

13.0 

-  13.9 

10 

60 

-  62 

2 

14.0 

-  14.9 

8 

57 

-  59 

4 

15.0 

-  15.9 

7 

54 

-  56 

9 

16.0 

-  16.9 

6 

51 

-  53 

6 

17.0 

-  17.9 

6 

48 

-  50 

8 

18.0 

-  18.9 

0 

45 

-  47 

4 

19.0 

-  19.9 

0 

42 

-  44 

4 

20.0 

-  20.9 

0 

39 

-  41 

5 

21.0 

-  21.9 

0 

36 

-  38 

3 

22.0 

-  22.9 

1 

33 

-  35 

3 

23.0 

-  23.9 

1 

30 

-  32 

1 

27 

-  29 

0 

24 

-  26 

1 

Mean 

14.4 

49.6 

S.D. 

2.7 

10.0 

principal ’s 

profile  was 

compared  with  the 

profile 

of  the 

mean  teache: 

t 


scores.  This  comparison  was  made  using  the  Cronbach  and  Gleser  (2) 


formula 

Bl22  ■  k  <  Ui  -  *32  >2  (2’  p-  459) 

That  is,  the  difference  between  the  principal’s  score  and  the  mean 
teachers’  score  was  found  for  each  dimension,  then  the  difference  was 


squared  and  summed  over  all  dimensions.  This  process  yielded  an  index 
of  dissimilarity  since  the  larger  the  index  the  greater  was  the  dis¬ 
parity  between  the  principal's  self- expectations  and  the  expectations 


of  his  staff  members. 


■■  '  ’ 


1 


. 


. 


141 


The  distribution  of  these  scores  for  the  fifty-six  schools  is 
shown  in  Table  XII.  The  distribution  is  bimodal,  but  does  have  a  fairly 
acceptable  range.  It  was  decided  to  normalize  this  distribution  and  to 
reverse  it  so  that  the  higher  score  represented  the  higher  degree  of 
similarity  between  a  principal  and  the  teachers  in  his  school. 


TABLE  XII 

FREQUENCY  DISTRIBUTION  OF  RAW  AND  NORMALIZED  TEACHER- 
PRINCIPAL  AGREEMENT  SCORES 


Raw  Scores 

Normalized 

Scores 

Interval  f 

Interval 

f 

0.0  -  2.9 

1 

72 

- 

74 

1 

3.0  -  5.9 

2 

69 

- 

71 

1 

6.0  -  8.9 

6 

66 

- 

68 

1 

9.0  -  11.9 

10 

63 

- 

65 

3 

12.0  -  I4.9 

8 

60 

- 

62 

3 

15.0  -  17.9 

6 

57 

- 

59 

6 

18.0  -  20.9 

4 

54 

- 

56 

4 

21.0  -  23.9 

10 

51 

- 

53 

8 

24.0  -  26.9 

3 

48 

- 

50 

6 

27.0  -  29.9 

1 

45 

- 

47 

8 

30.0  -  32.9 

3 

42 

- 

44 

2 

33.0  -  35.9 

0 

39 

- 

41 

6 

36.0  -  38.9 

2 

36 

- 

38 

3 

33 

- 

35 

2 

30 

- 

32 

1 

27 

- 

29 

0 

24 

- 

26 

1 

Mean 

16.8 

49.9 

S.D.  8.2  9.9 


The  normalized  scores  are  also  presented  in  Table  XII  along  with  the  raw 


scores. 


One  objection  that  might  be  raised  in  regard  to  this  index  is  that 


■■  .  ;i:J 

, 


-  •  5 


•  -  - 


- 


% 


■ 


j  »  .  •  i  i  ■  i  i 


' 


■ 


142 

by  using  the  mean  of  the  teachers"  scores  as  a  profile  for  the  whole 
staff,  one  may  have  produced  a  profile  which  no  teacher  closely 
approximated.  In  response  to  this  one  might  counter  that  in  the  face  of 
conflicting  expectations,  perhaps  the  best  course  for  a  principal  to 
steer  would  be  a  middle  course  which  gives  him  some  freedom  to  move 
in  either  direction  either  in  following  his  self- expectations  or  in  the 
process  of  modifying  the  expectations  which  his  teachers  held  for  him. 

It  would  seem  that  on  the  basis  of  such  logical  grounds,  one  can  defend 
the  use  of  the  procedure. 

III.  STATISTICAL  TECHNIQUES 

The  first  two  sections  of  this  chapter  and  the  statement  of  the 
hypotheses  to  be  tested  have  given  some  indication  of  the  procedures 
which  were  used  in  testing  the  hypotheses.  The  various  aspects  of  the 
treatment  of  the  data  to  this  point  served  to  prepare  the  data  so  that 
it  would  be  possible  to  test  these  hypotheses.  In  this  process  measures 
of  the  degree  of  relationship  between  variables  were  obtained  by  calcu¬ 
lating  the  Pearson  product  moment  correlation  coefficient  (r)  and  the 
correlation  ratio  (  77  ) .  Tests  of  the  significance  of  r,  and 
curvi linearity  as  well  as  tests  of  the  significance  between  rs  were  applied. 
In  order  to  determine  the  significance  of  differences  between  means  both 
analysis  of  variance  and  the  t  technique  were  utilized.  The  main  problem 
for  the  investigator  who  wishes  to  use  any  of  these  techniques  is  to 
recognize  the  assumptions  underlying  each  of  them  and  the  effects  which 
violations  of  the  assumptions  might  have  on  the  results. 


One  of  the  assiimptions  underlying  most  of  these  is  that  a  normally 


1  '  "  : .  ‘ 

,  ' 


rt  .  '  \  ■ :  i£ '  ■  '  jIC' oe  a 

,  c  ’•  ■ 

, 

1: 

' 

„ 


143 


distributed,  continuous  variable  or  variables  is  involved.  The  investi¬ 
gator  considered  that  normalizing  each  of  the  distributions  of  variables 
was  justified  on  the  assumption  that  these  variables  were  normally 
distributed  in  the  population.  Some  of  the  other  assumptions  specific 
to  each  of  the  techniques  are  outlined  below. 

Differences  Between  Means 

The  jt  technique.  The  _t  technique  is  used  for  determining  the 
significance  of  differences  between  group  means  when  only  small  samples 
are  involved.  This  technique  is  necessary  because  even  though  the  sampling 
distribution  of  means  is  normal,  the  departure  from  normality  increases  as 
the  sample  size  decreases;  that  is,  the  shape  of  the  distribution  varies 
with  the  number  of  degrees  of  freedom  (5,  pp.  98-99)*  The  application 
of  the  t  distribution  to  a  test  of  the  significance  of  the  ratio 
(M1  -  M2)/  Sdj,  assumes  that  the  scores  of  the  population  from  which  each 
mean  was  calculated  are  normally  distributed  and  also  that  the  variances 
are  equal  (5,  p.  105).  McNemar  points  out  that  the  techniques  commonly 
used  for  testing  each  of  these  assumptions  are  insensitive  to  these 
differences  in  small  samples  unless  the  differences  are  marked  (5,  p.  105). 
He  does  cite  evidence  from  an  empirical  study  carried  out  by  Boneau  (l) 
which  suggests  that  even  sizable  differences  in  variances  or  in  skewness 
may  not  have  serious  effects  on  the  technique  if  equal  Ns  and  two-tailed 
tests  are  used  (5>  p.  IO6-IO7). 

Analysis  of  Variance.  This  technique  is  generally  used  for  testing 


. 

■ 


■  •  ■  noo  I 


'• 


•• 

, 

•  ' 

■ 

- 

; 

'  ' 

.  ..... .  . 


144 


the  significance  of  differences  between  the  means  of  several  groups; 
it  can  be  applied  in  a  variety  of  situations  regardless  of  the  number 
of  groups  or  of  the  number  of  scores  in  each  group.  The  procedure 
involves  breaking  down  the  total  sums  of  squares  into  two  additive  parts: 
that  for  between  groups  and  that  for  within  groups.  Each  of  these  is 
divided  by  the  appropriate  degrees  of  freedom  to  yield  an  estimate  of  the 
population  variance.  The  parameter  estimate  based  on  the  between  groups 
sums  of  squares  is  divided  by  that  based  on  the  within  groups  sums  of 
squares;  the  resulting  ratio  is  tested  for  significance  against  the  F 
distribution.  If  the  two  estimates  of  the  population  variance  differ 
by  too  great  an  extent  from  that  which  would  be  expected  on  the  basis 
of  chance  differences,  it  is  considered  to  be  an  indication  that  there 
exist  some  significant  differences  between  group  means. 

Two  basic  assumptions  underlying  analysis  of  variance  are  (l) 
that  the  distribution  of  the  population  of  scores  is  normal,  and  (2) 
that  there  is  homogeneity  of  variance  from  group  to  group  (5,  pp.  264- 
265).  The  first  of  these  is  again  relatively  difficult  to  determine  for 
relatively  small  samples,  but  the  latter  can  be  evaluated  by  the  use  of 
Bartlett's  test.  Concerning  these  assumptions,  McNemar  states: 

Although  these  assumptions  are  incorporated  in  the 
mathematical  derivation  of  the  F  distribution,  there  is 
ample  evidence  that  marked  skewness,  departures  from  normal 
kurtosis,  and  extreme  differences  in  variance  (of  the  order 
1  to  4  to  9-  it  is  not  the  numerical  differences  but  the 
relative  sizes  of  the  variances  that  are  pertinent)  do  not 
greatly  disrupt  the  P  test  as  a  basis  for  judging  signifi¬ 
cance  in  the  analysis  of  variance.  (5?  p.  252) 

There  is  a  further  caution  to  be  observed  in  using  tests  following  an 

analysis  of  variance  to  determine  whether  selected  means  differ  significantly. 


- 

' 

, 

' 

...  .  .  '  ’  ’ 

•  '■/ ;  "  '  '  ;  h : 

' 

. 

■  ■'  :  ■'  I--'  ::  ;  :■  :>  -  ■  ■  1;  -  • 

,  '  ' 

' 

, 


' 


145 


McNemar  is  of  the  opinion  that  the  t_  test  is  appropriate  only  if  an 
a  priori  hypothesis  calls  for  making  the  comparison  (5,  p.  285). 

Measures  of  Relationship 

Pearson  r.  The  product  moment  correlation  coefficient  indicates 
the  amount  of  co-relation  between  two  variables  (5,  p.  109).  Computation 
can  be  carried  out  from  a  diagram  which  contains  the  bivariate  scatter  of 
the  scores  or  with  the  aid  of  a  calculator  on  the  scores  themselves. 

Some  of  the  basic  assumptions  are  linearity  of  regression,  homos cg  das ticity 
within  arrays,  and  normality  of  arrays;  these  conditions  are  more  likely 
to  obtain  if  the  marginal  distributions  are  not  badly  skewed  (5,  p.  134)* 

It  should  also  be  noted  that  correlation  coefficients  are  reduced  when 
the  range  of  talent  in  one  or  both  of  the  variables  is  restricted.  On 
the  other  hand  spuriously  high  coefficients  may  arise  from  undue  hete¬ 
rogeneity  in  the  group,  using  indexes  which  contain  a  common  denominator, 
correlating  averages,  and  from  part-whole  correlations  (3,  p.  441-442; 

5,  p.  162-164).  The  possible  effect  of  these  must  be  kept  in  mind  in 
interpreting  the  correlation  coefficient  which  is  obtained. 

The  significance  of  an  r  or  of  the  difference  between  rs  is  most 
readily  determined  by  transforming  the  coefficients  to  Fisher's  £  function 
(5,  pp.  139-140).  The  standard  error  formulas  in  each  case  make  the 
calculations  straight-forward. 

Correlation  Ratio .  The  correlation  ratio  (  )  gives  an  indication 

of  the  accuracy  with  which  predictions  of  one  variable  from  another  can  be 
made  on  the  basis  of  array  means.  It  is  particularly  appropriate  in  cases 


■ 

, 

, 

.  '  ' 

>  :  ;  - 

-  • 

<»  •  -  ? 


. 


where  regression  is  not  linear;  however,  it  has  one  disadvantage  in 
that  it  indicates  the  degree  of  relationship  but  not  the  form  of  the 


146 


relationship.  This  must  be  determined  from  a.  scatter  diagram 
(5>  p.  281 ).  The  calculations  involved  are  similar  to  those  used  in 
analysis  of  variance;  Mcllemar  has  outlined  in  detail  the  computational 
procedures  and  formulas  for  testing  the  significance  of  the  correlation 
ra„tio,  the  significance  of  linear  correlation,  and  testing  for  signifi¬ 
cant  departure  from  linearity  (5,  pp.  270-281).  For  eanh  of  these  an 
appropriate  F  ratio  is  determined  and  evaluated  on  the  basis  of  the 
distribution  of  values  of  F  which  might  be  obtained  by  chance.  In 
carrying  out  the  computations  in  the  following  chapter,  the  procedures 
outlined  by  Mcllemar  were  followed  closely. 


L  . 


r  '  ;• 


J  <:  .  m: 


o 


? 


0  - 


i'.'o  " 


X 


. .  c-  ■  o 


x 


L 


5 


i-4, 

u 


ri-  f 


c 


' 


REFERENCES  FOR  CHAPTER  VIII 


Boneau,  C.A.  ’’The  Effects  of  Violations  of  Assumptions 
Underlying  the  t  Test,"  Psychological  Bulletin,  57 : 

49-64)  January,  i960. 

Cronbach,  Lee  J.,  and  Goldine  C.  Gleser. "Assessing  Similarity 
Between  Profiles,"  Psychological  Bulletin,  50:456-473) 
November,  1953  • 

Garrett,  Henry  E.  Statistics  in  Psychology  and  Education. 

New  York:  Longmans,  Green  ancFCo.,  1938* 

Halpin,  Andrew  W.  The  Leadership  Behavior  of  School 
Superintendents .  ~ Second  Edition!  Chicago:  Midwest 
Administration  Center,  1959* 

McNemar,  Quinn.  Psychological  Statistics .  Third  Edition. 

John  Wiley  and  Sons,  Inc.,  1962'. 


:  -  . 


•  .  ■  -  -  - 

.  ?  • :  ■  ,  ■  •  -  :  ’  • 

'  o  .  .  .  -  ' 

. 

„  ■  . 

'  .  1 
,  7„  ■  ■  ■  ;  '  ‘ 

■ . •'  .  , 

:  . 

■  '  '  - 
,  -  - 


CHAPTER  IX 


TESTING  THE  HYPOTHESES 

In  this  chapter  the  relationships  between  ambivalence,  the  two 
dimensions  of  leader  behavior,  consensus  on  expectations,  and  teacher- 
principal  agreement  on  expectations  are  examined  in  detail.  The 
theory  which  was  outlined  to  guide  the  study  suggested  specific 
relationships  between  these  variables  which  would  enable  prediction 
of  such  variables  as  consensus  and  agreement  from  ambivalence  and  the 
two  dimensions  of  lea,der  behavior.  The  relationships  to  be  tested 
have  been  grouped  for  ease  in  analysis  and  in  the  interpretation  of 
results. 

I.  AMBIVALENCE  AND  LEADER  BEHAVIOR 

Testing  Hypothesis  One 

The  data  which  have  been  presented  revealed  that  principals 
differed  both  in  the  way  that  their  teachers  described  them  on  the 
Leader  Behavior  Description  Questionnaire  and  in  Ambivalence  as 
determined  by  means  of  the  self-report  instrument.  It  had  been 
hypothesized  that  Ambivalence  would  be  related  to  leader  behavior; 
the  specific  hypotheses  which  were  tested  are  as  follows: 

1.1  There  will  be  a  positive  relationship  between 
Ambivalence  and  the  Consideration  dimension. 

1.2  There  will  be  a  negative  relationship  between 
Ambivalence  and  the  Initiating  Structure 
dimension. 


v  ,  :  .  ■  .V  '  '  , 

•>  -  ■  _';c 


r  •  ■  •  •  7' 


:  '  :  j; 


>"  ■■  ■ 

•  '  /  •.  '• 

•  ; 

'  •  J  :  ■  '  .  i; 

'  .  .  •  •  '  ;  '  c .•  ■  ; 


•  >  '  .  .  7 


■ 


149 


In  order  to  test  these  hypotheses,  principals  were  classified 
into  two  groups  on  the  basis  of  their  Ambivalence  scores.  Those 
principals  who  scored  above  the  mean  were  considered  to  be  high  in 
Ambivalence  while  those  who  scored  below  the  mean  were  considered  to 
be  low  in  Ambivalence.  The  mean  Structure  and  Consideration  scores 
were  calculated  and  the  differences  between  the  groups  were  tested 
for  significance  by  using  the  t  technique.  The  data  necessary  for 
these  tests  are  presented  in  Table  XIII. 

TABLE  XIII 

MEM  STRUCTURE  MD  CONSIDERATION  SCORES,  VARIANCES,  AND 
SIGNIFICANCE  OF  DIFFERENCES  FOR  PRINCIPALS 
OF  HIGH  AND  LOW  AMBIVALENCE 


Leader 

Behavior 

Dimension 

Ambivalence 
High  Low 

N=28  N=28 

Diff. 

Between  SE^ 
Means 

* 

t 

M 

48.64 

51.52 

Structure 

2 . 68 

2.65 

1.02 

<L 

s 

125.95 

70.22 

M 

49-71 

50.11 

Consideration 

0.40 

2.59 

0.15 

(L 

s 

108.42 

79.81 

A-  v 

t  >  2.01  required  for  significance  at  .05  level  (two-tailed  test) 

It  is  evident  from  the  table  that  the  difference  in  mean 
Consideration  scores  was  almost  zero  and  was,  of  course,  highly 
insignificant .  The  difference  between  mean  Structure  scores  was  in 
the  predicted  direction  but  fell  short  of  being  statistically 
significant  a.t  the  .05  level  even  when  a,  one-tailed  test  was  used. 


_ _ 


2c  2 

’  i 

i  2 

'i  ::2c  i,1 

•  c  „  2 

2 

2 

2  i  ? 

:  OJ  J 

■  2 

c  2 

<: 

L 

2 

Y  .  r  C 

2  c 

2  n 

,  r..». 

.  u 

2 

'  Of 

'2  c  y2 

■  2  :  •  • 

o  .  '  2d  ■ :  ;  ■  •  .  2  •;  .2  -  ■.  .2 ’ 

'  •'  2  2 


-  T 


•'  1  ». 


:  . .  ;  ■  ■:  V  c.  ;  .  - ; 

2  c  2  c  -:2  c  c  ;  222  2. 

:2c.  .  .  - :  2  -.2  v  •:  2 


v  •.  '  •  2  - ;  2 

!  .  I  i:i :  ‘.222.  -’i  .  <  ;  i  ..  : 


<  '  •  . 


c  re 


c .  :  •  : 

Icfad 

'  :  2  c  r  •  c  \ 


C 


.  ' 


. '  .  v 


2c2  -  •:  '2 


<  i 


,  -• 


/  ; .  • 


150 


An  interesting  feature  of  Table  XIII  is  the  large  difference  in  the 
variance  of  both  Structure  and  Consideration  scores  for  the  Highs  in 
Ambivalence  as  compared  to  the  Lows.  These  differences  are  not 
significant  when  compared  using  the  P  test,  but  they  are  striking 
when  the  small  sample  involved  is  considered.  Although  these 
differences  were  not  anticipated  in  the  theory,  it  is  interesting 
to  note  that  the  Lows  in  Ambivalence  were  much  more  alike  in  both 
Structure  and  Consideration  than  we re  the  Highs . 

The  fact  that  there  were  no  significant  differences  between 
either  the  Structure  or  the  Consideration  scores  when  principals 
were  classified  on  the  basis  of  Ambivalence  was  not  entirely  unex¬ 
pected  in  view  of  the  third  hypothesis  in  this  set  which  stated  in 
effect  that: 

1.5  The  relationship  between  Ambivalence  and  Consideration 
and  between  Ambivalence  and  Structure  will  be  curvi¬ 
linear. 

In  order  to  test  this  hypothesis  it  was  necessary  to  calculate  both 
the  Pearson  pro duct -moment  correlation  coefficient  and  the  correla¬ 
tion  ratio  for  each  of  the  relationships.  In  order  to  make  it 
possible  to  carry  out  tests  of  significance ,  it  was  also  necessary 
to  calculate  each  of  these  from  the  same  scattergram.  The  results 
of  the  analysis  are  presented  in  Table  XIV  which  shows  that  there 
were  very  low  negative  correlations  between  Ambivalence  and 
Structure  and  between  Ambivalence  and  Consideration.  This  is  some¬ 
what  sxirprising  in  view  of  the  difference  between  Structure  scores 
for  the  Highs  and  Lows  in  Ambivalence  and  must  be  attributed  to  the 


;  ....... 


Ex.  •  ' 

;  ■  •;  .  v-  .v  J'o  T 

■  .V  •  .  ■- 


.  r. 


:  ( 


.  • 

>'  '•  •'  ■  ■  '■:■■■’  •  ■'  J  '  '  ’)  1.1 

-  •  ' '  •  :  .V  '  •  .  •  •  •  •  -  . 

•  c>: 

• '  •  ■  ■  ,  ! 

J‘: :  ■  •.  •  '  1  •: if  7  7  • 

7  ;  '  :>i  r £ 

.  •'  :  '  .  . 

’  7  .7. 


i  Vj 


151 


fact  that  there  was  a  very  high  variance  discrepancy  between  the  two 
groups  which  has  affected  the  product-moment  correlation  coefficient. 

TABLE  X IV 

LINEAR  AND  NONLINEAR  CORRELATION  COEFFICIENTS  FOR 
THE  RELATIONSHIP  BETWEEN  AMBIVALENCE  AND  TOO 
DIMENSIONS  OF  LEADER  BEHAVIOR  (N=56) 


Ambivalence 

and 

rAy 

H  Ay 

Structure 

-0.07 

0.45 

Consideration 

-0.06 

0.30 

As  had  been  expected,  the  correlation  ratios  (etas)  were 
numerically  larger  than  the  product -moment  correlation  coefficients. 
In  order  to  test  the  significance  of  the  correlation  ratios  as  well 
as  of  the  linear  correlations  and  the  departure  from  linearity,  a 
number  of  F  ratios  were  calculated  and  are  presented  in  Table  XV. 

An  examination  of  this  table  reveals  that  the  F  ratios  are  far  below 
those  required  for  significance  of  either  the  correlation  ratios,  the 
linear  correlations,  or  curvilinearity  regardless  of  the  variables 
concerned.  This  suggests  that  there  is  neither  a  linear  nor  a  curvi¬ 
linear  relationship  between  Ambivalence  and  the  two  measures  of 
leader  behavior  as  determined  by  the  instruments  used  in  this  study. 


Discussion 

Although  the  theory  for  the  hypotheses  tested  in  this  section, 
which  was  based  on  scanty  research,  suggested  that  there  might  be  a 
relationship  between  Ambivalence  and  style  of  leader  behavior,  the 


•  V  ;  •  ; 


...  .  .  .  :  :V  ■  . 


. 


.  . 


jI 


j  .  .  J  ;  ...... 


.  . 


...  >-■  ./  r  J  '  ■■  ; 


■  .  L  . .  1  L  '.  ' '  ...  ■  >  ...  .  ’.  '  ....  ... ..  ' ' . 

••  .  .  .  '.  ■:/  .'  •  ;  "  •: 


.... 


-  .  ..  .  .  .v  '  .  .  ’  \ .  *  •.  .  «  .  .  )  ».  V  • 


••  *  .....  * 


'  . 


. 


•  v.  o  : 

.  •  t  'J.:C  ■  J'  '  1 .  :  : 


1 


:  ■  • 


152 


analysis  carried  out  to  this  point  failed  to  reveal  any  significant 
relationships.  The  t_  technique  did  suggest  that  the  relationship 
hypothesized  for  Ambivalence  and  Structure  was  supported  at  least  in 
direction,  but  this  was  no  longer  evident  when  linear  and  curvilinear 
correlation  coefficients  were  calculated.  It  is  interesting  to  note 
that  the  relationship  between  Ambivalence  and  either  of  the  two 
dimensions  of  leader  behavior  is  much  lower  than  the  relationship 
which  has  been  found  to  exist  between  the  dimensions  themselves.^ 

TABLE  XV 

F  RATIOS  FOR  SIGNIFICANCE  OF  LINEAR  AND  CURVILINEAR 
RELATIONSHIPS  BETWEEN  AMBIVALENCE  AND  TWO 
DIMENSIONS  OF  LEADER  BEHAVIOR 


Ambivalence 

and 

Correlation 

Ratio 

h 

Linear 

Correlation 

P2 

Curvi- 

linearity 

F 

3 

Structure 

1.3,2 

0.25 

I.46 

Consideration 

0.54 

0.19 

0.59 

Degrees  of  freedom 

9/46 

1/54 

8/46 

F  required  for 
significance  at 
.05  level 

2.11 

4.02 

2.15 

Th is  sugge s t s  the 

possibility  that 

Ambivalence  might 

more  profitably 

be  considered  as  a  separate  dimension  of  leader  behavior  rather  than 


"In  this  study  the  correlation  between  the  two  dimensions 
was  .21. 


.  ■  .  >  . ■  r  . .  ■  c  ;  '■  ■ 

■  >  ■  •  •  -  :  ■  ■  :  ■  '  '  J  : 

’  ')■ 

•  ■  .  •  J-  ■  .  •  .  •  •  y 

'  ’ .  ■  V j;  '  :v  :  J\  •. 

"  '  •  '  .V  ■  ■  -  ■>  > 


•  V 


. 


■ 


. 


' 


‘ 


153 

as  a  variable  which  influences  other  dimensions.  It  is  also  quite 
possible  that  the  two  dimensions  of  leader  behavior  with  which  a 
relationship  was  hypothesized  are  not  appropriate  for  this  purpose. 

In  view  of  the  developments,  it  might  well  be  that  a  dimension  such 
as  Social  Sensitivity  would  be  rela,ted  in  a.  significant  way  to  the 
extent  to  which  a  principal  expresses  himself  as  experiencing 
difficulty  in  maicing  his  choices. 

II.  RELATIONSHIP  OF  STRUCTURE  AND  CONSIDERATION  TO 
AGREEMENT  AND  CONSENSUS 

The  study  was  also  concerned  with  the  relationship  of  teacher- 
principal  agreement  on  expectations  and  consensus  within  the  school 
to  the  leader  behavior  of  the  principal.  In  this  section  the  relevant 
hypotheses  are  examined  in  turn. 

Testing  Hypothesis  Two 

Some  of  the  existing  research  suggested  that  the  ability  to 
exhibit  behaviors  indicative  of  both  Structure  and  Consideration  are 
desirable  in  a  leader.  On  this  basis  it  was  hypothesized  that  each  of 
these  would  be  related  to  the  extent  to  which  there  was  Agreement 
within  the  school  concerning  expectations  for  the  role  of  the  princi¬ 
pal.  The  following  specific  hypotheses  were  tested: 

2.1  Agreement  will  be  positively  related  to  Consideration. 

2.2  Agreement  will  be  positively  related  to  Initiating 
Structure. 

In  order  to  test  these  hypotheses,  principals  were  again  divided  into 


■  L  '  7  v’o  ■  .  7  r  o  7  7  7 

)  ■ :  ■'  .  ' .  7  c:-  .  ■■  !  7  '7 

-s'  . 

' ..  7  V  •'  .  >  .  7  •  -  ' .7.7:  .•  . 

'  >  .  ■  77..<  .■  7  -  ■  .•  ■  J  .1  .7  ' 


. 


v  . 


...  ...  •  •  ,  %  :  ...  -i  J  .  '  \  .  >. 


'  ' 


: .  > 


'  .... 


-.1 


'  ■  ■■  •  V  >••  •  • 


' '  '  : 


:  '■  7  '.(U-  O  '  *r;  V  >:  'V 


t)  O' 


'  ...  y  '  '  .  .  ’  i  . 

'  '  J  -  :  7  ’  ,  .'  ■  ' 

>-v  : '  ■  •  ■  ■  7  '  1  .  .  •  ■  ■  <: ■  J 

•  .  .  -  ■  •  ■  o  7o  c  •  .  .  7  . 


■  .j..:-'.:  J 


:  ■  '  '  .  1 ,  t  u  .7 


/•i  .  .  ■  ■  . 


;)  .>•  wo  L  *  .* 


154 


two  groups  first  on  the  basis  of  their  Structure  scores  and  then  on 
the  basis  of  their  Consideration  scores.  The  mean  .Agreement  scores 
were  calculated  in  each  case  and  the  differences  were  tested  for 
significance .  The  data  in  Table  XVI  show  that  the  differences  not 
only  were  in  the  predicted  direction  but  also  tha,t  the  difference 
between  mean  Agreement  scores  when  principals  were  classified  accord¬ 
ing  to  Consideration  scores  was  significant  beyond  the  .05  level. 


TABLE  XVI 

MEM  AGREEMENT  SCORES,  VARIANCES ,  AND  SIGNIFICANCE 
OF  DIFFERENCES  FOR  PRINCIPALS  OF  HIGH  AND  LOW 
STRUCTURE  AND  CONSIDERATION 


Leader 

Principal 

Diff. 

* 

Behavior 

Classific ation 

Between 

sed  t 

Dimension 

High 

Low 

Means 

M 

52.11 

47.90 

Structure 

s2 

91.47 

95-76 

4.21 

2.58  1.62 

N 

27 

29 

M 

52.86 

4-6.78 

Consideration 

s2 

64.20 

113.52 

6.08 

2.51  2.42 

N 

29 

27 

* 

t  >  2. 

s 

,01  required  for 

significance  at  .05 

level  (two-tailed  test) 

The  only  disturbing  feature  of  this  is  the  reappearance  of  the  dis¬ 
crepancy  in  variance  of  the  Agreement  scores  for  the  two  groups. 
This  raises  some  question  as  to  the  validity  of  the  _t  test,  but  on 


.  ; 


;  u 


\  ;  :  '  • 


o' 


. 


-  ■.) 


■  *1 


■  .  : 


1  ;i- •  /  :  J'./IL 


. 


1 


■  ■ 

v 

‘  - 


,  . 


V 


. 


..  :  ■ 


>.  .  i  . 


•  ‘  1  v 


•  . 


■) 


j 


155 


the  basis  of  earlier  discussion  concerning  the  assumptions  underlying 
the  test  one  might  accept  this  as  being  indicative  of  a  significant 
difference . 

Although  Table  XVI  shows  that  there  is  no  significant  rela¬ 
tionship  between  Structure  and  Agreement,  the  t_  ratio  does  approaoh 
the  level  required  for  significance  in  a  one-tailed  test.  A  one- 
tailed  test  is  appropriate  in  view  of  the  hypothesis  which  is  being 
tested.  The  hypotheses  were  also  tested  using  a  correlational 
approach.  Table  XVII  shows  the  linear  and  curvilinear  correlation 
coefficients  for  the  relationships. 

TABLE  XVII 

LINEAR  AND  NONLINEAR  CORRELATION  COEFFICIENTS  FOR 
THE  RELATIONSHIP  BETWEEN  AGREEMENT  AND  TWO 
DIMENSIONS  OF  LEADER  BEHAVIOR  (N=56) 


Agreement 

and 

r  * 

Ay 

VAy 

* 

Structure 

0.28 

0.J8 

Consideration 

0.24 

0.51 

“X- 

Significant  at  .  03  level. 


It  is  evident  that  when  a  correlational  approach  is  used  the 
relationship  between  Structure  and  Agreement  appears  to  be  statisti¬ 
cally  significant  while  the  relationship  between  Consideration  and 
Agreement  falls  short  of  significance.  This  can  once  again  be 
attributed  to  the  fact  that  there  was  a  sizable  difference  in  the 
variances  when  the  _t  test  was  used  to  test  the  significance  of  mean 


.  >  1  . 
i '  ... .  ■  c  ;  J'  .  l'  • . 

.  ■■ 

-  .  -  r  ex  c  ..  ■ .  •  .■  '  ■  /.; . ...  ■.  J  •  ..  .  ■ 

;  .  qi  •  -  ,.k  :. 

-  J  ••  :  V 

.  '  ■  •  ,<  •  ■  i  •  x  . 

:  ■  ■  .  .  .  l' 

.  .  .  ' 

'  '  ..  ...  ' :  ' .  : 


. 


/  : 


•  :  t  -  :  ■ ' 


i '  i  <J  •  . 


ISC.’  X  R.  " '  ! t  ...  ' 


.  : 


i 


,  .  •  .  A 


156 

Agreement  scores  in  schools  classified  on  the  basis  of  principals' 
Consideration  scores.  The  significance  of  the  linear  and  curvilinear 
relationships  were  tested  in  the  usual  manner  and  the  F  ratios  are 
given  in  Table  XVIII.  The  only  significant  relationship  which  is 
evident  is  the  linear  correlation  of  Structure  and  Agreement.  It 
would  appear  that  the  variance  of  the  scores  within  arrays  was  great 
enough  to  forestall  the  significance  of  the  correlation  ratio. 
Although  the  F  ratios  involving  Consideration  are  fairly  high,  they 
fall  short  of  being  significant. 

TABLE  XVIII 

F  RATIOS  FOR  SIGNIFICANCE  OF  LINEAR  AND  CURVILINEAR 
RELATIONSHIPS  BETWEEN  WO  DIMENSIONS  OF 
LEADER  BEHAVIOR  AND  AGREEMENT 


Agreement 

and 

Correlation 

Ratio 

L 

Linear 

Correlation 

F 

2 

Curvi- 

linearity 

p? 

* 

F 

0.68 

4.47 

O.36 

Struc ture 

df 

11/44 

1/54 

10/44 

Reqd  F 

2.01 

4.02 

2.06 

F 

1.59 

5.58 

1.37 

Consideration 

df 

10/45 

1/54 

9/45 

Reqd  F 

2.06 

4.02 

2.11 

# 

Significant  at 

.05  level. 

■  XX?.  X 


.I 


....  r i 


157 


Discussion 

On  the  "basis  of  the  t_  test,  it  appeared  that  Consideration  was 
related  in  a  significant  way  to  the  extent  to  which  there  was  agree¬ 
ment  between  teachers  and  principals  on  the  principal  role.  The 
relationship  between  Structure  and  Agreement  fell  short  of  signifi¬ 
cance  at  this  stage.  However,  when  product-moment  correlation 
coefficients  were  calculated  the  significance  of  relationships  was 
reversed  and  it  appeared  that  Structure  was  related  in  a  significant 
way  to  Agreement.  One  might  conclude  from  this  that  neither  of  the 
two  dimensions  are  related  to  Agreement  in  any  consistent  way,  but 
the  investigator  wan  more  inclined  to  take  this  evidence  an 
suggesting  that  the  hypotheses  were  supported  in  a  tentative  way. 

It  was  also  hypothesized  that  Agreement  would  be  related  to 
the  style  of  leader  behavior  as  defined  by  scores  on  the  two  dimen¬ 
sions  considered  at  the  same  time.  The  results  of  the  analysis 
related  to  this  hypothesis  will  be  discussed  in  a  later  section  in 
which  style  of  leader  behavior  is  defined  more  broadly  than  was 
envisaged  in  the  original  hypothesis. 

Testing  Hypothesis  Three 

The  relationships  between  Structure  and  Consensus  and 
between  Consideration  and  Consensus  were  investigated  by  testing 
the  following  hypotheses: 

J.l  Consensus  will  be  positively  related  to  Consideration. 

3.2  Consensus  will  be  positively  related  to  Initiating 
Struc  ture . 


' 


... 


’ 


; 


; 


■ . 


-!■ 

V 


■  •.  ■■  ■  .' '. . 

..  f  . 

•  •  ...  \.  ■ .  ■  ■..  ;;  .  '/ 


■ 


;■  •  ,  •  •  .  J:  :r  :> 


1 


.  .'.  ;  ■  .V '  ■ 


■  - '  .  J:  .  ’  ." 


- 


« 


<c  >• 


158 


In  order  to  test  these  hypotheses,  the  principals  were  first  divided 
into  two  groups  on  the  basis  of  their  Structure  scores.  The  mean 
Consensus  score  was  calculated  for  each  of  these  groups,  and,  as  is 
evident  from  Table  XIX,  the  difference  falls  short  of  being  significant. 
In  fact  it  is  not  even  in  the  hypothesized  direction  and  suggests  that 
principals  who  are  high  in  Structure  are  not  able  to  achieve  as  much 
Consensus  in  their  schools  as  do  those  who  are  low  in  Structure. 

However,  it  must  be  noted  that  the  difference  is  not  statistically 
significant . 


TABLE  XIX 

MEM  CONSENSUS  SCORES,  VARIMCES,  MU  SIGNIFICMCE 
OP  UIFFERENCES  FOR  PRINCIPALS  OF  HIGH  MU  LOW 
STRUCTURE  M'U  CONSIDERATION 


Leader 

Behavior 

Dimension 

Princ ipal 
Classification 

High  Low 

Diff . 
Between 
Means 

seb 

* 

t 

M 

49-70 

50.31 

0.61 

2.62 

0.23 

Structure  S^ 

96.58 

94-66 

N 

27 

29 

M 

51.45 

48.48 

2 

Consideration  S 

113.45 

71.58 

2.97 

2.58 

1.15 

N 

29 

27 

t  >■  2.01  required  for  significance  at  .05  level  (two-tailed  test) 


i 


■  ■  ■  •  ■  .  .  ■  ■  -  : 


J  . 

•  :  •  ■  : .  . 


I  ■  •  ' :  , 


'  .  ; 


••• 


. 


. 


'  ■  .  J 


' )  ■  ■.  i 


.  • : 


.  , 


159 


Principals  were  again  classified  into  two  groups  on  the  basis 
of  their  Consideration  scores  and  the  mean  Consensus  scores  were 
calculated  for  the  two  groups.  Table  XIX  reveals  that  the  difference 
was  in  the  hypothesized  direction  but  fell  short  of  being  statistically 
significant.  Of  interest  again  is  the  extreme  difference  in  the  vari¬ 
ances  of  Consensus  scores  for  schools  in  which  principals  were  high 
in  Consideration  and  those  in  which  the  principals  were  low  in 
Considera/tion.  It  would  almost  appear  that  high  Consideration  scores 
tend  to  be  associated  with  a  wider  range  of  Consensus  scores  than  are 
low  Consideration  scores.  This  suggests  that  it  may  be  easier  to 
reduce  consensus  in  a  school  than  it  is  to  promote  it. 

The  theory  which  led  to  the  hypothesis  of  a  curvilinear 
relationship  between  Ambivalence  and  other  variables  also  suggested 
that  it  might  be  worthwhile  to  examine  the  relationships  between  the 
leader  behavior  dimensions  and  Consensus  using  the  correlation  ratio. 
Both  product -moment  and  correlation  ratio  coefficients  were  calculated 
from  a.  scattergram,  and  the  results  of  the  analysis  presented  in 
Table  XX  indicate  low  negative  linear  correlations  between  each  of 
Structure  and  Consideration  with  Consensus.  All  relationships  fell 
short  of  being  statistically  significant. 

Although  the  correlation  ratios  are  numerically  higher  than 
the  Pearson  rs,  the  F  ratios  in  Table  XXI  reveal  that  these  are  not 
statistically  significant.  This  is  also  true  in  the  case  of  the  test 
for  curvilinearity  of  regression  between  the  variables.  The  tests 
carried  out  to  this  stage  did  not  support  the  hypotheses  which  had 


J 

'  ■  :  I  - 

i 

' 

■qXi  ' 

■  ■ 

, 

>'  .. 

'  •'  ' 

t 

' 

.  | 

1 

;  •  ; 

'  .  M  v 

■ 

. 

.  i 

.  .c  ii;  . 

• 

. 

•  »  ... 

:■  ■ 

, .  1  ■ 

• 

!: 

■■  ' 

: 

■ 

-X  .  , 

'  ■ 

• 

' 

•  ■  : 

c  .. 

■  ■ 

;■  ;  J- 

'  •  •:  : ■ .. 

r 

j: 

.  : 

- 

,  ' 

i  i"  J 

■■  ' 

Ij  !  r-f 

i 

i 

>  '  %• 

£$  ■ 

. 

: 

• 

■  'l 

• 

,  ■ 

- 

:  '  ■'  '  .  ij  C 


.  •  ■ 


r 

.* .  *v 


.  • 


160 


been  proposed. 


TABLE  XX 

LINEAR  AND  NONLINEAR  CORRELATION  COEFFICIENTS  FOR 
THE  RELATIONSHIP  BETWEEN  CONSENSUS  AND  TWO 
DIMENSIONS  OF  LEADER  BEHAVIOR  (N=56) 


Consensus 

and 


rCy  ^  Cy 


Structure 


-0,13 


0.41 


C  ons  i  der  ati  on 


-0.06 


0.45 


TABLE  XXI 

F  RATIOS  FOR  SIGNIFICANCE  OF  LINEAR  AND  CURVILINEAR 
RELATIONSHIPS  BETWEEN  TWO  DIMENSIONS  OF 
LEADER  BEHAVIOR  AND  CONSENSUS 


Consensus 
and  . 

Correlation 

Ratio 

L 

Linear 

Correlation 

F 

2 

Curvi- 

linearity 

P3 

F 

0.83 

0.08 

0.90 

Structure 

df 

11/44 

1/54 

IO/44 

Reqd 

F  2.01 

4.02 

2.06 

F 

1.15 

0.02 

1.27 

Consideration  df 

10/45 

1/54 

9/45 

Reqd 

F  2.06 

4.02 

2.11 

161 


Discussion 

The  statistical  tests  which  were  used  failed  to  reveal  any 
significant  relationship  between  either  Structure  and  Consensus  or 
Consideration  and  Consensus.  When  correlation  coefficients  were 
calculated,  the  results  were  in  a  negative  direction  but  did  not 
differ  significantly  from  zero.  It  would  appear  that  neither 
Structure  nor  Consideration  taken  separately  are  related  to  the 
degree  of  consensus  which  exists  among  the  teachers  in  a,  school  on 
the  expectations  for  the  behavior  of  the  principal.  The  hypothesis 
relating  style  of  leader  behavior  and  Consensus  will  be  discussed 
in  a  later  section. 

III.  RELATIONSHIP  OP  AMBIVALENCE  TO  CONSENSUS 
AND  AGREEMENT 

Testing  Hypotheses  Four  and  Five 

To  this  point  it  appears  that  Ambivalence,  as  defined  and 
measured  in  this  study,  is  unrelated  to  either  of  the  two  dimensions 
of  leader  behavior.  It  was  also  hypothesized  that  there  would  be  a. 
positive  relationship  between  Ambivalence  and  the  extent  to  which 
there  was  consensus  on  expectations  for  the  behavior  of  a  principal 
among  the  teachers  in  a  school.  It  was  further  anticipated  that  the 
extent  to  which  a  principal  expressed  difficulty  in  making  his 
behavioral  choices  would  also  be  related  to  the  extent  to  which 
there  was  agreement  between  the  principal  and  the  teachers  in  his 
school  on  the  role  of  the  principal.  Stated  more  specifically  the 
hypotheses  are  as  follows: 


■  .  .  ..  . 

.  1 

:  A; 

.  .  :  • 

’  ‘  O 

&C,  ■ 

ri 

X  E  t 

A  'A  '  A. 

'  '■ 

'j.-'  r  .A 

. 

A'  : '  A 

. 

‘v-  ■  * 

-  ,  ; 

•.A: 

A.  A; 

.A.  ■ 

1 '  A. 

•.  :  A  J  •  A  •  ■ 

A  ■ .  • 

A  A  ■.  ■ .  a 

co  A 

-A  . 

i  '  .  ; 

. 

.  U 

0  , 

A  A  , 

A:  A  A  : 

Oi 

•  ;•  •  . 

.  ■ 

* 

A 

A 

. 

A  C'  !  :.  : 

-  :  v 

;  • :  . .-  :•  . 

; 

• 

3  .'.A 

'kJ 

•  * 

A 

-  ' 

Cl. 

■ 

■  v  .j;  A* 

■ 

A.  •.  'A 

•  A  vfjJ: 

A:  A 

i  i 

’ 

A 

A 

-  ; 

■  A:  iA  . 

' .  ‘  J  ’ 

A  -  ■  • :  A  A 

A.-.'  A 

(  a  r 

.  ■ 

A'y  A 

■i  i. 

A- 

j 

A  ■  '  A  , 

A.  A  .  ■ 

:  vA:A 

1 

> .  j  ' .J 

..'•.A.'  'Ac  .  •  ■  ■■■o  ' 

..  : 

• 

• 

v.  A  ■  A,  . 

...  A  ni  l 

•..■j:  '  . 

A  •  A’.'. 

.  ;  A.  .  . 

• 

.  .  J- 

■j  ,  o'  A 

:  A:  ;  . 

' 

i. 

:  ; '  j 

;  '?.  /  £ 

• .  id 

' 

• 

.A  A  : 

. 

j  co  loo  Ac: 

• 

162 


4.1  The  relationship  between  Ambivalence  and  Consensus 
will  be  positive. 

5.1  The  relationship  between  Ambivalence  and  Agreement 
will  be  positive. 

Principals  had  already  been  classified  into  Highs  and  Lows  in 
Ambivalence  in  order  to  test  the  initial  hypotheses,  and  now  mean 
Consensus  and  Agreement  scores  were  calculated  for  the  two  groups. 
Table  XXII  shows  means,  variances,  and  the  significance  of  differences 
between  the  means  for  both  Consensus  and  Agreement.  Although  both 
differences  were  in  the  predicted  direction,  neither  was  large 
enough  to  be  of  statistical  significance  at  the  .05  level. 


TABLE  XXII 

MEM  CONSENSUS  MD  AGREEMENT  SCORES,  VARIMCES,  MD 
SIGNIFICMCE  OP  DIFFERENCES  FOR  PRINCIPALS 
OF  HIGH  MD  LOW  AMBIVALENCE 


Le  ader 

Behavior 

Dimension 

Ambivalence 

High  Low 

N=28  N=28 

Diff. 

Between 

Means 

sed 

* 

t 

M 

51.04 

49-00 

Consensus  _ 

1.04 

2.60 

0.40 

nd 

D 

95-89 

93.00 

M 

51.07 

48.79 

Agreement 

2.28 

2.62 

0.87 

S^ 

112.28 

79.55 

t  >  2.01  required  for  significance  at  .05  level. 


It  must  be  recognized  that  such  differences  could  easily  arise  due  to 
chance  errors  in  sampling.  It  is  again  interesting  to  note  the  rather 


.‘.I 


•  m:  ■  '  ■  .  v 


:  :  ■  - 


: .  .  •  y.:  V:  :  .  :  .V  1: 


-  y  ..  .  ■.  >.  i  :  '  v  •;  , 


.  •;  :  1  -  '  'j  1  ;  r  ..  .  :  - 


iri- 


.r 


■  .  -  .  < 


r  .  •  ; .  . 

■  •  ■  •  * 


. 

.  :  » 

:  r  .  i  • . 

. 


. 


i  ■■  . ior  < 


.V  .  .  •  o' 


...  .  ' :  i  ;  • 


165 


large  difference  in  the  variance  of  Agreement  scores  for  the  Highs 
and  the  Lows  in  Ambivalence;  this  difference  is  not  evident  between 
the  two  sets  of  Consensus  scores. 

The  theory  also  suggested  that  the  relationships  between 
Ambivalence  and  the  two  other  variables  might  not  be  linear  and  the 
following  hypotheses  were  tested  next: 

4.2  The  relationship  between  Ambivalence  and  Consensus 
will  be  curvilinear. 

5.2  The  relationship  between  Ambivalence  and  Agreement 
will  be  curvilinear. 

In  order  to  test  these  hypotheses  both  linear  and  non-linear 
correlation  coefficients  were  calculated.  The  results  are  presented 
in  Table  XXIII.  These  indicate  fairly  low  linear  correlation  between 
Ambivalence  and  Consensus  with  a  higher  but  still  insignificant 
correlation  between  Ambivalence  and  Agreement.  This  would  suggest 
that  Ambivalence  is  more  related  to  the  extent  to  which  principals 

TABLE  XXIII 

LINEAR  AND  NONLINEAR  CORRELATION  COEFFICIENTS  FOR  THE 
RELATIONSHIP  BETWEEN  AMBIVALENCE  AND  TEACHER 
CONSENSUS  AND  AGREEMENT 


Ambivalence  and 

rAy 

n  Ay 

Consensus 

0.05 

0.40 

Agreement 

0.20 

0.42 

are  able  to  achieve  Agreement  between  their  self -expectations  and 
those  which  their  teachers  hold  for  them  than  they  are  able  to 


■'  r.:  .  c. . 


x  .  Jo:-  x  I  -v-x  •: 

.  :  J  .  . 

.x  .. 

x';.  Jr/  :  c  IS.h 


:  - 


.  -  ' .V/  v  .  . 

.  J-  f  o  J'  J' 


■  xt.  ■■  -  -  ■ 


- 


.  D 


.  I  X 


JUl 


■  .  :  '  . 


&  •  .)  v  ••  ' 


i 


. 


:  •  i  v 


r.  >  -  x 

X  J 


■  & 


L). 


164 


bring  a.bout  Consensus  among  staff  members. 

The  significance  of  linear  and  curvilinear  relationships  was 
again  tested  by  calculating  the  P  ratios  in  Table  XXIV.  It  is  evident 
that  although  the  ratios  for  the  relationship  between  Ambivalence  and 
Agreement  are  higher  than  those  for  Ambivalence  and  Consensus  they 
still  fall  short  of  statistical  significance . 


TABLE  XXIV 

P  RATIOS  POR  SIGNIFICANCE  OP  LINEAR  AND  CURVILINEAR 
RELATIONSHIPS  BETWEEN  AMBIVALENCE  AND 
TEACHER  CONSENSUS  AND  AGREEMENT 


Ambivalence 

and 

Correlation 

Ratio 

P 

1 

Linear 

Correlation 

P2 

Curvi- 

linearity 

P3 

Consensus 

0.95 

0.04 

1.06 

Agreement 

1.11 

2.19 

0.97 

Degrees  of  freedom 

9/46 

1/54 

8/46 

F  required  for 
significance  at 
.05  level 

2.11 

4.02 

2.15 

Discussion 

The  results  of  the  analysis  in  this  section  fall  into  the 
disconcerting  area  of  being  in  the  predicted  direction  and  yet  not 
statistically  significant.  While  recognizing  that  the  observed 
differences  could  have  arisen  by  chance,  it  is  still  interesting  to 
speculate  that  these  are  the  differences  which  were  expected  and 


■  ■  ;  ■  ■  d'  ;  x\  ■  • 

. o  ;  -V.  _  •. 

d  „  .  j‘  .  d  ..  c  i  d'  d' 


:•  ■  • 


- 


&  d 


v.c 


-  -  v. 


•-  ' 


1 T 


. 


: 


i- 


•  j:  •  ' 


-u  ,  ,  r 


j 


165 


might  well  have  been  significant  had  there  been  a  larger  sample.  One 
is  forced  to  conclude,  however,  that  Ambivalence  as  a  single  variable 
apart  from  other  variables  is  not  related  significantly  either  to  the 
extent  to  which  a,  principal  is  able  to  bring  the  mean  of  the  expecta¬ 
tions  which  teachers  hold  for  him  close  to  his  self-expec tations  or 
the  extent  to  which  he  is  able  to  bring  about  a  higher  degree  of 
consensus  within  his  staff  on  the  expectations  which  they  hold  for 
his  behavior. 


IV.  RELATIONSHIP  OF  STYLE  OF  LEADER  BEHAVIOR 
TO  CONSENSUS 

The  analysis  outlined  in  the  previous  sections  revealed  that 
neither  Ambivalence,  Structure,  nor  Consideration  were  related  in 
any  significant  way  to  Consensus.  Even  though  none  of  the  variables 
were  related  individually,  there  was  still  the  possibility  that  cer¬ 
tain  combinations  of  these  could  be  more  important  than  any  single 
variable  when  Consensus  was  involved.  Thus  it  would  have  been 
desirable  to  compare  Consensus  scores  in  schools  where  principals 
were  classed  as  being  high  in  all  of  Structure,  Consideration,  and 
Ambivalence  with  Consensus  scores  in  schools  where  principals  were 
low  in  all  of  these.  Such  a  direct  analysis  using  the  collected  data 
was  not  feasible  in  view  of  the  relatively  small  number  of  principals 
that  there  would  be  in  each  of  these  categories.  In  order  to  meet 
this  problem  it  was  decided  to  work  with  combinations  of  two  vari¬ 
ables  at  one  time  so  as  to  obtain  some  indication  of  the  effect  on 


Consensus  of  these  various  leadership  styles. 


. 


' 

,  ;  7  J--  c  '  .  !  d‘  "  7: 


.  i 


i  • 


-  7’  :  '  7 

.  ;  ■  '  -  .  . :  .  L  ..  I  .  C  '  !  ;  '  -  l  . :  . 

-  '■  j-  I.  •  7  : 

■  7  J'  :  7  ' 


1! . 


•  ;  . 
.  .  \ '  r 


J  "  7'  7s  J  :  7"  7:  7:77  •  ■  : 

7:  •  7  ■.  •  •  •  ,  7  7'  ,  .  .  7:  :  ■  t  .  7.  ■ 


7  -  7  .  7  .  :  7  7  v  ;  7  7:  7: 


-  7  7  :•  7  ■  7:  .  •:  7’  '  7  7-  7:  :  7:  7' 

.  .  7  7.  '  7  7  7.'  v 

'  :  ••  7  .  f  :  .  ;  7  '7.  ' 


cii  7  - 


1 66 


Hypotheses 

Originally  it  had  "been  hypothesized  that  Consensus  would  "be 
related  to  style  of  leader  "behavior  in  terms  of  Consideration  and 
Initiating  Structure  scores  only.  In  view  of  the  insignif icant  rela¬ 
tionships  between  Ambivalence  and  the  other  two  measures,  it  was 
decided  to  incorporate  this  variable  into  the  concept  of  style  of 
leader  behavior.  In  addition  to  testing  the  original  hypothesis: 

3.3  Consensus  will  be  higher  in  schools  where  the  principal¬ 
is  high  on  both  dimensions  of  leader  behavior. 

it  was  decided  to  test  the  following  hypothesis  a.s  well: 

3.4  Consensus  will  be  related  to  style  of  leader  behavior  as 
defined  by  Consideration,  Structure,  and  Ambivalence 
scores . 

Both  of  these  hypotheses  were  tested  in  the  context  of  the  same 
analysis . 

When  combinations  of  the  three  variables  Structure,  Considera¬ 
tion,  and  Ambivalence  are  taken  two  at  a  time  and  each  variable  is 
dichotomized,  it  is  evident  that  there  will  be  four  different  groups 
or  four  leadership  styles  each  time.  These  are  as  follows: 

(i)  C+S+;  C+S-;  C-S+;  C-S- 

C  =  Consideration 

(ii)  C+A+;  C+A-;  C-A+;  C-A-  S  =  Structure 

A  =  Ambivalence 

(iii)  S+A+;  S+A-;  S-A+;  S-A- 

The  mean  Consensus  scores  for  each  of  these  styles  of  leader  behavior 
are  given  in  Table  XXV.  It  must  be  kept  in  mind  that  the  principals 
in  each  of  the  three  cells  are  the  same  individuals  and  that  the 


Consensus  scores  are  for  the  same  schools  in  each  of  the  three  cells. 


V 


. 


; ' 


r  y.-\  ' 

k  yr  :  . 

vs  V  .  .  > 


.  . 


- 


.  . 

■  f  i: 


■  I 


' 


J:  : 

r  .  s'- 

^  •'  ; 
,  .  .•  . ’  ,!'  i 


. 


•  *  *  ’  >  o  '  : : 

s. 


... 


. 


..  .  .  ,  ■  ; 


167 


For  each  of  the  three  cells  an  analysis  of  variance  was  carried 
out  to  determine  whether  there  were  any  significant  differences  between 
means;  that  is,  an  analysis  of  variance  was  carried  out  on  the  data 
from  which  the  means  in  Table  XXV  were  calculated.  The  F  ratios  which 
were  obtained  are  given  in  Table  XXVI.  It  is  evident  that  only  one  of 
these  approached  significance  and  that  was  when  Structure  and  Consider- 
ation  were  combined  to  form  four-  different  styles  of  leader  behavior. 


TABLE  XXV 

MEAN  CONSENSUS  SCORES  BY  LEADERSHIP  STYLE  OF  PRINCIPAL  (N=56) 


High 

Struc ture 

Low 

Structure 

High 

Ambivalence 

Low 

Ambivalence 

High 

Consideration 

51.25 

51.69 

52.55 

50.28 

Low 

Consideration 

47.45 

49.19 

49.31 

47.71 

High 

Structure 

48.07 

51.21 

Low 

Structure 

53. 60 

46.78 

When  Consideration  was  combined  with  Ambivalence,  the  F  ratio  was 
below  unity  and  therefore  highly  insignificant .  From  this  it  would 
appear  that  style  of  leader  behavior  is  not  related  to  Consensus  in 
any  statistically  significant  way.  In  spite  of  this  some  interesting 


trends  are  discernable  in  Table  XXVI. 


; 


;  ;  j;  /  V  :  '  J.  ■.  . 

:  ■  ■■  ■  ‘  . 

.  ■  ■  :  rr  >  J  re  iv.  ]:■  /,:  r,  ..  ;  .  .  :  . 

■ 


:  :  '  .  .  •  '  .  u  ■  ‘  v  '  '  '  : 

.7/  '  ’  ':X:  o’; 


;  •  ;  •  .  {.-••• 

•  ■  >v  '  O'  ■  r  • 

.  >'■  •: r 

'  . 

,  .  J  ■ 

>  ’  1  V  -  -  '  ‘  '  ‘  U 


V  'v  ■ 


j; 


. 


'  ■:  J:  : 


.  ■  / 


. 


j-  <  .  ■ 

r-  J  i 


■’  ' 1 1  •  V 


168 


This  table  reveals  that  when  Consideration  was  combined  with 
either  of  the  other  two  variables,  the  principals  who  were  high  in 
Consideration  were  associated  with  higher  Consensus  scores  than  were 
those  principals  who  were  low  in  Consideration.  Both  Structure  and 
Ambivalence  exhibited  consistency  when  combined  with  Consideration 
but  not  when  combined  with  each  other.  The  lower  right  hand  cell 
would  seem  to  indicate  that  S-A+  and  S+A-  are  associated  with  a  higher 
degree  of  Consensus  than  are  the  S+A+  and  S-A-  styles  of  leader  be¬ 
havior.  The  differences  between  some  of  the  means  in  this  cell  are 
sizable . 

TABLE  XXVI 

F  RATIOS  OBTAINED  THROUGH  ANALYSIS  OF  VARIANCE  OF 
CONSENSUS  SC  ORES  GROUPED  BY  PRINCIPALS' 

STYLE  OF  LEADER  BEHAVIOR 

Structure  Ambivalence 

Consideration  2.49  O.58 

Structure  1*59 

df  =  3/52  F  >  2.79  required  for  significance  at  .05  level 

Discussion  and  Further  Analysis 

This  unexpected  finding,  an  example  of  the  operation  of 
serendipity,  aroused  enough  curiosity  to  cause  further  consideration 
to  be  given  to  the  data.  It  appeared  that  for  principals  who  were 
high  in  Structure,  Ambivalence  was  related  inversely  to  Consensus  but 
that  for  principals  who  were  low  in  Structure,  Ambivalence  and 


i  :  .. 

•'  ,  v  .  .  ■  :/  V.  •. 

■.  •  ■,  .  ■ 

•  ’  •  •  L-'  >  ,  :  J- 

..  :  ■'  .  . 

.  .  O  .  ..  «  v.  :  .  : 

•  ■  .  :  J'  j  n.r 

■■  - 


.. 


- 


;  :ot 


■  .  .« 


J  . 


Consenstis  varied,  directly.  The  same  sort  of  observation  could  be  made 
for  the  relationship  when  high  and  low  Ambivalence  rather  than,  high  and 
low  Structure  were  considered.  In  order  to  analyze  these  relationships 
further,  coefficients  were  calculated  to  determine  the  intercorrelations 
of  the  three  variables  for  both  high  and  low  Structure  principals. 

Table  XXVII  shows  that  for  principals  who  were  high  in  Structure,  the 
correlation  between  Ambivalence  and  Consensus  was  -0.29  while  for 
principals  who  were  low  in  Structure  the  correlation  was  0.33.  Although 
each  of  these  falls  short  of  being  significant  at  the  .05  level,  the 
difference  between  the  two  is  statistically  significant.  Consensus 
and  Structure  were  negatively  correlated  for  eanh  group  even  though 
the  range  of  Structure  scores  had  been  cut  in  half.  The  correlation 
between  Ambivalence  and  Structure  for  low  Structure  principals 
approached  significance  at  the  .05  level,  but  the  difference  between 
this  coefficient  and  that  for  the  other  group  was  not  statistically 
significant. 

The  data  with  respect  to  Ambivalence,  Structure,  and  Consensus 
are  of  considerable  interest  but  are  not  readily  explained  either  in 
terms  of  some  of  the  theory  developed  for  this  study  or  in  terms  of 
other  theories.  It  would  seem  more  fruitful  to  look  at  relationships 
between  the  three  variables  in  schools  classified  on  the  basis  of 
Consensus  rather  than  on  the  basis  of  Structure  of  principals  since 
the  latter  appears  to  have  little  if  any  relationship  to  Consensus. 

As  further  exploration  into  these  relationships,  the  schools 
were  divided  into  two  groups  on  the  basis  of  Consensus  scores  and 


■  -  ,  77.  •  •  '  ; 

"  c  '  '  .  •  ;■  7:7  7 ■  • ;  7 

;  . :  .  .  -  '  7'  ■  7  ■  .  »  •  ;  •  7'c.  v7  • 


i'j  '  L 

'  '0  r  '  .  7  •  7  .  7;  .7 7;  •  .  - ' 

. 

7  -  7  '■■■  7  '  7  ■  r/  •  .  •.. :  7  -  7 

Q  '  ■  '  ‘  J.' 

7-  7';  t  7  - -  7  J  7  IIVXX  ©I 

■  ' :  .7;  7  *  •• 

■ '  ■  -  .  '  ... :  ‘  7  '  7  .7  ...  ■  7.7  -.7.  c 

7  7 

7'  •  •  ;  77  ■  -i.vc'  •  7  7  ■  ;7  /  r-.7  '7;.  .  : 

..  .  '  .  7 

&  7  i  ■  ■  7  77  .  7‘  ■  ■  . 

■  . ;  .  ,  7 

7  c.  77  7.7  7  7  7  ,  7  c  •  7  7 

.  :  ■  •  7  c  .  \  .77  c  •  ■  ■  ■  7c  :o:7  r  • 

.7  ■■  r 

.  •  7  .  :ji  ■  I  ■  C  .  7' :  ■  -m  •...  7 

J  i  7  r 

•.  ;  •  .  •  -  ■  7  ©re  .!  :v7..7 '  rd'£ 

C  '  ■'  . . 

^  .  07  ■  d  ?0iv:,c  .7  .7  :  7. 

.7r-o7.7  77  7.  J- 

•  ;7  :  7:c-7  77:  o  :.±: 

,7-  ■  l  77.7  '  7: 

■  t  •  7 

c  :  /7  1  '  7  of  ■  77  f  7  •  - 

'  7:  7  .  ■  ■  7> 


7  *3 

:  .7  .  v 

o  7'  ■  ~  -  7' 

7  7.7  '  ■ : 

.77  q  7:  1  7-  ■  ]  7 

'  ;  •  .  .7  ■  t  7  ■  7'  -  v  ■  ■  7 

■  7  7.  ■■  ■  ■  •  -7 

•'  <  '  '•  ' 

-  •  7  7:  7''  ■ 

7  '7  7 

170 


the  intercorrelations  for  the  three  variables  were  again  calculated. 
The  correlations  between  Consensus  and  Ambivalence  and  between  Consen¬ 
sus  and  Structure  were  very  low  for  each  group  and  the  difference 

TABLE  XXVII 

INTERCORRELATIONS  OP  THREE  VARIABLES  AND  SIGNIFICANCE 
OF  DIFFERENCES  FOR  SCHOOLS  OF  HIGH  AND  LOW 
STRUCTURE  PRINCIPALS 


High 

Structure 

Low 

Structure 

Difference 

z 

SE 

Principals 

N=27 

Principals 

N=29 

r 

z 

z 

rA3 

0.04 

^0 

N-\ 

• 

O 

1 

0.40 

(.42) 

1.48 

rcs 

-0.21 

-0.28 

0.07 

(•07) 

0.25 

r 

AC 

-0.29 

0.33 

0.62 

(.64) 

* 

2.26 

A  = 

Ambivalence 

C  =  Consensus 

s  = 

Structure 

*Significant  at  .05  level. 

between  between  the  coefficients  in  each  case  was  not  statistically 
significant  as  is  shown  in  Table  XXVIII.  On  the  other  hand,  the 
correlation  between  Ambivalence  and  Structure  in  high  Consensus  schools 
was  -O.42  which  is  significant  at  the  .05  level,  while  the  correlation 
between  these  two  va  riables  in  low  Consensus  schools  was  O.3O.  The 
difference  between  the  two  rs  is  significant  beyond  the  .01  level  and 
thus  gives  every  indication  of  being  a  true  difference.  If  one  inter¬ 
prets  this  to  mean  that  the  situation  in  v/hich  a  principal  finds 
himself  is  important  for  his  behavior,  then  one  possible  explanation 


might  be  a,3  follows. 


'■ 


1  L 


(-’■ 


i  ;  '' 


■  • 


D. 


<!  J  :  . . 


■  ' 


■ )' 


171 


In  a  school  in  which  there  is  a  high  degree  of  Consensus  it  is 
not  appropriate  for  the  principal,  in  terms  of  the  situation,  to  he 
very  concerned  about  conflicting  expectations  since  these  are  at  a  very 
low  level  anyway.  In  such  a  case  the  principal  who  acts  as  if  his 
behavioral  choices  are  difficult  to  make  is  not  acting  in  accord  with 
the  reality  of  the  situation,  and  the  more  Ambivalent  he  is,  the  more 
likely  it  is  that  he  is  going  to  be  viewed  as  being  low  in  Structure. 

TABLE  XXVIII 

INTERCORRELATIONS  OP  THREE  VARIABLES  AND  SIGNIFICANCE 
OP  DIFFERENCES  FOR  HIGH  AND  LOW  CONSENSUS  SCHOOLS 


High 

Consensus 

Schools 

N=27 

Low 

Consensus 

Schools 

N=29 

Difference 

r  z 

z 

SE 

z 

rAS 

* 

-0.42 

0.50 

0.72 

O.76 

** 

2.68 

rcs 

-0.05 

0.15 

0.18 

0.18 

0 . 64 

rAC 

-0.08 

0.08 

0.16 

0.16 

0.57 

A  = 

Ambivalence 

S  =  Structure 

C  =  Consensus 

^Significant  at  .05  level. 
**Significant  at  .01  level. 


In  the  low  Consensus  school  it  is  more  appropriate  for  the  principal 
to  be  concerned  about  conflicting  expectations  for  in  this  case  it  is 
in  accord  with  the  reality  of  the  situation.  Now  it  is  the  principal 
who  is  more  Ambivalent  who  is  viewed  as  being  high  in  Structure.  The 
implications  of  this  analysis  is  that  Ambivalence  which  is  independent 


e.  ."o;. 


'•  ;r  i  • 


* 


i>i 


I  J.t/  «...  ....  s  »  ... 


.  I  . 


.. 

'•r  ::  ij 


.» 


. 


; )■;  '  > 


:  ■/ . 


•  .  I  • 


/  1  .  "J  l '  ■! 


K  i  >  J 


>d  it 


tl  ■ .  '  •.  N  .J.  r  J:  J-  .  v.':  ■  ■ 

.  i  J'  •  ( 


172 


of  the  situation  or  not  modified  by  the  situation  is  not  as  important 
for  appropriate  leader  behavior  as  is  that  which  coincides  with  a 
realistic  interpretation  of  the  environment. 

V.  RELATIONSHIP  OP  STYLE  OP  LEADER  BEHAVIOR  TO 

AGREEMENT 

The  previous  section  has  revealed  that  style  of  leader  behavior 
is  not  related  in  any  significant  wa,y  to  the  degree  of  consensus  which 
exists  on  expectations  for  the  behavior  of  a  principal  in  any  one 
school.  There  was  still  the  possibility  that  leadership  style  as 
defined  previously  might  be  related  to  the  extent  to  which  there  is 
agreement  between  teachers  in  a  school  and  the  principal  on  expecta¬ 
tions  for  the  role  of  the  principal. 

Hypotheses 

In  order  to  investigate  this,  the  following  hypotheses  were 

tested: 

2.4  Agreement  will  be  higher  in  schools  where  the  principal 
is  high  on  both  dimensions  of  leader  behavior. 

2.5  Agreement  will  be  related  to  the  style  of  leader  behavior 
as  defined  by  Consideration,  Structure,  and  Ambivalence 
scores. 

The  same  quadrant  approach  as  was  outlined  in  the  previous  section 
was  used  to  analyze  the  data  in  relation  to  these  hypotheses.  The 
mean  Agreement  scores  for  the  data  on  which  three  analyses  of  vari¬ 
ance  were  performed  are  given  in  Table  XXIX.  An  examination  of  the 
table  reveals  th.a,t  whether  Consideration  was  considered  in  combination 


with  either  Structure  or  Ambivalence,  mean  Agreement  scores  for  the 


■  ■ 


• '  ';t  , '  I  S  .  d-  •  'i  .  -  •  \r  .  :  . .. 

Il  .  '  i  .  ■  ; 


l  i'  .  : 


. 


'  >  i.  &  i  .  . 


JIV 


i: 


"  !  :  ) 


'  '  J.  ■  ;  .1  ft;.'  It  . 


c. 


ii 


173 


high  Consideration  group  were  consistently  above  the  mean  Agreement 
scores  for  the  low  Consideration  group.  The  same  consistent  relation¬ 
ship  exists  as  far  as  Ambivalence  is  concerned  but  does  not  hold  in  the 
ca.se  of  Structure.  It  should  again  be  kept  in  mind  that  the  mean 
Agreement  scores  for  each  of  the  three  cells  are  based  on  data,  for  the 
same  schools. 


TABLE  XXIX 

MEAN  AGREEMENT  SCORES  BY  LEADERSHIP  STYLE 
OP  PRINCIPAL  (N=56) 


High 

Structure 

Low 

Structure 

High 

Ambivalence 

Low 

Ambivalence 

High 

Consideration 

56.44 

48.46 

53.07 

52.64 

Low 

Consideration 

45.82 

47.44 

48.77 

44.93 

High 

Structure 

52.46 

51.78 

Low 

Structure 

49-87 

45-78 

The  analyses  of  variance  resulted  in  the  F  ratios  which  are 


presented  in  Table  XXX.  All  three  of  these  are  fairly  high,  and  one 
of  these,  that  when  Structure  and  Consideration  are  used  in  the  quad¬ 
rant  approach  to  leader  behavior,  is  significant  at  the  .05  level. 
This  indicates  that  there  are  some  significant  differences  between 
the  mean  Agreement  scores  included  in  the  upper  left  hand  cell  in 


. 


;• 


i  .-I 

£ 

' 


, 


i V.  ;-.o  .  S. 

"  r '  V  -J  >:  c  .'f  '  '  '■ 

-  >  .  ■  ;  :  ■  J  . 

c  '  :;v  &  i  >■:  1  ■  :  • 

;  /  •  ■,  •  : .  /  :  J  ' 

; 


.  !  .  v  i 


174 


Table  XXIX.  The  analysis  of  variance  data  is  presented  in  Table  XXXI 
for  this  part  of  the  analysis  only;  the  others  are  not  presented  in 
view  of  the  fact  that  no  other  significant  differences  were  revealed. 

TABLE  XXX 

F  RATIOS  OBTAINED  THROUGH  ANALYSIS  OP  VARIANCE  OP 
AGREEMENT  SCORES  GROUPED  BY  PRINCIPALS' 

STYLE  OP  LEADER  BEHAVIOR 


Structure 

Ambivalence 

Consideration 

* 

3.86 

2.22 

Structure 

1.27 

df  -  3/52  P  >2.79 

required  for  significance  at 

.05  level. 

■^Significant 

at  .05 

level . 

TABLE  XXXI 

ANALYSIS  OP  VARIANCE  OP  AGREEMENT  SCORES  FOR 
STRUCTURE  AND  CONSIDERATION 

Source 

df 

Sums  of  Squares 

Variance 

SD 

Between  Groups 

3 

990.97 

330.32 

Within  Groups 

52 

4,452.75 

85.63 

9.25 

Total 

55 

5,443.72 

F  =  3.86  ( 

.05  > 

P  >  .01) 

In  order  to  determine  specifically  where  significant  differences 
existed,  a  t  test  was  used  to  check  the  differences  between  the  mean 


175 


Agreement  score  for  schools  in  which  the  principals  were  described  as 
having  the  S+C  +  style  of  leader  behavior  and  mean  Agreement  scores 
associated  with  other  styles  of  leader  behavior.  Although  the  same 
standard  deviation  of  9*25  was  used  in  calculating  the  standard  error 
of  the  difference,  the  Ns  did  differ  from  group  to  group,  and  this 
resulted  in  differents  SE^s.  The  calculated  standard  errors,  t_  ratios, 
and  resulting  probabilities  are  given  in  Table  XXXII.  The  table  indi¬ 
cates  that  the  probability  of  obtaining  such  difference  between  means 
by  chance  in  all  three  cases  is  less  than  one  out  of  twenty.  It  can 
be  concluded  that  the  mean  Agreement  scores  in  schools  where  the 
principal  has  been  described  as  being  high  in  both  Consideration  and 
Structure  are  significantly  greater  than  in  those  schools  in  which  the 
principal  has  been  described  as  having  any  other  style  of  leader 
behavior. 

TABLE  XXXII 

SIGNIFICANCE  OF  DIFFERENCES  BETWEEN  MEAN  AGREEMENT 
SCORES  FOR  S+C+  GROUP  AND  AGREEMENT  MEANS  OF 
THREE  OTHER  GROUPS 


Mean  of  ShXJ-h  (n=16) 
minus  mean  of 

Niff. 

Be tween 

means 

seb 

t 

P 

C+S-  (N=13) 

7.98 

5-79 

2.11 

.02<Pb.05 

C-S-  (N=l6) 

9.00 

3-27 

2.75 

PC. 01 

C-S+  (N=ll) 

10.62 

3.62 

2.93 

PC. 01 

.  ;  ■  '  .  : 

■.  x  ■  .  ■ 

.  * 

-  .  .  -  >  .  •  i  m  5  . 

.  V  >  'X  .•  .  •  :  j  : 


1 


- 


•  > 


- 


" ■  J;  ■  !  1  -  r  \  ■  .  I  ■■  ..c:  J;-. 

■  ■  i':  \  >'  .  .  i  . 


. 


.  .  •  - .  .ji 


.  . 


.  .  I 


;  .  >  n 

:  i  a:  i  iri 


176 


Discussion 

It  would  seem  that  although  neither  the  style  of  leader  beha.vior 
nor  any  single  variable  associated  with  this  is  related  to  Consensus, 
the  extent  to  which  there  is  correspondence  between  the  average  expecta¬ 
tions  of  the  teachers  in  a  school  and  the  principal's  self -expectations 
is  related  to  his  style  of  leader  behavior.  If  there  is  a  causal 
relationship  here  it  has  far-reaching  implications  for  it  would  give  a 
definite  indication  that  a.  principal  is  not  at  the  mercy  of  the  expecta¬ 
tions  which  his  teachers  hold  for  him  but  that  he  can  in  a  very  real 
way  influence  their  expectations  and  bring  them  close  to  his  self¬ 
expectations.  That  is,  this  can  be  done  provided  that  he  can  exhibit 
that  style  of  leader  behavior  which  has  been  considered  as  desirable 
behavior  on  the  basis  of  other  criteria,,  namely,  high  Consideration 
and  Initiating  Structure  in  interaction.  A  step  in  the  direction  of 
investigating  possible  causal  relationships  is  taken  in  the  remaining 
tv/o  sections  related  to  the  analysis  of  data. 

VI.  STYLE  OF  LEADER  BEHAVIOR,  TENURE ,  AMD  CONSENSUS 

In  the  original  statement  of  the  problem  and  outline  of  the 
hypotheses  no  provision  was  made  for  considering  the  effect  of  tenure  - 
the  length  of  time  which  a  principal  had  been  associated  with  a 
particular  group  of  teachers  -  on  any  of  the  other  variables.  During 
the  analysis  of  the  data  in  the  previous  section  it  became  apparent 
that  it  would  be  desirable  to  investigate  the  relationship  of  tenure 
in  combination  with  the  variables  Structure,  Consideration,  and 
Ambivalence  to  Consensus  and  Agreement.  It  was  felt  that  such  an 


.  .  ■  .  ' 

;  •  -  >  ■  ■  '  i  .  . 

--  ■  ■ 

■  -  1  •  ■  :  : 

.  '  .  :  :  ■ 

:  •  ■'  £  £  j 

-  •  v  ■  ,  '  <’  •  i  :  '•  .  jj;  . 

:cv  •  :--o  Q.  i  ;  o  f  c  J.  j  «  cir.  ilciiif; 

^  ;•  ■  •  ■.  :r  J:i  ,  .  c  ■  .  :> 

■'  :  .... 

■  i  •  v  <  ::r  •  -  •  ,  '  J  , 

I  ?  ..  .  ■  .  •'  j 

■  .t  rJ:  o-ru&c  yd’’  ■ 

'•  v  o  .  .r . •:  v 

c .  '  I-..  ■  ■  .  ;:J ;  ■  J' 

•  »-  . 

•  >  i  l  *  » 

:A\ '  :  u'd’".;1';  V  .  ; 


177 


analysis  might  indicate  possible  causal  relationships  even  though  the 
direction  of  causality  would  probably  still  be  open  to  question.  This 
section  will  examine  the  relationship  of  Tenure  and  the  other  variables 
to  Consensus  while  the  following  section  will  concern  itself  with  their 
relationship  to  Agreement. 

Testing  Hypothesis  Six 

The  hypothesis  which  was  formulated  and  tested  is  as  follows: 

6.  Consensus  scores  will  be  related  to  the  interrelationship 
of  leader  behavior  variables  and  Tenure. 

The  first  step  in  testing  the  hypothesis  was  to  divide  the  principals 

into  two  groups  on  the  basis  of  an  index  of  average  tenure  with  the 

present  staff  which  wa,s  obtained  by  calculating  the  arithmetic  mean 

of  the  number  of  years  which  eaoh  teacher  in  the  school  had  been 

working  with  the  present  principal.  The  mean  Consensus  scores  were 

then  calculated  for  four  groups  obtained  when  the  two  groups  classified 

on  the  basis  of  Tenure  were  further  divided  on  the  basis  of  Structure 

scores.  In  the  same  way  Consensus  scores  were  also  calculated  for 

four  groups  established  when  principals  were  classified  according  to 

Tenure  and  Consideration  and  also  for  classifications  based  on  Tenure 

and  Ambivalence.  The  results  of  this  analysis  were  presented  in 

Table  XXXIII.  It  should  be  kept  in  mind  tha.t  the  analysis  in  each  of 

the  three  sections  of  the  table  is  based  on  the  same  data. 

An  examination  of  the  table  reveals  that  for  principals  who  are 
high  in  Structure,  Consensus  increases  with  Tenure  while  it  decreases 
with  Tenure  for  those  who  are  low  in  Structure.  The  same  observation 


' 


(. 


Ij  L  ' 


.. 


'C  0.1  ■  ‘ 


-  £.;  : 


!  ’ 


.)  ■  h 


1.  x 


,•  -j:  •. 


X1( 


j; 


..  •■  . : 


■  u  0  i.‘. 


£ 


o'  :  ;< 


c  .  -  S.  ■■  i'o 


c  ■ '  ;  • 


e.:;  .  •  •  .  : 

.  •  J:c  i 


-  ,  •  -  ■■  ■ 


.1 1 


%  ■  ■ 


; 


.... 


TO: 


c 


•  - 


1?8 


can  be  made  with  respect  to  Ambivalence,  but  the  opposite  effect  is 
true  in  the  case  of  Consideration.  An  analysis  of  variance  was  carried 
out  to  determine  the  significance  of  these  differences  in  each  case. 


TABLE  XXXIII 

MEAN  CONSENSUS  SCORES  BY  STYLE  OF  LEADER  BEHAVIOR 

AND  TENURE 


Leader 

Behavior 

High 

Tenure 

Low 

Tenure 

Difference 

High 

Structure 

52.00 

47.87 

4.15 

Low 

Structure 

48.44 

52.62 

-4. 18 

High 

Consideration 

49-58 

55.12 

-5.74 

Low 

Consideration 

50.47 

46.OO 

4.47 

High 

Ambivalence 

52.28 

49-78 

2.50 

Low 

Ambivalence 

47.64 

50.56 

-2.72 

The  F  ratios  in  Table  XXXIV  show  that  none  of  these  were  significant 
in  spite  of  what  appear  to  be  sizable  differences  in  the  means  accord¬ 
ing  to  Table  XXXIV.  Two  of  the  three  differences,  although  not 
significant,  were  in  the  direction  one  might  have  hypothesized  on 
the  basis  of  the  theory  that  high  Structure,  Consideration,  and 
Ambivalence  taken  individually  would  be  related  to  a  high  degree  of 


;■? 


, 


-v‘. .. j  ><■  i'  ■ 


.  .  : 


>  ■ 


; 


'  , 


•  • 
.  .1 


■  k 


. 


)■ 


C 


Consensus. 


179 


The  relationships  were  investigated  further  by  calculating  the 
pro duct -moment  correlations  coefficients  between  Tenure  and  Consensus 
on  the  basis  of  each  of  six  classifications;  that  is,  the  high  and  low 
groups  in  each  of  Structure,  Consideration,  and  Ambivalence. 

TABLE  XXXIV 

F  RATIOS  OBTAINED  THROUGH  ANALYSIS  OF  VARIANCE  OF 
CONSENSUS  SCORES  GROUPED  BY  PRINCIPALS’ 

STILE  OF  LEADER  BEHAVIOR  AND  TENURE 


Tenure 

and 

F 

df 

F  required 
for 

significance 

Structure 

0.81 

Consideration 

1.21 

3/52 

2.80 

Ambivalence 

0.50 

The  correlation  coefficients  together  with  differences  between  the 
high  and  low  groups  are  presented  in  Table  XXXV.  Although  none  of  the 
coefficients  or  the  differences  are  significant  at  the  .05  level,  the 
difference  in  the  case  of  Structure  does  approach  significance. 

Discussion 

Both  when  means  were  calculated  and  when  the  correlational 
approach  was  used,  the  analysis  involving  Structure  revealed  a  relation¬ 
ship  which  could  be  explained  in  these  terms:  The  longer  that 
principals  who  are  high  in  Structure  associate  with  a  particular 
group  of  teachers,  the  more  Consensus  there  is  within  the  group  of 


.. 


o 


r 


:  ' 


'1 


:: 


« 


'■ 


:rp 


.  . . 


I 


c 


f  '  •  itii  ! 


O' 


V  3  I  ■ 

I'  I  ■  V.'' .  ’ 


- 


V  .  . 


•  •  ;•  r  * 


;r,i~  .  ; 


'  < 


/:  .  • 


180 


teachers  on  role  expectations  for  the  principal.  This  does  not  hold  if 
the  principal  is  low  in  Structure  for  then  the  amount  of  Consensus  de¬ 
creases  as  the  length  of  association  increases.  When  the  means  were 
calculated  a  similar  explanation  might  have  been  made  for  the 
relationships  involving  the  .Ambivalence  variable;  however,  the 
correlation  coefficients  do  not  support  such  an  explanation  to  as 
large  an  extent  as  in  the  case  of  Structure. 

TABLE  XXXV 

LINEAR  CORRELATION  OP  TENURE  WITH  CONSENSUS  FOR  BOTH 
HIGH  AND  LOW  STYLES  OF  LEADER  BEHAVIOR  WITH 
SIGNIFICANCE  OF  DIFFERENCES 


Leader 

Behavior 

Dimension 

High 

y 

r 

Low 

y 

Difference 
r  (z) 

SE 

D 

Diff 

z 

SE^ 

D 

Structure 

0.18 

-0.52 

0.50 

(.51) 

1.80 

(N=27  ) 

(N=29) 

Consideration 

-0.21 

-0.01 

0.20 

(.20) 

.285 

0.71 

(N=29) 

(N=2? ) 

Ambivalence 

-0.05 

-0.14 

0.09 

(.09) 

0.52 

(N=28) 

(N=28) 

Required  for 

significance  at  . 

,05  level 

1.96 

One  would  hardly  have  expected  to  find  the  observed  relation¬ 
ships  involving  Consideration,  Consensus,  and  Tenure  on  the  basis  of 
reasoning  similar  to  that  above.  One  possible  explanation  for  the 
observed  results  would  be  to  suggest  that  whatever  effect  Consideration 
does  have  on  Consensus  is  not  so  much  a  function  of  time  as  it  is  in 
the  relationship  between  Structure  and  Consensus.  Perhaps  it  may  even 


*  ^  ^ 


•  • 

c  .  .. '  ..  ■ '  . . 

CV'.C  .hi. 


...  - 

'  .;r>  \ 


•v  •(!  -  j;:  :,.r,  ' 


i  >  • .  ■  '  f  i : 


. 


.1  ■  .) 


- :  . . 


turn  ' 


■■  J- r.  ,  r-.  '  '  v  • 


•  V  to  />'  L  O  i 
i  \ .  ■  :  J  ■  J  ■ 


181 


be  sound  to  suggest  that  high  Consideration  does  have  a  negative  effect 
over  a  period  of  time  on  Consensus  and  that  it  is  possible  for  leaders 
to  go  too  far  in  this  direction.  Some  of  the  earlier  analysis  did 
suggest  that  Consideration  may  be  related  to  Consensus  and  perhaps  one 
reason  why  this  showed  up  more  strongly  than  did  any  relationship  in¬ 
volving  Structure  was  due  to  this  difference  in  influence  of  Tenure. 

VII.  STYLE  OF  LEADER  BEHAVIOR,  TENURE,  AND  AGREEMENT 
Testing  Hypothesis  Seven 

The  same  reasons  as  were  put  forth  for  the  analysis  involving 
Consensus  resulted  in  investigating  the  effect  of  Tenure  on  the  rela¬ 
tionship  between  various  aspects  of  leader  behavior  and  Agreement. 

Groups  of  principals  were  again  formed  to  create  categories  of  leader 
behavior  variables  and  Tenure  in  order  to  test  the  following  hypothesis: 

7.  The  relationships  between  Agreement  and  leader  behavior 
variables  will  be  influenced  by  Tenure. 

Table  XXXVI  shows  mean  Agreement  scores  for  each  of  the  classifi¬ 
cations  and  also  shows  the  differences  between  means.  With  the 
exception  of  the  low  Consideration  group,  these  differences  are  in 
the  direction  one  might  have  anticipated  on  the  grounds  that  those  in 
the  higher  classification  would  be  associated  with  higher  degrees  of 
Agreement  the  longer  that  they  were  associated  with  a  particular  group 
of  teachers.  The  data  from  which  the  mean  scores  were  calculated  were 
also  subjected  to  analyses  of  variance  in  order  to  check  on  the  sig¬ 
nificance  of  observed  differences.  The  resulting  F  ratios,  which  are 
given  in  Table  XXXVII,  indicated  that  none  of  the  differences  were 


■  '  vv  .  .■  ;  ,::;i  ;i  .7  y  •  r:  o;i' 

-  ./  v  ■  '  ■■ .  ;  :  •(  •  •  v 


i: 

;  .  .\  •  ■'  A 

< 1  i  . . 

'  .  ■  J  ■  .t  •  • / 

j  •  ... 

.  V-  '  ./•  >■?•-  .!  :  .•  ;  .v.  •  •  • .  -  •  .  :.  ‘V 

oj  7/  jj  >  :•  •  'i.  :- 7 

.  ;  .77  .77  ■  .77'  Y  7  7  ‘  r: .  ■  ; 

. 

t  .■  :■  7  y  '  ;i ;  c 7:  7v 

.  r.  '  ■  ■  7  1  h.i  j:-  v  ■.  .7/  >ir 


7-  7  .  :  .  ■ 

.m 

,  ■, i o  -. 

■.  ■  •  :  : ■  i  ..r  ;  -  ■  J.  .7  .'  .17 :  A 

■ 

l  i  «>rl  :  .  ■  '  1 

7  ?  t  •  ■.  7  •  •  :  ■  7'  :  r:  \y  c 

■  ■  '  .77  .  .  .7  •  .  . 

'.I  .  J.'.  ’  i  Y  .  ■  V 


■: 

:  ■  7  1  ...  .  ■  ■  A  '.7 

■  >  :  ■  ■■ 

7  •  '  «  '•  -  :  V.o  '  .'  •; .  ••  7  -.Jo'..'  <  :  ■ 

A.  .  . 


182 


significant . 

In  order  to  investigate  the  relationships  further,  product- 
moment  correlation  coefficients  were  calculated  between  Tenure  and 


TABLE  XXXVI 

ME AN  AGREEMENT  SCORES  BY  STYLE  OP  LEADER  BEHAVIOR 

AND  TENURE 


Leader 

Behavior 

High 

Tenure 

Low 

Tenure 

Difference 

High 

Struc ture 

55.00 

49.80 

5.20 

Low 

Structure 

46.69 

49-38 

-2.69 

High 

Consideration 

54.00 

51.94 

2.06 

Low 

Consideration 

47.00 

46.5O 

0.50 

High 

Ambivalence 

52.64 

49.50 

3.14 

Low 

Ambivalence 

47.86 

49.71 

-1.85 

Agreement  for  each  of  the  six  groups.  Table  XXXVIII  shows  that  both 
for  high  and  low  Structure  groups,  Agreement  and  Tenure  correlated 
relatively  highly  with  each  other.  The  correlation  between  the  two 
variables  for  the  high  Structure  group  was  0.50  which  is  significant 
beyond  the  .01  level,  while  for  the  low  Structure  group  the  coeffi¬ 
cient  was  -O.3I  which  fell  short  of  significance  at  the  .05  level. 


’  ; 


. 


V 


: 


. 


. 


:  f  1  ■ 


. 


. 


. 


■  ;•  . 


. 


183 


The  difference  between  the  two  was  highly  significant.  The  relation¬ 
ships  were  not  as  marked  when  Consideration  was  involved  but  were  in 
the  same  direction  as  in  the  case  of  Structure;  the  da,t a,  for  Ambivalence 
revealed  no  trend  about  which  one  might  speculate. 


TABLE  XXXVII 


P  RATIOS  OBTAINED  THROUGH  ANALYSIS  OF  VARIANCE  OP 
AGREEMENT  SCORES  GROUPED  BY  PRINCIPALS'  STYLE 
OP  LEADER  BEHAVIOR  AND  TENURE 


Tenure 

and 

P 

df 

P  required 
for 

significance 

Structure 

1.68 

Consideration 

1.95 

3/52 

2.80 

Ambivalence 

0.55 

TABLE  XXXVIII 

LINEAR  CORRELATION  OP  TENURE  AND  AGREEMENT  FOR  BOTH 
HIGH  AND  LOW  STYLES  OP  LEADER  BEHAVIOR  WITH 
SIGNIFICANCE  OP  DIFFERENCES 


Leader 

r 

Difference 

SE-p,  Diff 

D  z 

Behavior 

Dimension 

High 

y 

Low 

y 

(z) 

r 

SJSD 

* 

(.87) 

* 

Structure 

0.50 

(N-27) 

-0.31 

(N=29) 

0.81 

3. 08 

Consideration 

0.13 

(11=29) 

-0.14 

(H=27) 

0.27 

(.27) 

0.283  O.95 

Ambivalence 

-0.01 

(1=28) 

-0.14 

(N=28) 

0.13 

(•13) 

O.46 

■^Significant  at  .01  level 


. 


•  ,  •.  ;  -r  ‘ 

.  v  .  .  •  id  .'.  ' 

;  •  '.  .  ■  ■ 

.  i  o  ... :  ,::r  ;  :  o  ■ ■  •.  -  :n; 


■ .  .  ’ :  .  ::  ..  ■.. 

» 


t  X  . 


.  . 

. 


:  .  ;  ;  :  .  :  .  :  .  :  .• 


■  • '  '  ■■  v  - 
‘  : 

:  '  ."  . 


. 

.  ,  .  a .  ■  :  : 

.  .  ..  ,  .  a  .  i  1  A.  .  c 

.  .  :  ■  .  :  :  :  :  : 


■  ..  ;  l. 


•  \  '  •  . 


. 


’ '  *  - 

.  - 

. 


. 


184 


Di  sous si on 

The  analysis  used  in  this  section  tends  to  support  the  idea  that 
whether  a  principal  is  high  or  low  in  Structure  will,  over  a  period  of 
time  in  one  school,  influence  the  amount  of  Agreement  in  that  school. 
The  type  of  relationship  which  seems  evident  is  the  same  as  that  out¬ 
lined  in  the  previous  section.  The  relative  effect  of  Consideration  is 
difficult  to  determine  for  the  results  of  the  correlational  approach 
conflict  to  some  extent  with  that  involving  the  calculation  of  means. 

In  the  same  way,  the  part  that  Ambivalence  might  play  is  difficult  to 
speculate  upon  on  the  basis  of  the  previous  analysis. 

The  relatively  small  N  in  the  sample  precluded  combining  vari¬ 
ables  such  as  Structure  and  Consideration.  For  future  study  it  would 
seem  reasonable  to  hypothesize  that  there  would  be  marked  differences 
between  Agreement  in  schools  characterized  by  high  Tenure  and  high 
Consideration  and  Structure  and  those  characterized  by  high  Tenure  and 
low  Consideration  and  Structure.  Proper  test  of  siich  relationships 
would  require  samples  considerably  larger  than  that  involved  in  this 
study. 

VIII.  SUMMARY 

This  chapter  has  included  restatements  of  the  hypotheses  which 
were  tested,  the  methods  used  in  testing  these  hypotheses,  and  the 
results  of  the  analysis.  The  intercorrelations  of  five  variables  are 
presented  in  Table  XXXIX;  these  will  serve  as  a,  framework  around  which 
the  summary  will  be  built. 


c 


.  ■'  i. 


■  ■ ::  c '  X1'  / 


. 


u  . ’ r  (V  ■  ) 


'■ 


crj' 


■- 


« 


0 


JC  vi 


i’  .  -  x  ■  ■  x  c  •'  , 


dS  i 

'  v  xii  ■  .  >  .  '  ;  xc 


c  -  J  x  x.  '  x 


.  :•  •  O’  :  'i  ; 


ox 


- 


. 

.•  ,  u 


• ,  X  •  .  ’  .  ’  ' 


•  v..;  i  .X'  x  vxi: 


.  x. 


•  •  :  :  X  . 

.  J'  . ’  :  x 

: 

: 


185 

■Ambivalence ,  Structure ,  Consideration 

The  analysis  which  was  carried  out  failed  to  reveal  any 
significant  relationships  between  Ambivalence  and  either  Structure 
or  Consideration.  Table  XXXIX  shows  that  the  linear  correlations 
between  va  riables  did  not  differ  from  zero.  However,  when  the  degree 
of  Consensus  on  expectations  among  teachers  in  a,  school  was  taken  into 
account,  there  did  appear  to  be  a  relationship.  Indications  were  that 
principals  classed  as  being  high  in  Ambivalence  in  situations  which 
did  not  call  for  this  were  described  by  their  teachers  as  being  low 
in  Structure  while  principals  who  were  high  in  Ambivalence  in  situa¬ 
tions  for  which  this  wan  appropriate,  namely  low  Consensus  schools, 
were  described  as  being  high  in  Structure. 


TABLE  XXXIX 

CORRELATION  MATRIX  FOR  FIVE  VARIABLES 


Structure 

Ambivalence 

Consensus 

Agreement 

Consideration 

0.21 

-0.06 

-0.06 

0.24 

Structure 

-0.07 

-0.15 

* 

0.28 

Ambivalence 

0.05 

0.20 

Consensus 

* 

0.52 

^Significant  at  .05  level. 


Ambivalence ,  Agreement ,  Consensus 

The  relationship  between  Ambivalence  and  Agreement  was  not 
statistically  significant  but  was  in  the  hypothesized  direction. 


; 


/ 


■ 


: 


-  '  '  ' 


'  . .  .  .'.  '■ 


•  '  J7  !;  ■  . .  ; 


■ 


c 


.  u  ■  ' 


186 


The  correlation  coefficient  of  O.OJ  between  Ambivalence  and  Consensus 
is  highly  insignificant. 

Structure,  Consideration,  Agreement 

Table  XXXIX  shows  that  the  relationship  of  both  Consideration 
and  Structure  to  Agreement  were  relatively  high  and  that  in  the  latter 
case  it  was  statistically  significant.  VJhen  the  quadrant  approach  to 
style  of  leader  behavior  was  used,  the  C+S+  style  was  found  to  be 
associated  with  significantly  higher  Agreement  scores  than  the  other 
three  styles  of  leader  behavior.  The  relationship  was  further  sup¬ 
ported  when  the  Tenure  variable  was  introduced.  For  principals  who 
we re  described  as  being  high  in  Structure,  Agreement  increased  with 
Tenure  but  for  those  who  were  low  in  Structure,  Agreement  decreased 
as  Tenure  increased. 

5 true ture ,  Consideration,  Consensus 

Individual  relationships  between  Structure,  Consideration  and 
Consensus  did  not  approach  significance  but  when  styles  of  leader 
behavior  were  considered,  there  was  a  tendency  toward  significant 
differences.  Tenure  was  found  to  be  positively  correlated  with 
Consensus  for  principals  described  as  being  high  in  Structure,  but 
the  same  relationship  was  negative  for  principals  described  as  being 
low  in  Structure. 


Discussion 

A  problem  in  drawing  inferences  from  analysis  such  as  that 
summarized  above  results  from  effects  due  to  the  interrelationships 


'  '  . 

--  *  -  .  .  ■ 

■'  ■'  '  , '  ' : 

y:'v: 

■:  .v  k.:. 

•  C  .  .  V.  '  ■  J 

V  ■  ■  .  •  : 

.  •  .  <  :  '  ••  .•  .  • :  .  '  /:  .  ■  •  ■  .>  :  .  •  .  .■  ■ 


'  ?  '.v.  \  o ■.>:  . 

; 

■  V .  . 

■  J  •  •  ■  ..  :>•  ■>'  .  . 

•  '  :  •'  •  •  ' .  •  .  •  c  •.! 

.  .  ••  J. 

'  •  '  :  •  '•/  '  )■  ■  k 

■  ■  '  '  >'  :  ’  :  v.  '  /'  •  .  >  .c 


187 


of  the  variables.  In  order  to  explore  this  effect  a  number  of  partial 
rs  were  calculated  but  it  was  found  that  these  did  not  differ  greatly 
from  the  coefficients  in  Table  XXXIX  nor  did  they  serve  to  aid  in  the 
interpretation  of  the  results. 


.  •  ■'  "■  ■  >;  •  .  ■  c.[  •  ...  .  ■  ■... 

•  ...  ,r  ■  ..  .' 

•  .  ■ ;/  ■.  •  ■ 

- 

AS 


' 


CHAPTER  X 


SUMMARY  AND  CONCLUSIONS 

This  study  was  designed  for  the  purpose  of  exploring  the 
possibility  of  refining  some  of  the  methodology  used  in  the  study 
of  role  conflict  in  general  and  conflict  in  the  role  of  Alberta, 
school  principals  in  particular.  It  was  hoped  that  this  exploratory 
study  would  suggest  how  further  progress  might  he  ma.de  in  helping 
practicing  administrators  to  deal  with  this  problem  and  that  it 
would  also  suggest  improved  procedures  for  analyzing  the  effect 
which  principals  have  on  the  expectations  for  their  role  and  the 
effect  which  role  conflicts  have  on  them.  In  this  final  chapter 
the  main  features  of  the  investigation  are  summarized  and  some 
general  implications  for  both  the  student  of  educational  administ¬ 
ration  and  the  practicing  administrator  are  proposed. 

I.  SUMMARY  OF  THE  STUDY 

An  analysis  of  the  theory  and  the  methodology  of  scale 
analysis  suggested  that  this  might  be  a  useful  approach  for  the 
study  of  conflicting  expectations.  For  one  reason,  it  offers  a 
method  whereby  quantitative  variables  are  derived  from  data  which 
are  essentially  qualitative.  This  appealed  to  the  investigator  for 
such  a  quantitative  variable  is  of  obvious  value  both  to  the 
administrator  who  desires  to  know  more  about'  the  structure  of 
the  expectations  held  for  him  and  to  the  researcher  who  might 


. 


V 


•  \  .o'  •  ;  J;.  .  X  ■  y-  ;  ! 

>•  i  ■  V  ■  x;  • 

-  '  J  oJL  :  Xrr  1 ; 

;  -  -  -  '  .  i '  -  -  :  ; 

■  •  ’  •  -  t.  >•.?■.  -  o'-  ...  ' .  "  .  -.•••■  i  r  T,-  -  •  ;  v; 

:  ■  ‘  J'  ■  :  .';c • 

'■  •  '  cv  ' .  '  ■ ...  j.  r  J  -  ■.  • 

>  ‘  '  ; 

'  •  •  •  l'<:  '  -  -  •  •  '  V  '  ..  ;  ; 

'  •'  v  .1  ••  V.  •  ;.i  :  ;.J' 

...  v  .  .  .  .  r j.l* 

‘ -  cold- On*  - 

;  *  X 

»  ■■  >■  r'i or  ",  J-  .  :  -  ,  .  •  . ; 

•  ' .  ;  '  •  •  • 

’•  c  -  ■ 

-  '•  •  '  ■  p  .  =.v;  v  ■  J' 

•  '  '  »  ....  i'.  .. 

'y  j  '  '  :  -  V'  ■  .i ■.  ■  ■  ■  ■  ,i>-  •. 

' 

. 


189 

wish  to  use  it  in  testing  hypotheses.  It  was  thought  that  such  a  study 
might  yield  results  of  value  to  those  concerned  with  administrative 
phenomena  and  to  those  interested  in  the  analysis  of  role  conflict  in 
a  variety  of  situations. 

A  review  of  the  available  literature  indicated  that  scale 
analysis  had  been  used  with  considerable  success  in  a  variety  of 
problem  areas.  Although  it  is  still  subjected  to  considerable  criti¬ 
cism  on  both  theoretical  and  practical  grounds,  the  results  of  the 
empirical  application  of  the  method  suggested  that  it  had  proven  it¬ 
self  to  be  of  value  as  a  research  tool.  For  this  reason  it  was 
considered  suitable  for  the  proposed  study  of  role  conflict  in  the 
pr  inc  ip  al  sh  ip . 

The  essential  characteristic  of  scale  analysis  is  that  it 
provides  a  test  of  the  uni dimensionality  of  a  series  of  items  based 
on  the  theory  that  when  such  a  scale  does  exist,  it  is  possible  to 
reproduce  individual  responses  from  the  quantitative  variable  which 
has  been  derived  to  characterize  the  respondent’s  attitude  on  the  set 
of  items.  In  actual  use  scale  analysis  has  proven  itself  to  be  of 
value  in  circumstances  where  it  wan  not  possible  to  achieve  the 
necessary  degree  of  reproducibility.  However,  the  existence  of  a 
unidimensional  scale  does  provide  the  researcher  with  a  quantitative 
variable  that  is  meaningful  and  also  provides  him  with  information 
about  the  structure  of  the  particular  area  he  is  investigating. 

The  most  important  step  in  the  use  of  scale  analysis  takes 
place  long  before  any  analysis  is  actually  done;  that  is,  the 
definition  of  the  universe  of  content  which  will  be  tested  for 


‘ 


i  .  j;  t 

:  •  '  r  ■.  •  r-  o  \r;  .•  .nU. 

'  i  ‘  -L J  ■ ■  i  .j  .  ! 

•'  ■  ...  •_  ;  j  .  ;  •  >  :  ;  £  ; 

■ 

•  :  ■  ■  '  r  :-r 

.  .’.  .  ' 

'  •  "  :■  i.  ’  .•  j; 

>•  •  ' 

. :  . .  .  :  :  ■  c  :  J'  J  j' 

.  ..  ••  •  '  '  ■  .  •  I;  .  ;  :y.o\  <  •  •  . 

1  :  '  ^ 

J 

.  LO  •  :.  ;  .  J  :  ■  ■;  ,/  y 

y  i  •  ;  : 

•  ■  :  ■  o  ■  .'  !  ■ ;  ■ 

'.  '  '.  <  J  .  j 

>  '  '  '.  5  •  -  olJ 

.  .  •  '  '  -  e.  r 

s'.  '  I  '  I: 

.  ’  '  ■  '  '  :  ■■  '  .  J  ' .  jptOxJ’/rnn  r  - 


190 


unidimensionality.  This  is  so  important  because  the  a  priori  definition 
will,  to  a  large  extent,  determine  the  success  of  the  scale  analysis. 

If  the  area  has  been  defined  inappropriately  it  will  be  extremely 
difficult  and  perhaps  impossible  to  extricate  a  quantitative  variable 
with  which  to  describe  the  responses.  The  first  step  is  an  extremely 
hazardous  one  for  the  individual  who  is  not  aware  of  what  an  extremely 
rigorous  definition  of  a.  scale  is  involved  in  the  use  of  the  Guttman 
model  (l,  4)* 

In  the  present  study  the  dimensions  or  role  conflict  proposed  by 
Seeman  (2,  3)  were  selected  and  subjected  to  tests  for  unidimensionality 
using  the  scale  analysis  procedures.  Seeman  proposed  that  leaders  in 
organizations  were  subjected  to  role  conflicts  in  particular  areas  be¬ 
cause  there  were  value  conflicts  in  the  culture  in  these  very  same  areas. 
Four  such  possible  conflict  areas  surrounded  values  concerning  hierarchy 
as  opposed  to  equality,  dependence  as  opposed  to  independence,  univer- 
salism  versus  particularism,  and  means  as  opposed  to  ends.  Items  were 
constructed  which 'were  considered  to  be  valid  for  one  of  the  four 
hypothesized  dimensions.  In  general  agreement  or  disagreement  with  an 
item  involved  a  choice  in  the  direction  of  one  or  the  other  polar 
extreme  in  the  areas  mentioned.  The  items  were  pretested  using  samples 
of  undergraduate  education  students,  graduate  students,  and  teachers 
and  principals  in  actual  teaching  situations. 

Responses  were  obtained  from  7&5  teachers  and  sixty  principals 
located  in  sixty  centralized  non-urban  schools  in  Alberta..  A  sample 
of  one  hundred  responses  drawn  at  random  from  the  completed  question- 


191 

naires  was  tested  for  scalability  using  the  scalogram  board  technique. 
The  analysis  revealed  that  the  items  were  not  scalable  in  the  way  that 
had  been  anticipated.  That  is,  the  hypothesis  of  uni dimensionality  in 
each  of  the  four  areas  of  expectations  was  not  supported.  By  means  of 
a  careful  study  of  the  items,  the  investigator  was  able  to  define  sub- 
areas  which  approximated  scales  much  more  closely  than  did  those 
originally  tested  for  unidimensionality. 

Two  sets  of  items  were  classed  as  quasi  scales  from  among  those 
which  had  been  constructed  for  the  hierarchy  versus  equality  category. 
One  set  consisted  of  some  very  specific  items  concerning  the  relation¬ 
ship  of  principals  to  teachers  in  various  aspects  of  work  in  the  school 
the  second  set  consisted  of  much  more  general  items  concerning  social 
distance  between  teachers  and  principals  generally.  Clear  definitions 
of  these  subareas  were  limited  by  the  type  of  items  which  had  been 
constructed  originally  and  thus  left  much  to  be  desired. 

The  authority  items  were  also  reduced  to  more  scalable  subareas. 
One  of  these  consisted  of  items  dealing  with  the  extent  to  which 
teachers  should  be  dependent  upon  the  principal  for  assistance  with 
their  discipline  problems.  The  second  set  consisted  of  items  con¬ 
cerning  the  extent  to  which  the  principal  should  give  the  teachers 
freedom  in  various  aspects  of  their  work.  It  was  also  possible  to 
define  two  subareas  in  the  items  related  to  the  universalism  versus 
particularism  dilemma.  The  first  set  consisted  of  items  concerned 
with  the  extent  to  which  a  principal  could  have  closer  social 
relationships  with  some  teachers  than  with  others  while  the  second 


J-  '■  • 


' 

. 


;  -  *  .  .  >  '  .  ■ 


-  '  '  ;  ■  '•  •'  ;  J  • 

-  -  il  X  ' 

-  •• ■  ''  /:  .  >•  /  J-  •  I.  •/;  : 

-  >■  '  '  "  •  '  ’■■■■■  ' 

*  •  '  •  1  •  •  •  :  '  v  i  .  •  . 

•  ;  -CC-JC  ■  ■}  .  j;  ,  J  - 


■  J:t  ’  >  j; 


■  '  ;  ;  i  ..  •  r 

:  ;  •  '  •  •  ' 

:  '  -  •  '  r.:  r.  ■:  ;\v,i  J-  ‘.o , 

w*t  ..  ■  ;  ; 

*  -  t  '  ;  V 

>•  >•  v 

1  . . '  ■  >'  '  1  '  :  •  i  '  C  >  '  .  J  ,  c 


■  &  i  ; 


: 

o  '  •  .  •  '  j-:  ■’<  ;  •  •;  ;•  ■ 


-•>'  .  ■  rt;i 


192 

set  concerned  the  extent  to  which  he  should  take  individual  differences 
into  account  when  assigning  duties  to  teachers.  The  set  of  items  dealing 
with  the  means -ends  dilemma  proved  to  he  quite  unsatisfactory;  one  quasi 
scale  was  formed  from  these. 

Instead  of  only  four  areas  or  dimensions  as  had  been  hypothesized 
originally,  there  had  now  been  established  seven  dimensions  on  which  it 
was  possible  to  define  ea,ch  teacher's  position  with  a  quantitative  vari¬ 
able.  Although  none  of  the  sets  of  items  satisfied  the  criteria  for  a. 
scale,  there  seemed  to  be  sufficient  grounds  for  considering  each  of 
these  as  a  quasi  scale.  The  results  of  the  scale  analysis  were  used  to 
construct  a  key  which  was  used  in  scoring  all  of  the  questionnaires. 

From  the  scores  obtained  by  individual  teachers  it  was  possible  to 
derive  indices  of  the  amount  of  consensus  which  existed  among  the 
teachers  in  a  school  and  also  of  the  extent  of  teacher -principal  agree¬ 
ment  concerning  expectations  for  the  role  of  the  principal.  These 
indices  were  required  for  testing  a  number  of  hypotheses  which  had 
been  formulated.  The  interrelationships  of  the  scores  on  the  variovis 
scales  were  not  investigated  since  this  was  not  one  of  the  purposes  of 
the  study. 

Reference  has  now  been  made  to  the  hypotheses  which  were  tested 
utilizing  the  results  of  the  scale  analysis  and  the  scores  and  indices 
derived  from  the  questionnaires.  It  had  been  theorized  that  even 
though  all  principals  were  subject  to  role  conflict,  some  principals 
would  be  able  to  deal  with  role  conflict  more  effectively  than  others. 

It  seemed  logical  to  hypothesize,  partly  on  the  basis  of  some  earlier 


.  ;  i  •  J:. . 


■  o  ••  ,:■■■■! 


■  .  •  .  .r  •  '  .. 

.  .  'r  • 

»  ••  .  jf' :  •  :  V 

'v':  d  ■  v  l 

•  l.':  .  /; 

■'  :.i  ■  o i  e  ±  <  .  ’  ■  c  i  k  .  '  f  sldi: 

‘  ■’  '  rr.  '  d  :  .•  . 

o  >  •  ;  -  .'xd\ :  &  ■  '  , :  • :  - 

-  •  •  J\  »  '  •  j; . 

■  L  '■  u  O;  .  SJ  I  ;  £  ■  .  ft  r  d 

d  ■■■!  ■  ••  '  dd  ~  -.  d .  d:  .1:  ;  d  ,  c  :  ;d  -  \d 

5j  d  ■  '■  i::  •  d  •  noo  1  fm  ■  oi:  v± 


■  '  -CO/  XI  ~i:  :  ■  •  dx©  .  5  lo  :  \ 

jj  S.  >i  o  ti  ■’  :  »C  d 

■:  d  d  .  •:'■■■'  /  \  V.  •'  "  ■  ■  •  - 

'  x  "•  ■  ,  o  • .  •  •  ’  \d  .  '  -  .  d  •  •: 

.  -  ■.  ■  ••  d  d  .  '  -  •  ■  d  •  ■  j  ■  -  'd  d  •  • 

■ 

.  ■ '  d  ‘  -  doc;  ’t  sham  /to© 

'  f  -  V  u;  •  J:  •  7-,  .  c  •  '&  .  d  •  '  d 

o  :  ■  1 

■  .i .  (  :  ....  r.'.  ,  •',! 

*  .V  ■  ■  .  r  oi  ...  £  . 


193 


studies,  that  differentials  in  the  intensity  with  which  principals 
held  their  self -expectations  would  he  related  to  the  way  in  which 
teachers  described  them  on  the  Leader  Behavior  Description  Question¬ 
naire.  The  variable  of  Ambivalence  was  defined  in  terms  of  the 
intensity  with  which  self -expectations  were  held;  an  index  of  this 
was  obtained  from  a  self-report  instrument  in  which  principals 
indicated  how  difficult  it  had  been  for  them  to  decide  on  their 
responses  concerning  expectations  for  the  role  of  the  principal. 

These  responses  were  subjected  to  scale  analysis  from  which  it  wan 
possible  to  derive  a  quantitative  variable  with  which  to  characterize 
the  Ambivalence  of  individual  principals. 

Subsequent  tests  of  the  hypotheses  failed  to  reveal  any  con¬ 
sistent  relationship  of  this  variable  to  either  of  the  two  leader 
behavior  variables  Consideration  and  Initiating  Structure .  Nor  was 
Ambivalence  found  to  be  related  significantly  either  to  the  extent  to 
which  the  expectations  of  the  teachers  in  a  school  corresponded  to 
the  principal’s  self -expectations  or  to  the  extent  to  which  there 
was  consensus  among  the  teachers  themselves. 

It  was  found,  however,  that  when  schools  were  classified  as 
being  either  high  or  low  in  Consensus,  Ambivalence  was  positively 
related  to  Structure  in  low  Consensus  schools  but  was  related 
negatively  to  Structure  in  the  high  Consensus  schools.  This  suggests 
that  Ambivalence  in  a  situation  which  would  seem  to  call  for  it  has 
a  different  effect  on  the  way  that  teachers  describe  the  behavior  of 
a  principal  than  it  has  in  situations  which  do  not  demand  that  the 


-  •  .  ..  • 

;•  - 

-  '  -  r  ;r  ■.  .'.-'k  •••  •:  .•  i 

•  .  ■  .:  '  '  :  :  .  ■  ■ 

'  ':rl  ■  .  •/• 

-  '  l  ■  •  -  *  x  •  ••  EJ  •  '..  • 

•- 

•  :  ; 

■  ;  --  ■  '  i  '  • 

■  '  :  •'  .V  :v 

•  ’  •  .  ■  ’  .V;  '  •  ■  ,t 

--noo  \  _ 

:  ■  ■’  ■  v  Jr. 

o  .  ■  ^ :  '■  t  ‘  _ .  .  :  . . ; '  -•  •  • 

:  "  .  ' '  '  ;/  r.  ?' 

■"  I  •  :  ‘  ' '  ■  '  •  •  ...  •  </  i  ■:  -  ..  C.. :  • ' :  ■  ■  ;  V'  j':  £X 

■ ;  1  1  ■ 

’  •'  •  .  '  „  .  '  ;  ■  x 

:  c  :  i';  i.-.'j 

-  '  ■:  .1  c  c-n 

;  '>&  •  ■.  ' j  •  •,  1  - v  'J- 

. ■  '  ■  .  • 


194 


principal  be  high  in  Ambivalence.  Obviously,  the  situation  is 
important  in  determining  whether  or  not  Ambivalence  on  the  part  of 
the  leader  is  appropriate  for  him  in  terms  of  desirable  leader 
behavior. 

Other  hypotheses  concerned  the  relationship  of  the  leader 
behavior  of  the  principal  to  the  amount  of  consensus  which  existed 
among  teachers  and  the  amount  of  agreement  between  teachers  and 
principals  on  the  role  of  the  principal.  The  analysis  revealed  a 
higher  relationship  between  style  of  leader  behavior  and  Agreement 
than  between  style  of  leader  behavior  and  Consensus.  The  style  of 
leader  behavior  characterized  by  high  scores  on  both  dimensions  of 
the  Leader  Behavior  Description  Questionnaire  was  found  to  be  associ¬ 
ated  with  significantly  higher  Agreement  scores  than  any  other  style 
of  leader  behavior.  When  the  analysis  took  into  account  the  average 
length  of  time  which  a  principal  had  been  associated  with  his  present 
group  of  teachers,  it  was  found  that  for  principals  who  were  high  in 
Structure,  Agreement  and  Consensus  correlated  positively  with  Tenure 
while  for  those  who  were  low  in  Structure  the  correlations  were 
negative.  The  relationships  in  the  case  of  either  Consideration  and 
Ambivalence  were  not  as  clear  cut  as  this. 

II.  CONCLUSIONS  AND  IMPLICATIONS 

In  this  section  of  the  chapter  are  outlined  a  number  of  con¬ 
clusions  derived  from  the  study  and  some  possible  implications  for 


both  the  researcher  and  the  administrator. 


■  •  .  ■ 


.. 


.  .  ■  .  ■  ;  !  • 


•  ■  .  ••  V  .  "c 

■.  .  J:c  '  d'J  .  ■  ■  -  - 

■  ■  ■  I  -.J  •  • 

. 

1o  Qixox  .  X'  u  obvc  o  ‘  tv.  -  ■  iol 

.‘.x  ’[  i to  V  ■  •  '  •  i> 

■  .  V'j  '  .V  ;/  ■. 

.  ■■  •  ■  x  ;  os  ■  . Tc  xv  vba  %o 

■  ■  ‘  TV  T  ■:  ■  ■  •  :  TO  '  i£oI 

v  •  '•  .  •.  •  ‘ 

u"  l  - '  i  1  :T ' ?  •  ’ 

"  •  ■  ■  • :  w:  >.j  L:  o  - > r ' ;/  J . ; 


v  .. :  ■  •->.  '  . '■ 


.  :  :  : 


:  '  •  ’ .  •  •  •  . )  .  .  :  ■ 1  , 


:  ■  n/.o  ;!  •  :  :  r  '.j 


Conclusions 


195 


1.  Scale  analysis  appears  to  be  a  promising  approach  to  studying 
the  structure  of  the  expectations  which  are  held  for  the  incumbents  of 
certain  positions  in  formal  organizations .  However,  the  appropriateness 
of  the  Guttman  model  depends  to  a  large  extent  on  how  sharply  the  hypo¬ 
thesized  dimensions  are  defined  and  on  the  adequacy  of  pretests.  Even 
though  the  content  areas  originally  defined  in  this  study  were  not 
scalable,  the  study  does  indicate  which  sets  of  items  did  approach 
scalability. 

2.  Scalogram  analysis  does  provide  a  useful  means  for  quantify¬ 
ing  the  expectations  held  for  a  certain  position  when  a  unidimensional 
scale  is  shown  to  exist.  From  such  quantitative  variables  it  is 
possible  to  derive  other  indices  for  testing  hypotheses,  or  the 
quantitative  variables  themselves  may  be  used  for  testing  hypotheses. 

5.  The  intensity  with  which  a  principal  holds  his  self- 
expectations  does  appear  to  be  relaled  to  the  way  in  which  teachers 
perceive  and  describe  his  behavior  as  a  leader;  however,  the  relation¬ 
ship  appears  to  be  highly  complex  and  depends  in  part  upon  the  situation 
in  which  the  teacher  and  principals  are  operating.  The  extent  to  which 
principals  express  difficulty  in  choosing  among  behavioral  alternatives 
appears  to  be  related  favorably  to  leader  behavior  only  if  it  involves 
an  accurate  assessment  of  the  situation. 

4.  Style  of  leader  behavior  seems  to  be  related  to  the  extent 
to  v/hich  there  is  consensus  among  teachers  and  teacher-principal 
agreement  on  the  role  of  the  principal;  however,  the  relationship  to 


-■  ...  ■  •  •' .  : ■.  :  '  ovl  .  .  :j;  ;  .  : ;  ;•  ■ 


■  ■ 


. .  .  ..  . .  ;  ..  ‘  •  * 

V  v ■  oi’i.i' 


'  ;  :  J'  .  .  -r  J  ■  c  r  •  . ;  • 

o  -  o  . .  i  ■  .  j  .  ■'  ;  '*  .  '  .  ‘  .  ( 

'  ■  •  '  ;  -  -■  .  .  .  ;  « 

-  ■  . .  1  o  ^  . .  .  L- .  .  .  '  *  .  ’  .  '  ^ 


"  '  -  •  -  ■  Jc  '  •  .  : 

■  .  .....  ...... 

-  -  •>./:  c. 

;  c  :  -  -  •  •  ..  ..  ’  .  .  si 

“  , ,  -  - 

•  .  t. 

*■  •  '  ’  ••  •  ......  .w  ..  ...... 

:  -•  ■'  $  ■  ;  ■  £  -  :■  .  . 

•  ■  c  v  Vi  :  ' 

r  ■  ■'  -.  ..  ■  '  ...  ...  .  c;.f  ■  J  .  •  ■  ;  .  .  •  j:  '  , 

,  .  •  .  ... 

a- .  ..  .  .  -  :  „  •. ;  •  •  ,•  .  .-  ,  .  . 

•  ;  V  •  /.  -  ;  ;  .  \[  /;  ••  ;  ., 

'.  1  •  ■  .  ■  ■ 

•  v'  M.;. 

•'  •'  -  -  ■'  J’  i  •  .  J- 


•  -  .  ■  ; 


196 

Agreement  was  found  to  be  higher  than  that  of  style  of  leader  behavior 
to  Consensus.  The  study  revealed  that  style  of  leader  behavior  charac¬ 
terized  by  scores  high  in  both  Initiating  Structure  and  Consideration 
to  be  associated  with  a  high  degree  of  agreement  between  teachers  and 
principal  on  the  expectations  for  the  principal’s  role.  Whether  it  is 
the  high  degree  of  agreement  which  makes  it  possible  for  him  to  behave 
effectively  as  a.  leader  or  whether  his  general  behavior  as  a  leader 
makes  it  possible  for  him  to  influence  the  expectations  of  teachers  in 
other  areas  as  well  is  difficult  to  state.  No  doubt  the  two  axe 
interrelated,  but  the  theoretical  background  of  the  study  would  suggest 
that  desirable  leader  behavior  is  probably  a,  significant  factor  in  the 
interrelationship . 

5.  The  interrelationship  of  leader  behavior  and  the  Consensus 
and  Agreement  variables  was  also  illustrated  by  the  finding  that  for 
those  principals  who  were  high  in  Structure,  both  Agreement  and 
Consensus  correlated  positively  with  the  length  of  time  that  the 
principal  had  been  associated  with  the  particular  group  of  teachers. 
These  same  correlations  proved  to  be  negative  in  the  case  of  principals 
who  were  low  in  Structure.  This  suggests  that  although  tenure  does 
influence  the  other  two  variables,  the  direction  of  the  influence 
seems  to  depend  upon  the  leader  behavior  which  the  principal  is  able 
to  exhibit. 

Implications  for  Research 

This  exploratory  study  suggests  that  the  utility  of  scale 
analysis  in  analyzing  the  expectations  held  for  schoo]  administrators 


197 


should,  be  investigated  further.  This  could,  be  done  either  through 
studies  designed  specifically  for  determining  the  structure  of  the 
expectations  or  in  connection  with  studies  designed  for  testing 
various  hypotheses.  Of  particular  interest  would  be  a,  far  tor  analysis 
of  the  quantitative  variables  obtained  from  a  series  of  uni dimensional 
scales.  Of  course,  the  factor  analysis  might  also  be  carried  out  on 
the  items  themselves,  but . this  investigator  would  favor  the  suggested 
approach  because  the  analysis  would  be  carried  out  on  quantitative 
variables  which  had  already  been  shown  to  have  some  meaning  in  terms 
of  the  expectations  which  they  represented. 

Future  studies  could  benefit  from  the  experience  in  constructing 
unidimensional  scales  illustrated  in  this  study  and  could  probably 
avoid  some  of  the  problems  which  have  been  described.  A  careful 
selection  of  items  with  proper  pretesting  and  careful  selection  of 
response  categories  which  could  be  readily  dichotomized  would  be 
advisable.  If  these  steps  were  properly  carried  out,  much  of  the 
analysis  could  be  accomplished  by  using  the  machine  procedures  which 
have  been  developed.  This  would  eliminate  the  step  which  was  most 
time-consuming  in  this  study  and  would  make  the  use  of  scale  analysis 
that  much  more  attractive. 

The  reactions  of  principals  to  conflicting  expectations  needs 
to  have  much  more  intensive  investigation.  The  use  of  paper  and 
pencil  techniques  for  determining  difficulty  of  choice  or  intensity 
of  self -expectations  or  Ambivalence  was  useful  for  exploratory  pur¬ 
poses.  The  next  step  would  appear  to  be  an  analysis  of  the  psychologi¬ 
cal  meaning  of  this  variable  and  a  study  of  it  or  related  variables 


:  :  .V  '  '  .  x. ;  ■:>  •  ■ :  .i.J 

y  ■  '  -!  ■  ;  :  lx  •  ■■  - 

•  ■  ■  '  -  .  '.v  ■  r; 

.  s  .  . 

y  .  .  !'  .  J:  v  J  -  .  v  •  o;  . 

:  .  J  I  ;  .)•  tJC  ■ 

’  '  '  '  ■  :  rr  .  \ 

; 

'  -  -  ..  ■'  •. ‘1  .  >  .  -  ••  •. •  /  :  v  .1  :• : 

•  •  -  1  J"  /.  ..'x-m  .V  .a;/' 

■  1  •  ■  :  VjS d  crfvi  ' .  is  M  v.  ■ 

xc  .  .  o  Jb  '■  •  leqcri  .  (:  fjfc  lo  i 

.  .  ■  ■  .  .  jV  ■ 

:  r  ■'  ■  >  '  :  ;  .  • 

:  r- 

.  ■  ■  J 

.  '  .  J: 

-  '  >  •  ' :  .  ji  a  :  J  ■ 

'  .j:;.'  ’  :  ; 

•  -  -  1  ;  i  ■  i ; 


198 

through  reactions  in  simulated  situations  or  perhaps  by  using  different 
instruments  which  have  been  designed  to  measure  variables  closely 
related  to  what  has  been  defined  as  Ambivalence  in  this  study. 

If  a  self-report  instrument  were  used  again  in  a  related  study, 
some  instruments  other  than  those  used  in  this  study  should  be  used  in 
an  effort  to  determine  differences  in  the  actual  or  perceived  behavior 
of  the  administrator.  It  has  already  been  indicated  that  some  measure 
such  as  an  index  of  social  sensitivity  on  the  part  of  the  leader 
would  probably  have  been  more  useful  than  the  LBDQ,  dimensions.  The 
suggestion  here  is  that  not  only  must  the  meaning  and  measurement  of 
ambivalence  be  defined  but  the  theory  of  the  relationship  of  this 
variable  to  significant  outcomes  in  teacher-principal  interaction 
must  be  developed  and  then  adequate  means  obtained  for  testing  hypo¬ 
theses  derived  from  the  theory. 

Consensus  among  teachers  and  teacher -principal  agreement  on 
expectations  for  the  role  of  the  principal  could  also  be  subjected  to 
more  analytical  treatments.  Studies  might  be  designed  to  attempt  to 
clarify  the  causal  relationships  involved  as  well  as  to  study  the 
correlates  of  variations  in  the  degrees  of  Consensus  and  Agreement. 

It  would  be  interesting  to  know  definitely  if  principals  are  able  to 
influence  expectations  held  for  them  and  if  so,  by  what  process  this 
is  communicated  to  the  teachers.  Some  work  might  also  be  done  on 
predicting  teachers'  expectations  for  the  role  of  the  principal  from 
personal  variables  such  as  training  and  experience. 

As  in  the  case  of  an  exploratory  study,  this  one  ends  by 


' 


; 


; 


■■ 


- 


■ 


■- 


■  .  ... 


•  1  -v 


. . 


'  ••  ’  •-  ■  -  -  ■  ••  <  :/  ■  v 


■  ■  J  .V  ■  ■ 

.  .  ■  ,  :  ;■  ■ 


•  .  ■  i  ■. .  .  :  '  .  , 

-  :  J.  ;  q 


i  ■  ;  . 

::  1 '  . . 


/■  '  ..-JI  J'  '  j:.': 

-  . 


i  i  .  o.,.j  h 


:  •  r 


r 


199 

suggesting  more  hypotheses  than  were  included  in  the  design  of  the 
study.  Perhaps  this  is  just  another  one  of  those  many  problems 
operative  in  the  present  stage  of  research  in  educational  administ¬ 
ration.  Research  has  tended  to  gloss  over  too  many  areas  without 
giving  them  the  attention  which  they  deserve  in  view  of  their 
implications  for  practice.  The  problem  'is  complicated  by  definitive 
statements  concerning  research  findings  which  are  premature  to  say 
the  least.  Eanh  and  every  relationship  which  is  considered  to  be  of 
importance  to  the  field  must  be  studied  a  number  of  times  in  a 
variety  of  situations  as  indeed  they  must  in  any  field  of  study. 

Implications  for  Admini s  tr  at  or  s 

This  study,  like  ary  single  study,  holds  only  limited  impli¬ 
cations  for  the  practice  of  administration.  It  does  offer  some 
encouragement  in  that,  there  appear  to  be  discernible  and  meaningful 
structures  in  the  expectations  which  teachers  hold  for  administrators 
and  that  these  structures  might  be  readily  determined  by  fairly  simple 
techniques  if  research  is  extended  in  this  area.  If  an  administrator 
had  adequate  knowledge  about  the  general  structure  of  expectations 
and  some  of  the  more  specific  expectations  of  a  particular  group,  he 
would  at  leant  be  able  to  predict  how  his  behavior  or  actions  will 
be  received  and  thus  be  able  to  govern  himself  according  to  his  aims 
and  purposes.  Although  this  study  does  not  offer  very  much  that  is 
definite  in  this  respect,  it  does  indicate  that  such  optimism  is  not 


unwarranted. 


■  )  . 


■ 


£  ;  J  oJ‘  'I;o 

■  ■  .■  o- ■  o/s  o  ■ 


J  i  ■  .  o 

.o’; '  ' 

- 


■  — '  '  :  .V  •-  J  ;  ( 

“  -  1  •  •'  ■  '  :  v  .  :  ■:  -  o.';  .  v 


. 


.  >  ■  j '  ,  • ;■ 


i 


x  ;  - 


■ 


’)  A,'  1 : ’  i  ' 


* 


-  .oo  ■ 


•  O''  • 


o  I.[  ■  '  1  J 

»  ■■  -  • 

' 


200 


The  study  also  indicates  that  some  administrators  may  he  able 
to  modify  and  influence  the  expectations  which  significant  others 
hold  for  them  in  a  variety  of  areas.  At  the  present  time  the  best 
speculation  that  can  be  made  is  that  this  accrues  to  the  leader  who 
is  able  to  exhibit  appropriate  leader  behavior  over  a  period  of  time 
during  which  he  is  interacting  with  his  staff.  If  appropriate 
leader  behavior  is  not  forthcoming,  the  indications  axe  that  con¬ 
ditions  with  respect  to  Consensus  and  Agreement  will  continue  to 
deteriorate  the  longer  that  the  principal  is  in  the  school. 

On  the  basis  of  the  demonstrated  relationship  between  leader 
behavior,  Consensus,  and  Ambivalence,  it  seems  reasonable  to  suggest 
that  the  principal  of  a  school,  or  any  administrator  as  far  as  that 
goes,  should  make  a  conscious  effort  to  assess  the  extent  to  which 
there  is  agreement  between  himself  and  his  teachers  concerning  his 
role;  furthermore,  this  will  also  involve  an  assessment  of  the 
degree  of  consensus  among  the  teachers.  Perhaps  the  former  is  more 
important  than  the  latter  for  reasons  which  will  be  discussed  below. 
If  there  is  considerable  discrepancy  between  the  expectations  of  the 
teachers  and  the  principal's  self-expec tations ,  there  is  some  evidence 
that  a  principal  may  be  able  to  do  something  about  the  problem  by 
exhibiting  appropriate  leader  behavior  generally. 

The  fact  that  the  teachers  in  a  school  are  not  in  complete 
agreement  on  the  role  of  the  principal  should  not  be  a  source  of  too 
much  distress  for  an  administrator  because  it  may  be  that  this  is 
what  makes  it  possible  for  him  to  function  at  all  if  one  agrees 


. :  ■ 

j- 

.  . 

. 

:■( 

. 

* 

.  .■  .  v'  .. 

' 

- 

• 

'  J  ;  ' 

ut  ■  / 

i: 

•  J'  . 

■  - 

-  c 

. 

-  - 

.  > 

'•  i 

•  J- 

. 

• 

. 

• 

. 

iX 

.  . 

■  . 

■  .  ' .  •  c  . 

-> 

r.  .  ■  \  ) 

.  •  .  J 

.  : 

‘ 

■■  ■ 

i  i'j 

. 

,f  J;: 

• 

. 

f  vf  ■  . 

:  ■'  ' 

201 


with  the  analysis  by  Stouffer  (5).  The  very  fact  that  teachers  dis¬ 
agree  opens  up  a  band  of  behavior  for  the  administrator  within  which 
he  can  perform  his  duties.  It  may  be  that  he  will  seldom  be  able  to 
please  everyone  entirely;  on  the  other  hand,  he  may  seldom  displea.se 
everyone  either.  Even  though  there  is  a  band  of  acceptable  behavior 
open  to  him,  the  a.dministra.tor  might  do  well  not  to  rely  blindly  upon 
its  existence  without  making  a.  conscious  effort  to  determine  its 
limits.  To  be  thoroughly  familiar  with  the  expectations  held  by 
subordinates  is  probably  not  unwise  strategy  for  an  administrator  to 
follow  as  he  functions  within  the  organization  or  as  he  seeks  to 
modify  the  expectations  in  order  to  function  effectively.  It  may 
very  well  be  that  the  administrators  who  practice  this  are  the  ones 
described  favorably  as  leaders  by  their  subordinates. 

Both  the  student  of  administration  and  the  practicing 
administrator  are  well  aware  that  administration  involves  many 
difficult  tasks.  In  the  past  some  of  the  research  has  done  little 
more  than  verify  the  fact;  however,  there  are  indications  that  if 
research  is  persistent  in  various  a.spects  of  the  many  problems, 

results  in  the  form  of  clues  for  further  research  will  give  way  to 

\ 

results  which  are  empirically  useful.  This  is  not  to  say  that 
administration  will  be  less  challenging;  the  challenge  will  be  the 
greater  when  it  has  been  made  explicit  and  demonstrated  to  be 


manageable. 


-  J:  ;  . 

-  .  '  J .  J  '  •  J  "1  : 

■  .  ■  J  J'  J:  J 

''  J:  ' V  1  '  V  '  i  v  v  -v  :  -  v;  v 

?  ■  J  ! •.  >;i  , 

■  •  -  ■  J:  '  \  r  ;  v." 

.  .  , 

J  •  i: J 

■  Jr  .  a; • 

. .  .  • 

■ 

'  V  •  J  J  .  J;  i  (j  ' 

J'J:  J;  ■  J  '  Jm  ‘  'Y.'_ 

'  Jv '  -v  J'  vJJ:  v  '  J 

-  J: '  v  v  -vi  •  ;i 

J  ''  J  J  v:  J  •  .  J  J;  J: 

■  J  •  r'vv:  J.'  •.  y:;  ■;  c.  ■ 

J'v  :  --  .JlvuJ..)' /  .-J;  ■  .v  .Jv vw' 

•  '  ■  :J  J: 

J.  J:  .  J:  i  .  )  ij'v 

r  .  -  Jr  '  J;.  o 

J'  ■  J'J'  J' ;  -  '  ;■  '  ~  '..J;  '■ 

■  "  o'-.tJ  JvJ:  •'  ■  v  J  v  •,  .;J-  v.  "  J'Jviv 

• 

V:  '  ••  •' 

JJ  •  J'  J:  J-  J  J-  ■  ;j- 

-'J  '  J:  "  J.J'  J-  J. 

■  J‘  •  '  '  ■  ■  '  J:  -J.  5,  ■ 

.  J  ;J 

J'  „  J  J; 

'  ?  ••  V  -  ' 

c 

■  '  Jr '  '  '  J:  J  J  J,  ■  J:  • • 

"  ' :  .  £  'J  J  r  ■ 

✓ 

J'J  '  £  J  ‘ 

i  J  J-  '  tl  J:  ; 

' 

J:  ■  J  ■  /■■••• 

REFERENCES  FOR  CHAPTER  X 


Guttman,  Louis.  "The  Cornell  Technique  for  Scale  and. 
Intensity  Analysis."  Educations,!  and  P s,y o hoi ogic al 
Measurement ,  7^24  7-279 >  Summer,  1947- 

Seeman,  Melvin.  "Role  Conflict  and  Ambivalence  in 
Leadership,"  American  Sociological  Review, 

18:  373-380,  Au'gust,  1953. 

Seeman,  Melvin.  Social  Status  and  Leadership .  Columbus, 
Ohio:  The  Ohio  State  University,  I960. 

Stouffer,  Samuel  A.,  et  al.  Me asur ement  and  Prediction. 
Vol .  IV  of  Studies  in  Social  Psychology  in  World  War  II . 
4  vols.  Princeton,  New  Jersey:  Princeton  University 
Press,  1950. 

Stouffer,  Samuel  A.  "An  Analysis  of  Conflicting  Social 
Norms,"  American  Sociological  Review,  14:  707-717) 
December,  1949* 


BIBLIOGRAPHY 


BIBLIOGRAPHY 


A.  BOOKS  AND  MONOGRAPHS 


Campbell,  Roald  P. ,  John  E.  Corbally,  Jr.,  and  John  H.  Ramseyer. 

Intro  due  tion  to  Educational  Admini  s  tr  ah  i  on .  Second  Edition. 
Boston:  ALlyn  and  Bacon,  Inc.,  1962. 

Campbell,  Roald  F. ,  and  Russell  T.  Gregg  (eds.).  Administrative 
Behavior  in  Education.  New  York:  Harper  and  Brothers,  1957* 

Cantril,  Hadley  (ed.).  Ganging  Public  Opinion.  Princeton,  New 
Jersey:  Princeton  University  Press,  1944* 

Edwards,  Allen  L.  Techniques  of  Attitude  Scale  Construction. 

New  York:  Appleton-Century-Crofts,  Inc.,  1957* 

Festinger,  Leon,  and  Daniel  Katz  (eds.).  Research  Methods  in 
the  Behavioral  Sciences ♦  New  York:  Holt,  Rinehart  and 
Winston,  1953* 

Fromm,  Erich.  Escape  from  Freedom.  New  York:  Farrar  and  Rinehart, 
Inc . ,  1941 • 

Garrett,  Henry  E.  Statistics  in  Psychology  and  Education. 

New  York:  Longmans,  Green,  and  Company,  1958* 

Griffiths,  Daniel  E.  Administrative  Theory.  New  York:  Appleton- 
Century-Crofts,  Inc.,  1959* 

Gross,  Neal,  Ward  S.  Mason,  and  Alexander  McEachern.  Explorations 
in  Role  Analysis:  Studies  of  the  School  Super intendency  Role . 

New  York:  John  V/iley  and  Sons,  Inc.,  1958* 

Guilford,  J.  P.  Psychometric  Me thods .  Toronto:  McGraw-Hill  Book 
Company,  1954* 

H alp in,  Andrew  W.  (ed).  Admini s tr at i ve  Theory  in  Education. 

Chicago:  Midwest  Administration  Center,  1958. 

Halpin,  Andrew  V.  The  Leadership  Behavior  of  School  Superintendents . 
Second  Edition.  Chicago:  Midwest  Admini strati on  Center,  1959* 

Hemphill,  John  K.  Situational  Factors  in  Leadership.  Columbus, 

Ohio:  The  Ohio  State  University,  1949* 

Henry,  Nelson  B.  (ed.).  The  Dynamics  of  Instructional  Groups. 

Fifty-ninth  Yearbook  of  the  National  Society  for  the  Study  of 
Education,  Part  II.  Chicago:  University  of  Chicago  Press,  i960. 


205 


Lazarsfeld,  Paul  F.  (ed.).  Mathematical  Thinking  in  the  Social 
Sciences .  Glencoe,  Illinois:  The  Free  Press,  1954- 

Lindzey,  Gardner  (ed.).  Handbook  of  Social  Psychology,  Vol .  I. 

Reading,  Massachusetts:  Addis on -Wes ley  Publishing  Company,  1954* 

McNemar,  Quinn.  Psychological  S tali sties ♦  Third  Edition. 

Hew  York:  John  Wiley  and  Sons,  Inc.,  1962. 

Merton,  Robert  K.  Social  Theory  and  Social  Structure .  Glencoe, 
Illinois:  The  Free  Press,  1957* 

Myrdal,  Gunnar ,  Richard  Sterner,  and  Arnold  Rose.  An  .American 
Dilemma, :  The  Negro  Problem  and  Modern  Democracy.  New  York: 

Harper  and  Brothers,  1944* 

Parsons,  Talcott,  and  Edward  A.  Shils  (eds.).  Toward  a  General 

Theory  of  Action.  Cambridge,  Massachusetts:  Harvard  University 
Press,  1951* 

Riley,  Matilda,  White,  John  W.  Riley,  Jr.,  and  Jankson  Toby. 

Sociological  Studies  in  Scale  Analysis ♦  New  Brunswick,  New 
Jersey:  Rutgers  University  Press,  1954* 

Seeman,  Melvin.  Social  Status  and  Leadership.  Columbus,  Ohio: 

The  Ohio  State  University,  I960. 

Selltiz,  Claire,  et  al.  Research  Methods  in  Social  Relations . 

New  York:  Henry  Holt  and  Company,  1959* 

Stogdill,  Ralph  M. ,  and  Alvin  E.  Coons  (eds.).  Leader  Behavior: 

Its  Description  and  Measurement .  Columbus,  Ohio:  The  Ohio 
State  University,  1957* 

Stouffer,  Samuel  A.,  et  al.  The  American  Soldier:  Adjustment  During 
Army  Life .  Yol.  I  of  Studies  in  Social  Psychology  in  World  War  II . 
4  vols.  Princeton,  New  Jersey:  Princeton  University  Press,  1949- 

Stouffer,  Samuel  A.,  et  al.  Measurement  and  Prediction.  Vol.  IY 
of  Studies  in  Social  Psychology  in  World  War  II.  4  vols. 

Princeton,  New  Jersey:  Princeton  University  Press,  1950* 

Torgerson,  Warren  S.  Theory  and  Methods  of  Scaling.  New  York: 

John  Wiley  and  Sons,  Inc.,  1958* 

Williams,  Robin  M. ,  Jr.  American  Society:  A  Sociological  Inter¬ 
pretation.  New  York:  Alfred  A.  Knopf,  1951* 


c 


<  * 


206 


B.  ARTICLES 


Ake,  James  N.  "Rapid  Machine  Procedures  for  Determining  Scalability 
of  any  Number  of  Questions,"  Public  Opinion  Quarterly,  26: 
121-125?  Spring,  1962. 

Back,  Kurt,  Reuben  Hill,  and  J.  Mayone  Stycos.  "Interviewer 
Effect  on  Scale  Reproducibility,"  American  Sociological 
Review,  20:445-446,  August,  1955* 

Bidwell,  Charles  E.  "The  Administrative  Role  and  Sa/tisfaction 
in  Teaching,"  Journal  of  Educational  Sociology,  29:41-47? 
September,  1955 - 

Boneau,  C.A.  "The  Effects  of  Violations  of  Assumptions  Underlying 
the  t  Test,"  Psychological  Bulletin,  57:49-64?  January,  I960. 

Borgatta,  Edgar  P.  "An  Error  Ratio  for  Scalogram  Analysis," 

Public  Opinion  Qparterly ,  19:96-100,  Spring,  1955* 

Brim,  Orville  G. ,  Jr.  "Attitude  Content-Intensity  and  Probability 
Expectations,"  American  Sociological  Review,  20:68-76, 

February,  1955* 

Burchard,  Waldo  W.  "Role  Conflicts  in  Military  Chaplains," 

American  Sociological  Review,  19:528-555?  October,  1954* 

Campbell,  Ernest  Q. ,  and  Alan  C.  Kerckhoff.  "A  Critique  of  the 
Concept  ’Universe  of  Attributes’,"  Public  Opinion  Quarterly, 
21:295-505,  Summer,  1957- 

Campbell,  Merton  V.  "Teacher -Principal  Agreement  on  the  Teacher 

Role,"  Administrator's  Notebook,  Vol.  7?  ho*  6,  February,  1959* 

Case,  Herman  M.  "Guttman  Scaling  Applied  to  Centers'  Conservatism- 
Radicalism  Battery,"  American  Journal  of  Sociology,  58:556-565? 
May?  1955* 

Chase,  Francis  S.  "Professional  Leadership  anu  Teacher  Morale," 
Administrator  *  s  Notebook,  Vol.  1,  No.  8,  March,  1955* 

Chase,  Francis  S.  "How  to  Meet  Teachers'  Expectations  of 
Leadership,"  Administrator ' s  Notebook ,  Vol.  1,  No.  9? 

April,  1955* 

Chase,  Francis  S.,  and  William  W.  Savage  (eds.).  "Attitudes  of 
Teachers  and  Administrators,"  Administra.tor ' s  Notebook , 

Vol.  5?  Ho.  1,  September,  1954* 


20? 


Coleman,  James  S.  "Multidimensional  Scale  Analysis,"  Meric  an 
Journal  of  Sociology,  65:255-263,  November,  1957* 

Cronbach,  Lee  J. ,  and  Goldine  C.  Gleser.  "Assessing  Similarity 
Between  Profiles,"  Psychological  Bulletin,  50:456-473? 
November,  1953* 

Danielson,  Wayne  A.  "A  Data  Reduction  Method  for  Scaling 

Dichotomous  Items,"  Public  Opinion  Quarterly,  21:377-379? 

Pall,  1957- 

Davis,  James  A.  "On  Criteria  for  Scale  Relationships,"  American 
Journal  of  Sociology,  63:371-380,  January,  1958. 

Perneau,  Elmer  P.  "Which  Consultant?"  Administrator 1 s  Notebook, 

Vol.  2,  No.  8,  April,  1954- 

Pleishman,  Edwin  A.  "The  Measurement  of  Leadership  Attitudes 
in  Industry,"  Journal  of  Applied  Psychology,  37:153-158, 

June,  1953  - 

Pord,  Robert  N.  "A  Rapid  Scoring  Procedure  for  Scaling  Attitude 

Questions,"  Public  Opinion  Quarterly,  14:507-532,  Fall,  1950. 

Getzels,  J.W.  "A  Psycho-Sociological  Framework  for  the  Study  of 
Educational  Administration,"  Harvard  Educational  Review, 
22:235-246,  Pall?  1952. 

Getzels,  J.W. ,  and  E.  G.  Guba.  "Role,  Role  Conflict,  and 

Effectiveness,"  American  Sociological  Review,  19:164-175? 
April,  1954* 

Getzels,  J.W. ,  and  E.  G.  Guba,.  "The  Structure  of  Roles  and 
Role  Conflict  in  the  Teaching  Situation,"  Journal  of 
Educational  Sociology,  29:30-48?  September,  1955* 

Getzels,  J.W.,  and  E.  G.  Guba.  "Social  Behavior  and  the 
Administrative  Process,"  School  Review,  65:423-441? 

Winter,  1957* 

Green,  Bert  P.  "A  Method  of  Scalogram  Analysis  Using  Summary 
Statistics,"  Psychome tr ika ,  21:79-88,  March,  1956. 

Green,  Norman  E.  "Scale  Analysis  of  Urban  Structures:  A  Study 
of  Birmingham,  Alabama,"  American  Sociological  Review, 

21:8-13,  February,  1956. 

Gullahorn,  John  1.  "Measuring  Role  Conflict,"  Meric  an  Journal 
of  Sociology,  61:299-303?  January,  1956. 


208 


Guttman,  Louis.  "The  Cornell  Technique  for  Scale  and  Intensity 
Analysis,"  Educational  and  Psychological  Measurement, 

7:247-279)  Summer,  1947* 

Hall,  Robert  L.  "Social  Influence  on  the  Aircraft  Commander's 

Role,"  American  Sociological  Review,  20:292-299)  June,  1955* 

Halpin,  Andrew  ¥.  "The  Leadership  Behavior  and  Combat  Performance 
of  Airplane  Commanders,"  Journal  of  Abnormal  and  Social 
Psychology,  49:19-22,  January,  1954* 

Halpin,  Andrew  W.  "The  Superintendent's  Effectiveness  as  a 

Leader,"  Administrator's  notebook,  Vol.  7)  Ho.  2,  October, 

1958. 

Hencley,  Stephen  P.  "The  Conflict  Patterns  of  School  Superin¬ 
tendents,"  Administrator 1 s  Notebook,  Vol.  8,  Ho.  9)  May,  I960. 

Henry,  Andrew  P. ,  and  Edgar  P.  Borgatta.  "A  Consideration  of 

Some  Problems  of  Content  Identification  in  Scaling,"  Public 
Opinion  Quarterly,  20:457-469)  Summer,  1956. 

Jacobson,  Eugene,  ¥.¥.  Charters,  Jr.,  and  S.  Lieberman.  "The 

Use  of  the  Role  Concept  in  the  Study  of  Complex  Organization, " 
Journal  of  Social  Issues ,  7:18-27)  Humber  5)  1951* 

Korber,  George.  "Comments  on  'Role  Conflict  and  Personality'," 
American  Journal  of  Sociology ,  52:48-49)  July,  1951* 

Laulicht,  Jerome.  "Role  Conflict,  the  Pattern  Variable  Theory, 
and  Scalogram  Analysis,"  Social  Forces,  35:250-254)  March, 

1955- 

Lesser,  Gerald  S.  "Application  of  Guttman' s  Scaling  Method  to 

Aggressive  Fantasy  in  Children,"  Educ ati onai  and  Psychological 
Measurement,  18:545-561,  August,  1958. 

Me Dill,  Edward  L.  "A  Comparison  of  Three  Measures  of  Attitude 
Intensity,"  Social  Forces,  38:95-99)  December,  1959* 

Mehling,  Reuben.  "A  Simple  Test  for  Measuring  Intensity  of 

Attitudes,"  Public  Opinion  Quarterly,  23:576-578,  Winter,  1959* 

Menzel,  Herbert.  "A  Hew  Coefficient  for  Scalogram  Analysis," 

Public  Opinion  Quarterly,  17:268-280,  Summer,  1953* 

Motz ,  Annabelle  Bender.  "The  Role  Conception  Inventory:  A  Tool 
for  Research  in  Social.  Psychology,"  American  Sociological 
Review,  17:465-471,  August,  1952. 


c 


•  « 1  <  ‘  ? 


Moyer,  Donald  G.  "Leadership  that  Teachers  Want," 
Notebook,  Vol.  3,  No.  7,  March,  1955- 


Administrator ' s 


209 


Nye,  Ivan  F. ,  and  James  F.  Short,  Jr.  "Scaling  Delinquent  Behavior," 
American  Sociological  Review,  22:326-331?  June,  1957* 

Pearson,  Richard  G.  "Plus  Percentage  Ratio  and  the  Coefficient  of 
Scalability,"  Public  Opinion  Quarterly,  21:379-580? 

Fall,  1957- 

Perry,  Stewart  E. ,  and  Lyman  C.  Wynne.  "Role  Conflict,  Role 

Definition,  and  Social  Change  in  a  Clinical  Research  Organization, 
Social  Forces ,  38:62-65,  October,  1959* 

Riley,  Matilda  White,  and  Jackson  Toby.  "Subject  and.  Object  Scales: 

A  Sociological  Application,"  American  Sociological  Review, 

17:  287-29 6,  June,  1952. 

Riley,  Matilda  White,  John  W.  Riley,  Jr.,  and  Marcia  L.  Toby. 

"The  Measurement  of  Consensus,"  Social  Forces,  31:97-106, 
December,  1952. 

Schubert,  Glendon  A.  "The  Study  of  Judicial  Decision-Making  as 
an  Aspect  of  Political  Behavior;  Scalogram  Analysis," 

American  Political  Science  Review,  52:1014-1017,  December,  1958. 

Schuessler,  Karl  F.  "Item  Selection  in  Scale  Analysis," 

Meric  an  Sociological  Review,  17:183-192,  April,  1952. 

Scott,  John  Finley.  "Two  Dimensions  of  Delinquent  Behavior," 

American  Sociological  Review,  24:240-243?  April,  1959* 

Seeman,  Melvin.  "Role  Conflict  and  Ambivalence  in  Leadership," 
American  Sociological  Review,  18:373-380,  August,  1953* 

Southall,  Aidan.  "Operational  Theory  of  Role,"  Human  Relations, 
12:17-34,  February,  1959- 

Steiner,  Ivan  D.  "Scalogram  Analysis  as  a  Tool  for  Selecting 
Poll  Questions,"  Public  Opinion  Quarterly,  19:415-424? 

Winter,  1955* 

Stogdill,  Ralph  M.  "Personal  Factors  Associated  with  Leadership: 

A  survey  of  the  Literature,"  Journal  of  Psychology ,  25:35-71? 
January,  1948. 

Stover,  Robert  E.  "The  Measurement  of  Change  in  a  Uni dimensional 
Attitude  by  Guttman  Scale  Analysis  Techniques,"  Public 
Opinion  Quarterly,  22:116-122,  Summer,  1958. 


.  . 


c  • 


: 


<c 


210 


Stouffer,  Samuel  A.  "An  Analysis  of  Conflicting  Social  Norms," 
American  Sociological  Review,  14:707 -717?  December,  1949* 

Stouffer,  Samuel  A.,  and  Jackson  Toby.  "Hole  Conflict  and 

Personality,"  Meric  an  Journal  of  Sociology,  56:595-406, 
March,  1951. 

Strauss,  Anselm  L.  "The  Development  and  Transf ormation  of  Monetary 
Meaning's  in  the  Child,"  American  Sociological  Reviev/, 
17:275-286,  June,  1952. 

Sutcliffe,  J.  P. ,  and  M.  Haberman.  "Factors  Influencing  Choice 
in  Role  Conflict  Situations,"  American  Sociological  Review, 
21:695-705,  December,  1956. 

Toby,  Jackson.  "Some  Variables  in  Role  Conflict  Analysis," 

Social  Forces,  50:525-527,  March,  1952. 

Vernon,  Glenn  M.  "Inquiry  into  the  Scalability  of  Church 
Orthodoxy,"  Sociology  and  Social  Research,  59:524-527? 

May,  1955. 

Wallin,  Paul.  "A  Guttman  Scale  for  Measuring  Women’s  Neighbor¬ 
liness,"  American  Journal  of  Sociology,  59:245-246, 

November ,  1955* 

Vispe,  Lauren  G.  "Sociometric  Analysis  of  Conflicting  Role- 
Expectations,"  American  Journal  of  Sociology,  61:154-157? 
September,  1955*  Erra,ta  61:565?  January,  1956. 


C .  UNPUBLISHED  MATERIALS 


Cheal,  John  E.  "Role  Conflict  in  the  Principalship  of  the 

Composite  High  School."  Unpublished  Master’s  thesis,  The 
University  of  Alberta,  Edmonton,  1958* 

Greenfield,  Thomas  Barr.  "Teacher  Leader  Behavior  and  its  Relation 
to  Effectiveness  as  Measured  by  Pupil  Growth."  Unpublished 
Master's  thesis,  The  University  of  Alberta,  Edmonton,  19 6l. 

Keeler,  Bernard  T.  "Dimensions  of  Leader  Behavior  of  Principals, 
Stair  Morale  and  Productivity."  Unpublished  Doctoral  thesis, 
The  University  of  Alberta,  Edmonton,  1961. 

McBeath,  Arthur  Groat.  "Teacher  Leader  Behavior  and  its  Relation 
to  Teacher  Effectiveness."  Unpublished  Master's  thesis,  The 
University  of  Alberta,  1959. 


/ 


.  -  t 


c 


*  ' 


$ 


c 


r 


Stone,  Carol  Larson.  "A  Machine  Method  for  Scaling  as  Many  as 
Twelve  Dichotomies,"  Pullman,  Washington:  Washington 
Agricultural  Experiment  Station,  Institute  of  Agricultural 
Sciences,  State  College  of  Washington,  August,  1958* 
(Mimeographed) . 

Warren,  Philip  John.  "Leadership  Expectations  of  the  Principal 
in  Newfoundland's  Regional  and  Central  High  Schools  as 
Perceived  Toy  Principals  and  Staffs."  Unpublished  Master's 
thesis,  The  University  of  Alberta,  Edmonton,  1959* 


APPENDIX  A 


INSTRUCTIONS  TO  TEACHERS 


EXPECTATIONS  FOR  PRINCIPAL  LEADER  BEHAVIOR 


215 


INSTRUCTIONS  TO  TEACHERS 


The  questionnaires  in  this  envelope  are  being  distributed  in 
connection  with  some  research  which  1  am  carrying  out  as  part  of  my 
studies  in  educational  administration  at  the  University  of  Alberta. 

The  research  project  is  concerned  with  the  various  expectations 
which  teachers  might  hold  for  the  behavior  of  a,  principal,  with  the 
way  that  principals  view  these  sane  expectations,  and  also  with-  the 
relationship  of  these  to  the  way  that  principals  actually  behave  as 
leaders  in  their  schools.  I  would  be  very  grateful  if  you  would 
assist  me  in  this  research  by  completing  each  of  the  enclosed  ques¬ 
tionnaires.  Even  though  there  seem  to  be  a  large  number  of  items, 
you  will  note  that  the  response  called  for  will  not  involve  a  large 
amount  of  time  for  each  item. 

Each  of  the  teachers  in  your  school  has  been  asked  to  complete 
the  questionnaires  as  have  the  teachers  in  more  than  fifty  schools 
similar  to  yours  in  other  parts  of  the  province.  The  questionnaires 
contain  no  identifying  marks  and  individual  respondents,  principals, 
and  schools  will  remain  completely  anonymous.  There  will  be  no  way 
in  which  anyone  reading  the  research  report  could  possibly  identify 
individual  schools  let  alone  individual  respondents.  It  is  very 
important  that  I  receive  complete  returns  from  each  teacher  in  each 
of  the  schools.  I  hope  that  you  will  be  able  to  find  the  time  to 
help  me  in  carrying  out  this  research. 

The  two  questionnaires  each  contain  complete  instructions, 
but  the  following  notes  on  each  one  might  be  helpful. 

Expectations  for  Principal  Leader  Behavior 

This  questionnaire  has  been  designed  for  surveying  the  opinions 
of  teachers  and  principals  concerning  the  behavior  of  principals. 

There  are  no  right  or  wrong  responses  in  the  sense  that  teachers 
should  either  agree  or  disagree  with  a  specific  item.  This  study  is 
concerned  with  finding  out  what  the  expectations  are  and  not  whether 
the  expectations  are  right  or  wrong. 

Leader  Behavior  Description  Questionnaire 

For  the  purposes  of  this  study  LEADER  refers  to  the  PRINCIPAL 
in  your  school  and  GROUP  refers  to  the  TEACHERS  in  your  school. 

Please  keep  this  in  mind  as  you  respond  to  the  items. 

You  need  not  write  the  name  of  the  leader,  the  group,  or 
your  name  on  this  questionnaire. 

This  questionnaire  has  been  designed  for  the  purpose  of 
enabling  members  of  a  group,  in  this  case  the  teachers  in  a  school, 
to  describe  the  behavior  of  their  leader,  in  this  case  the  principal 
in  the  school.  It  is  not  a  test  and  not  an  evaluation;  it  is  what 
the  title  indicates,  and  that  is  a  description  of  the  behavior  of 
a  leader.  It  has  been  used  in  several  previous  studies  in  this 


:•  '  r;  ;  ■ 

-  J-  '  ;  •  >  rr.  .i; 

.  '  •  •  >  ;  ••  ;  .  • 

''  •  •’  ■  •'•••  '  -  ■  •  ■  -■  v  •  ' :  ;  ;  ..  r:  .  ;  •  ;?  .■  -vj 

•  ■  ■  ;  :  :  .:  ’  ;  V  ;; 

.  ■  '  .’  '  ;.f  - y 

,  .  ;■ 

0  :  ■  •  ..<■  :  '  ■  • 

«  >  ■  ;!  u  . 

'  .  .  .  ■  i  ■  -  '  or _ j  ■  _ 

:  r  '..  r  uC1  .r  -  v  .  ’ 

•  v-  '  ....  *  ;•  r  or'  ...•  .1  V:  .  /for, 

J  •  v . '  •  , 

'  -  ‘  '  .;  \rr  •  j  r 

<  t  •  '  ■  ■  '■  ■  i  '  o  ■  ■ 

-  -  -  '  - 

•  -  ■  :  '  -  '  ' 

.  -  »  >  <  •  r  ■  :  ..'  '  v 

..  o.  •  V.  r  :  '  •  .  ;]■  .  r  ;  •  yj;  :  .  J'  •  ;/  ,v  if  /; 


.  '  *  •/  ..  v  ■  .  i ' 

:  ■  v  .  . :  .  .  '  :  :  \  r-  \  ■ 

'  ■;  ;.".V  !  ’  • ..  ■; cot  ' , . 

:•  '  T)  i '  z 

•  '  .  .  .  :  :  ■  ;  .  ...  /■ 


■  -  ■  i*  ii  j:  ... 

:  <  .  ■  :>  ■ 

..j’;:  j:, >;!  c. 7 

>  r  ■  j:  i‘ 

■  .'  '  .■  -'i.a  •  ,  or i  ,v  '. 

.  .’.  .'  • 

i  .  .1  ■ :.  .'  )  •..  ■  *  .  ; 


2 


214 


province  and  to  a  much  larger  extent  in  the  United  States.  Its  use 
in  Alberta  has  been  approved  by  the  Division  of  Educational  Adminis¬ 
tration  and  by  the  Alberta  Teachers’  Association.  The  fact  that 
your  principal  has  distributed  these  questionnaires  to  you  is  an 
indication  that  he  has  no  objections  to  having  the  teachers  in  the 
school  describe  his  behavior.  Please  make  certain  that  you  respond 
to  each  of  the  items  on  the  questionnaire. 


Let  me  assure  you  again  that  individual  schools  and  respon¬ 
dents  will  remain  totally  anonymous.  Do  not  write  your  name,  that 
of  your  principal,  or  the  name  of  your  school  on  either  of  the 
questionnaires . 

I  am  interested  in  individual  points  of  view  and  would 
encourage  you  to  respond  to  all  items  without  consulting  your  fellow 
staff  members  about  their  responses.  As  has  been  staled  before, 
there  are  no  right  or  wrong  answers  as  far  as  this  study  is  concerned. 


WHEN  YOU  HAVE  COMPLETED  THE  QUESTIONNAIRES  PLACE  THEM  IN  THE 
ENVELOPE  AND  SEAL  IT.  THEN  HAND  IT  TO  THE  STAFF  MEMBER  WHO  HAS 
BEEN  DESIGNATED  TO  RECEIVE  HIE  ENVELOPES  AND  HE  OR  SHE  WILL 
FORWARD  ALL  ENVELOPES  FROM  YOUR  SCHOOL  TO  ME. 


I  would  be  very  grateful  if  you  would  be  able  to  find  the 
time  to  complete  the  questionnaires  within  the  next  few  days.  May 
I  extend  my  thanks  in  anticipation  of  your  cooperation. 


Sincerely, 


Erwin  Miklos 


■ ;  ; 


SI  I)  • 


•'  '  -  •  .  '•  •  ■.  ■:  ■  ■  ;  -  ■ 

■  .•  /  •  •  ; 

'  ’  . 
J:  '  . 

:  '  :  •  ■  i 

*  .  .  . .  . .  ■  : 

«  .  ■  . ;  . 

. 


... 


' .  .  i  '■ 


::  ..  .  :  .  .  ■  : 

I1; .  ■  v.  ■: 


.  .  ; 


. 


r-  ■  ■ 


EXPECTATIONS  FOR  PRINCIPAL  LEADER  BEHAVIOR 


SECTION  A 

The  items  in  this  questionnaire  are  expectations  which 
teachers  or  other  individuals  might  hold  for  the  behavior  of 
school  principals.  You  are  asked  to  indicate  agreement  or 
disagreement  with  each  of  the  items.'  If  you  feel  tha,t  your 
responses  would  depend  upon  the  type  of  school  concerned,  then 
think  in  terms  of  the  school  in  which  you  are  now  teaching. 

DIRECTIONS: 

a. .  READ  each  item  carefully. 

b.  DECIDE  whether  you  strongly  agree,  agree,  disagree,  or 
strongly  disagree  with  the  statement. 

c.  CIRCLE  ONE  of  the  five  responses  following  the  item  to 
indicate  your  reaction  to  it. 

SA  -  Strongly  Agree;  A  -  Agree;  U  -  Undecided;  D  -  Disagree; 
SD  -  Strongly  Disagree 


START  HERE: 

1.  Within  the  school  a,  principal  should  try  to  be  just  another 
member  of  the  group  in  his  job  relations. 

2  .  A  principal  should  not  encourage  teachers  to  look  to  him  for 
help  in  controlling  their  classes. 

3.  The  principal  should  support  a  teacher’s  action  in  the  class¬ 
room  under  all  circumstances. 

4.  In  assigning  duties  a  principal  should  be  cognizant  of  the 
out-of-school  responsibilities  of  a  teacher. 

5.  A  principal  should  keep  a  certain  professional  distance 
between  himself  and  the  teachers  in  his  school. 

6.  A  principal  should  intervene  immediately  if  he  believes  that 
a  teacher  is  having  discipline  problems. 

7.  A  principal  shoiild  not  deliberately  gather  evidence  against  a 
student  who  he  believes  influences  other  students  adversely. 

8.  A  principal  should  not  expect  all  teachers  to  take  part  in 
supervising  hallways,  playgrounds,  and  extra-curricular 
activities . 

9.  A  principal  should  not  attempt  to  obtain  any  special  privilege 
from  the  board,  such  as  might  relate  to  housing,  unless  he  is 
also  willing  to  work,  for  having  these  privileges  extended 

to  teachers. 


\  ’  ■.  • 


•  1  >  ;  i  " 


;  m  ■  • 


■  .  :  i  .  }f\  • 

*  '  *  '  . .  . . 


V.\  ■ 


z 


'  ■  <:_  f  .•!  ■ ,  i '  1  fi<  ■ 

ic  ■  j  ■ 1  V  7..r 

v  •  •• '  .  ''  •  ; 

\  '  '  ■ 

■  ' 

:  til'  '■  b  . 


,  *  •  '*  ’  "  '  -  •  - 


.  ■ 

- 


216 


10.  A  principal  should  visit  regularly  the  classes  of  those 
teachers  who  are  weak  in  discipline  and  exert  his  personal 
influence  in  an  attempt  to  keep  the  classes  under  control. 

11.  A  principal  should  refuse  to  assist  students  with  their 
studies  if  they  come  to  him  with  the  complaint  that  a  teacher 
is  not  giving  them  enough  help. 

12.  A  principal  should  do  personal  favors  for  staff  members. 

13.  A  principal  should  spend  as  much  time  supervising  playgrounds, 
hallways,  and  extra-curricular  activities  as  any  of  the 
teachers  in  the  school. 

14*  A  principal  should  encourage  teachers  to  refer  serious  behavioral 
problems  to  him  only  as  a  last  resort. 

15 •  A  principal  should  not  hesitate  to  take  the  side  of  a  pupil 
if  he  feels  that  the  pupil  has  received  unfair  treatment  at 
the  hands  of  a  teacher. 

16.  A  principal,  should  not  visit  the  homes  of  some  teachers  any 
more  often  than  he  visits  the  homes  of  other  teachers. 

17.  A  principal  should  submit  to  the  superintendent  written 
reports  on  each  teacher’s  effectiveness. 

18.  A  principal  should  allow  teachers  to  work  out  their  class¬ 
room  problems  by  themselves. 

19.  A  principal  should  call  regular  stall*  meetings  even  though 
'  he  may  not  have  anything  of  importance  to  discuss  with  the 

staff. 

20.  A  principal  should  take  into  account  such  personal  obligations 
of  teachers  as  family  responsibilities  when  assigning  extra¬ 
curricular  duties. 

21.  Even  though  he  may  have  less  training  and/or  experience  than 
some  teachers,  the  opinions  of  a  principal  should  carry  more 
weight  than  those  of  the  teachers  in  the  school. 

22.  A  principal  should  visit  classes  regularly  to  determine  the 
quality  of  teaching, 

23.  A  principal  should,  delay  action  until  there  is  staff  consensus 
before  proceeding  with  a  staff  project. 

24.  A  principal  should  call  either  all  or  none,  and  not  just  some, 
of  the  teachers  in  the  school  by  their  first  names. 


9  t  ■  . 


:axv 

: 


.  '  ..  :  ;  v' 


' 


I  '  .  iO.  9 


.  '  - .  '...a  . . .  '  •  r'  ■  .9  o'  ■  h  i  , .  j  :  i  .. ; 

^  C--  . 


■  •  v  *■ 


.  „ 


•  .i.O 


h  :  : 


' 


■ : :  O'  ..  ■  .  ’O:  . 

.  •  .! 


r,o' 


■ 


;)■  ■. 


217 


25.  A  principal  should  think  of  himself  as  being  one  of  the 
teachers  in  the  school. 

26.  A  principal  should  expect  teachers  to  submit  to  him  for  his 
approval  copies  of  all  examinations  which  they  plan  to 
administer. 

27-  A  school  may  have  developed  certain  traditions  over  a  period 
of  time  which  an  incoming  principal  should  not  try  to  change. 

28.  A  principal  should  refuse  to  assist  teachers  who  apply  to  the 
board  for  some  special  concession  such  as  permission  to  leave 
school  a  few  days  before  the  end  of  the  term  so  that  they  can 
have  a  longer  holiday. 

29-  A  principal  should  try  to  achieve  a  higher  standard  of  living 
than  that  of  the  teachers  in  his  school. 

30.  A  principal  should  encourage  teachers  to  carry  out  various 
special  class  activities  such  as  room  parties  without 
consulting  him. 

31.  A  principal  should  exert  pressure  on  the  superintendent  in 
order  to  have  an  ineffective  teacher  removed  from  his  staff. 

32.  A  principal  should  not  have  close  relatives  on  his  teaching 
staff. 

33*  in  a  fairly  large  school  it  is  reasonable  for  a  principal  to 
ask  teachers  to  make  appointments  to  see  him  rather  than  to 
come  to  his  office  whenever  they  wish. 

34*  A  principal  should  not  do  anything  to  help  a  teacher  who  is 
having  teaching  difficulties  until  the  teacher  comes  to  him 
for  help. 

35*  A  principal  should  always  defend  his  school  against  adverse 
criticism  even  if  in  his  opinion  the  criticism  is  justified. 

36.  A  principal  should  not  concern  himself  with  the  personal 
backgrounds  of  his  teachers. 

37*  Teachers  should  not  call  the  principal  by  his  first  name. 

38.  A  principal  should  permit  teachers  to  adapt  courses  to  the 
needs  of  their  classes  in  whatever  ways  they  may  see  fit. 

39*  A  pi'incipal  should  never  resort  to  asking  certain  teachers 
in.  the  school  about  the  attitude  or  work  of  one  of  their 
colleagues  in  an  effort  to  learn  more  about  him. 


j;: 


' 


.  .  '  :  J;.  . 


-  1  ■  •  -  ■  '  '  '  . 

•  -  -  -  V. 


•  ■  .  :  :  ■  . ' 

j  i 

1  .1  "  •  ’ 

■  ij.ro 

i'l '  -  .V  V.i  i  • 

.  ;.c  i 

; 

. 

i  t 

ju  > 

■  '  ■■  ;  i  >1  i  ■  ' 

• 

11  .  .  ;  . 

218 


40.  A  principal  cannot  carry  out  his  duties  effectively  if  he 
has  close  friends  on  the  staff. 

41.  A  principal  should  receive  a  salary  considerably  above  that 
of  the  highest  paid  teacher  in  his  school. 

42.  If  a  teacher  has  a  grievance  against  the  superintendent  or 
the  board,  the  principal  should  try  to  settle  the  differences. 

43»  A  principal  may  have  to  be  critical  of  a  teacher  to  a  super¬ 
intendent  without  informing  the  teacher  about  it. 

44*  A  principal  should  restrict  his  relationships  with  staff 
members  to  the  formal  requirements  in  order  to  avoid 
preferential  treatment  of  some  teachers. 

45*  1  principal  should  not  strive  to  achieve  a  higher  social 

position  in  the  community  than  that  of  the  teachers  in  his 
school . 

46.  Teachers  should  expect  principals  to  deal  with  all  parental 
complaints  before  they  reach  the  teacher. 

47*  A  principal  should  recognize  that  there  is  greater  value  in 
running  a  school  democratically  than  in  doing  tasks  more 
quickly  by  less  democratic  means. 

48.  A  principal  should  invite  teachers  to  his  home  only  as  a 
group  if  he  invites  them  at  all. 

49*  A  principal  should  spend  more  out  of  school  time  with  other 
leaders  in  the  community  than  with  teachers. 

50.  A  principal  should  expect  teachers  to  adopt  uniform  rules 
and  regulations  for  all  classes  in  the  school. 

51.  A  principal  may  be  justified  in  dealing  directly  with 
students  who  are  giving  a  teacher  difficulty  without 
informing  the  teacher  of  his  actions. 

52.  A  principal  should  not  discuss  his  personal  problems  with 
any  of  his  teachers. 

53.  A  principal  should  try  to  spend  some  time  each  day,  either 
during  recesses  or  after  school,  in  the  staff  room  wi thins 
teachers . 

54.  A  teacher  should  not  be  expected  to  consult  the  principal 
before  administering  corporal  punishment  to  a  pupil. 


' r  J; 

. 


•  V  . 


•  •'  ■  ■' 

•  :  '  •  )  '  '  ; 

:  .  :  ; 

' '  '  i  o  .  rr  ■  /  Ir,(  1:  ,.< 


219 


55*  A  principal  should  not  hesitate  to  go  against  the  wishes  of 
his  teachers  on  a  matter  of  school  policy  if  he  considers 
it  necessary. 

56.  If  a  principal  ha,s  lunch  at  school  he  should  associate  with 
those  teachers  who  axe  also  at  school  during  the  lunch  hour 
as  one  of  their  equals. 

57*  A  principal  should  he  more  like  a  superintendent  than  like 
a  teacher. 

58.  In  delegating  some  definite  non-teaching  responsibilities  to 
teachers,  a  principal  should  give  them  full  authority  to  act 
as  they  see  fit. 

59.  A  principal  may  be  justified  in  deliberately  building  up 
tension  between  himself  and  a  teacher  who  he  believes  is 
ineffective  in  an  attempt  to  force  the  teacher  to  leave 
the  school. 

60.  A  principal  should  ignore  the  fact  that  a  staff  member  has 
submitted  incomplete  or  incorrect  credentials  to  the  board 
for  salary  purposes. 

61.  If  a  principal  has  some  minor  matter  to  discuss  with  a  teacher 
after  school,  he  should  go  to  the  teacher's  classroom  rather 
than  call  the  teacher  to  his  office. 

62.  Prior  to  issuing  reports  to  parents,  teachers  should  submit 
a,ll  comments  and  gradings  to  the  principal  for  his  approval. 

63.  A  principal  should  not  feel  obligated  to  consult  all  teachers 
on  a  school  matter  if  he  can  gain  a,  general  impression  of  the 
views  of  the  staff  by  consulting  only  a  few  teachers. 

64.  A  principal  should  be  more  concerned  with  qualifications 
than  with  teacher  preferences  when  assigning  teaching  duties. 

65.  If  a.  teacher  and  the  principal  disagree  over  a  matter  of  class¬ 
room  procedure,  it  is  reasonable  for  the  principal  to  expect 
the  teacher  to  yield  to  his  point  of  view. 

66.  Even  the  decision  of  whether  or  not  a  teacher  participates 
in  policy  development  should  be  left  to  the  teacher. 

67.  A  principal  should  not  attempt  to  gain  special  privileges 

for  the  teachers  in  his  school  by  developing  close  relationships 
with  the  superintendent  or  members  of  the  board  for  that  purpose. 


'  '  :  '  •'  >, '  j;;  f j; ■  ;  ■ 

-•  ;  -•  '  '■  ■  '  >  ;  .  W  .,.  : 

■' .  ;  '  :  .  :  .;j  f  .  /;.  (  ,  '  ,  . 

-  .  '  ■  ■ 

’  •  o'  .  ]'j i'  /;■ 

. 

^  ■'  •  ••  '  .'.  •  ■  • 

1  :  '  •  -  ■  -  -  '  ■  -  -  ...  .  ••  .  /' 

•  '  V. 

■  ’  '  ■  • 

-  •  '.  i 

.  )  j  ••  .  i  r  j : 


'  o  .'  .1  ' '  .'  .  :  i  f 


;  ‘  "  ..  '  ; 

'  1  '  .  ?  • 

■  '  .  ...  r 


■  :  - v  J-'  ...  ;  ■' .<0 

- •  ;  J'  J  j:I  .)  :  '.  .  -  »j;  ir . (:  ;  < 

•  ■  ■'  •  . '  •  ■/  :  •  .  ■  . 

-  if  •• 

;  -•  . 

•  ■  ’  '  ;  •  •  ’. 


i  1. 


■  n.  •.  .  ■]>  • 

*  •  •  •  • .  v  ■  .  J- 

■  -  ;  .  ‘. . 

•  ■  :  ^  o'  (  .  /  ’  :  i.J 

‘  ’  i  ••  'I  T  fciJ  •  '  C  !  >'•  ■  '  . 


220 


68.  A  principal  should  show  concern  for  the  personal  welfare  of 
each  staff  member. 

69.  A  principal  should  keep  his  relations  with  the  staff  on  an 
informal  level. 

70.  It  is  reasonable  for  a  principal  to  expect  new  teachers  to 
fit  themselves  into  the  traditional  ways  of  the  school. 

71.  A  principal  need  not  inform  a  teacher  against  who  there  has 
been  a  parental  complaint  if  he  has  satisfied  the  parent  who 
made  the  complaint. 

72.  A  principal  should  make  it  possible  for  those  teachers  who 
have  been  in  school  longer  to  have  a  greater  voice  in  school 
affairs. 

73*  In  "the  school  and  in  the  community  a  principal  should  act  as 
if  teachers  were  on  the  same  social  and  professional  level 
as  he  is. 

74*  The  principal  should  expect  teachers  to  consult  with  him 
before  making  any  final  decisions  about  promotions. 

75*  4  principal  should  not  hesitate  to  depart  from  official 

procedures  if  it  means  that  certain  tasks  will  be  carried 
out  more  effectively. 

76.  A  principal  should  be  willing  to  give  special  privileges 
to  teachers  who  are  very  effective  in  the  classroom. 

77*  The  principal  of  a  school  need  not  spend  as  much  time  or 
the  same  hours  at  school  as  do  the  teachers  because  the 
nature  of  his  work  is  different. 

78.  A  principal  should  encourage  teachers  to  look  to  him  for 

guidance  concerning  appropriate  modes  of  dress  and  conduct 
in  the  community. 

79*  A  principal  need  not  involve  the  staff  in  formulating  school 
policies  in  order  to  run  a  school  effectively. 

80.  A  principal  should  refuse  to  discuss  a  teacher’s  personal 
problems  with  that  teacher. 

Please  make  certain  that  you  have  responded  to  every  statement 

SECTION  B 

The  following  information  is  required  for  checking  the 

representativeness  of  the  sample.  It  will  NOT  be  used  for 

attempting  to  identify  respondents.  Please  answer  all  questions. 


^  '  v.  • '  ■  •  d  " 

r  ■'  ■  .  c  •  .  d  .  ■  ,':c  • 

1  .  xd  :l  .  o  _  t 

V.  k. 

-  >'  .  d  d  ic  : 

.  o  d;  l  /  ■.  v  ■  '  d'  •  c  v  - 

.  d:  ' 

: x  .  d d;  -  j/; c  id' 

-  ■'  -  .  X:  i. : 

u  .  '  L'  i  %  .  .  ..  .  .  •' 

-  •  ■  ■  x  x 

.  dc  V :  ■  '  d\.  -i  d  .•  d'.':  d*  :  ;c  d.  ■ 

;  .  "  x  .  r  x  d  d  d  d;  '  ,d  ; 


-  d  d  .  •  d:;  d 

...  /  c  v  x 

;  ■  ...■■■  i  .  _::i  d: .. 

:  •:  ;  •  d  ')  .•  ;  d  •  ■  <  x 

d  .  d  :  d  ■  !  :  '  d:;.  d. 

d  •  '  ,  d: 

r, 

d:  d 

<  . O  :  d  d.  dx  .  .. 


. :  ■  c 


221 


1.  Sex:  Male  ...  Female  ... 

2.  Please  check  one  of  the  following  to  indicate  your  present  age: 

Under  20  . . .  .  21-30  ...  31-40  ... 

41-50  ...  Over  50  ... 

3.  Please  check  one  of  the  following  to  indicate  your  total  years 
of  teaching  experience  to  June,  1962: 


2  or  less  . . . 

11-15  ... 


3-5  ... 

16-20  ... 


6-10  ... 
20+  ... 


How  many  complete  years  of  teacher  training  do  you  have?  Circle 
one. 


1 


7 


7+ 


Circle  each  grade  in  which  you  are  teaching  this  year: 

123456789  10  11  12 

Circle  the  grade  in  which  you  do  most  of  your  teaching.  If 
your  time  is  equally  distributed,  circle  more  than  one  grade, 

123456789  10  11  12 

If  you  teach  in  a  departmentalized  junior  or  senior  high 
school,  please  indicate  your  subject  specialty. 


8.  Including  the  present  year,  for  how  many  years  have  you  been 

teaching  in  this  school?  _  years. 

TEACHERS  ANSWER  No.  9;  PRINCIPALS  ANSWER  No.  10 

9.  For  how  many  of  the  years  in  which  you  have  been  teaching  in 

this  school  has  the  present  principal  of  the  school  held  that 
position?  (include  the  present  year)  _  years. 

10.  Including  this  present  year,  for  how  many  years  have  you  held 
the  position  of  principal: 

(a)  of  this  school _ years 

(b)  of  other  schools  _  years 

years 


Total 


I 


" 


...  •  •  •  ;  /•  •  ■ 

^  ?  h  i:  <■"'  r 


Q 


APPENDIX  B 

LEADER  BEHAVIOR  DESCRIPTION 

QUESTIONNAIRE 


22J 


LEADER  BEHAVIOR  DESCRIPTION  QUESTIONNAIRE  ITEMS 

DIRECTIONS: 

a.  READ  each  item  carefully. 

b.  THINK  about  how  frequently  the  leader  engages  in  the 
beha,vior  described  by  the  item. 

c.  DECIDE  whether  he  always,  often,  occasionally,  seldom 
or  never  acts  as  described  by  the  item. 

d.  DRAW  A  CIRCLE  around  one  of  the  five  letters  following 
the  item  to  show  the  answer  you  have  selected. 

A  =  Always  B  =  Often  C  =  Occasionally  D  =  Seldom  E  =  Never. 

1.  He  does  personal  favors  for  group  members. 

2.  He  makes  his  attitudes  clear  to  the  group. 

3.  He  does  little  things  to  make  it  pleasant  to  be  a  member  of 
the  group. 

4.  He  tries  out  his  new  ideas  with  the  group. 

5.  He  acts  as  the  read  leader  of  the  group. 

6.  He  is  easy  to  understand. 

7.  He  rules  with  an  iron  hand. 

8.  He  finds  time  to  listen  to  group  members. 

9.  He  criticizes  poor  work. 

10.  He  gives  advance  notice  of  changes. 

11.  He  speaks  in  a  manner  not  to  be  questioned. 

12.  He  keeps  to  himself. 

13.  He  looks  out  for  the  personal  welfare  of  individual  group 
members. 

14.  He  assigns  group  members  to  particular  tasks. 

15.  He  is  the  spokesman  of  the  group. 

16.  He  schedules  the  work  to  be  done. 

17.  He  maintains  definite  standards  of  performance. 


•?rur  a.L  'YoiiaoJ  .l'j  y.IJ/ioyr,  >a'i.  v/ori  J ;  . .  . : 

« 

5  •'  •  ; 

»  '  -  .  v 

‘ 


•  : -i!‘ 


- 


•:  ; 


■ 


•  ■>>  0>;  ..  l 


- 


: 


/  :  • 

■ 


' 


.  ; 

.  '  :  ■  i.‘;  :  ;?  ■  •  j;  • 


•  '  -  iJ  ' :  •  ■  J"  .  •  v  ».  ;  ; 

f  ■  v  ')•  ■  • 

„  ■  ;  .  , ,  ;  ’  *-  ’  ; 

*  -  •  ’  J.  -  O.  [%'  »  ’  ‘v  ’  / 

.  v  X'j">  :j  >  -  ,r  •.  <:•  ■  t  t 

. 

r  o  id'  v;.- Yj.  ,/  e:!o<  f.  • 

.  (  • 

-  J  ■  '  :G  ■ 

: r  j;  ■ 

.  .  >"■  <  >\.:o  v  t 

i  •  '  •  '  ■  '  '•  ' 


.  : 


224 


18.  He  refuses  to  explain  his  actions. 

19«  He  keeps  the  group  informed. 

20.  He  acts  without  consulting  the  group. 

21.  He  hacks  up  the  members  in  their  actions. 

22.  He  emphasizes  the  meeting  of  deadlines. 

23.  He  treats  all  groups  members  as  his  equals. 

24.  He  encourages  the  use  of  uniform  procedures. 

25.  He  gets  what  he  asks  for  from  his  superiors. 

26.  He  is  willing  to  make  changes. 

27.  He  makes  sure  that  his  part  in  the  organization  is  understood 
by  group  members. 

28.  He  is  friendly  ajid  approachable . 

29.  He  asks  that  group  members  follow  standard  rules  and  regulations. 

30.  He  fails  to  take  necessary  action. 

31.  He  makes  group  members  feel  at  ease  when  talking  with  them. 

32.  He  lets  group  members  know  what  is  expected  of  them. 

33 «  He  speaks  ac  the  representative  of  the  group. 

34*  He  puts  suggestions  made  by  the  group  into  operation. 

35*  He  sees  to  it  that  group  members  are  working  up  to  capacity. 

36.  He  lets  other  people  take  away  his  leadership  in  the  group. 

37 •  He  gets  his  superiors  to  act  for  the  welfare  of  group  members. 

38.  He  gets  group  approval  in  important  matters  before  going  ahead. 
39*  He  sees  to  it  that  the  work  of  group  members  is  coordinated. 

40.  He  keeps  the  group  working  together  as  a  team. 


■■  ;  ‘  •’ 

' 

■  . 

■ 

.  \  ••  J  :  ■ 

• 

■ '  -  ■ 

:  •  ' 

» 

rci 

• 

- 

* 

» 

• 

'  "■  ■  '  ,  &  :  J  • 

-  i 

'  . 

.  .  .  .  . . 

. 

'  i 

' 

• 

• 

■ 

..  ■  ■  'j  ■ 

. 

' 

>■  /  •  ;j  •  :J*  :  . 

• 

«  . :  : 

r  ' 

* 

•  -  •  •  ■■■ 

■  ■  ■  ■:  f  -..  : 

* 

u  V.'  i  . 4 :  i 

. 

. 

- 

» 

1 

.  ' 

.  (  : 

’’  ... 

,  >•  ■  •  ':  *  /  .  .  . 
c;  t  .  : 

'  v  J 

•  •  •  ' 


APPENDIX  C 

INSTRUCTIONS  TO  PRINCIPALS 


PRINCIPAL’S  QUESTIONNAIRE 


226 


INSTRUCTIONS  TO  PRINCIPALS 

1.  Please  hand  one  of  the  enclosed  envelopes  to  each  of  the  teachers 
in  your  school,  I  would  be  very  grateful  if  you  would  assure  the 
teachers  that  this  study  is  being  carried  out  with  your  coopera¬ 
tion  and  that  you  have  no  objection  to  the  questionnaires  which 
are  being  used. 

2.  Designate  a  staff  member  to  whom  teachers  will  hand  the  sealed 
envelopes  containing  completed  questionnaires.  Since  one  of  the 
questionnaires  refers  specifically  to  you  as  principal  of  the 
school,  some  of  the  teachers  might  feel  more  comfortable  about 
completing  it  if  a  staff  member  rather  than  the  principal  were 
in  charge  of  handling  completed  questionnaires.  The  staff 
member  who  is  in  charge  of  this  may  wish  to  keep  a  record  of  the 
teachers  who  have  handed  him  or  her  sealed  envelopes  but  no 
identifying  mark  should  be  placed  on  the  envelopes  themselves. 

5.  Plea.se  encourage  teachers  to  complete  the  questionnaires  within 
a  week  after  they  have  been  distributed.  When  all  the  envelopes 
have  been  returned  to  the  designated  staff  member  they  should  be 
placed  in  the  large  self-a.ddressed  envelope  and  mailed  to  me.  I 
should  have  left  enough  money  with  you  to  cover  the  cost  of 
mailing;  it  is  not  necessary  for  you  to  refund  unused  postage. 


4-  Please  complete  the  PRINCIPAL'S  QUESTIONNAIRES: 

You  ac  principal  are  requested  to  complete  the  two  question¬ 
naires  contained  in  the  envelope  marked  "Principal".  These  two 
questionnaires  are  to  be  completed  in  conjunction  with  each  other. 
After  you  have  responded  to  an  item  on  the  expectations  question¬ 
naire  you  are  required  to  indicate  on  the  second  questionnaire  how 
difficult  it  was  for  you  to  decide  upon  your  response.  The 
principaf 's  questionnaires  form  an  important  part  of  this  study  and 
it  is  essential  that  both  are  completed  in  full. 

When  you  have  completed  the  questionnaires  place  them  in  the 
envelope  which  has  been  provided  and  seal  it.  This  envelope  can  then 
be  placed  with  the  teachers'  envelopes  for  mailing  to  me. 


5.  Many  thanks  to  you  and  the  members  of  your  staff  for  assisting 
me  in  carrying  out  this  research. 


,  •  . 

;  '  •  1 

t 


'■  :.i  i liv  ■yvcm.oyrj'  mr;nV  o.f  'xsJnocn  &  9d£xrj,>j:ay(i 

•  $  .  . 

<■  . . : 

i?,i 

-  ■  .r  ‘  ' •  .  ft.;-  ;  :r  ■:> 

... 

-  '• 
v.‘-  .  .  ■  .  •  '.  o.: 

•  -  . 

•'  '  -  •  \r  -  I  v  .  V 


'  .  i 


-v-  .  ..  r  ;  '  . x  ■  '<  ■ ,  •  ■ 

. :  1  , 

•  . 

■  1 


. 


.  £  <  ; 
V  '  •  ■ 


PRINC IPALS «  QUESTIONNAIRE 


In  responding  to  the  items  in  the  expectations  questionnaire  you 
will  probably  experience  varying  degrees  of  difficulty  in  deciding 
upon  whether  or  not  you  agree  with  an  item  and  the  extent  to  which 
you  agree  or  disagree  with  it.  This  form  enables  you  to  indicate 
how  difficult  it  was  for  you  to  decide  about  your  responses. 

Please  complete  this  form  at  the  same  time  as  you  complete  the 
expectations  questionnaire  according  to  the  following  directions: 

a.  READ  each  item  in  the  expec tatiofip  questionnaire. 

b.  DECIDE  whether  you  strongly  agree,  agree,  disagree,  or 
strongly  disagree  with  the  statement. 

c.  CIRCLE  one  of  the  five  responses  following  the  item  to 
indicate  your  reaction  to  it. 

d.  ASK  yourself  how  difficult  it  was  for  you  to  decide  upon 
your  response. 

e.  CIRCLE  one  of  the  five  numbers  5>  4>  2,  or  1  in  the 

corresponding  item  on  this  form  to  indicate  how  difficult 
it  was  for  you  to  decide.  Think  of  the  numbers  as  being 
on  a  scale  like  this: 


4 


Moderately 

Difficult 


2 


Not  At  All 
Difficult 


1 


Difficult 


The  scale  is  repeated  on  the  following  pages  to  help  you 
remember  it. 


Please  CIRCLE  ONE  response  for  each  item. 


' 

'  ■  -  " '  •  ;  ;  •  v  o:  ■  .  ■  \ 

; '  '  ■  -  -  •  :  •  •  c  . 

d  ■  to'  .  c  sjss! 

• ;  ;  .  •'  ■  •  :  ■ '  ;  ■  •  ■  ..  ;  -  >-  ■ 

(  '  • 

• ;  -  :  .  1 ! 

■ 

*  ^  -  .  •  '  I .  . 

■'  ;  V :  >ri  ' ;  r 

. 


.  ■ 


228 


-  2  - 


HOW  DIFFICULT  WAS  IT  FOR  YOU  TO  DECIDE? 

5 _ 4 _ 3 _ 2 _ 1 

Very  Moderately  Not  At  All 

Difficult  Difficult  Difficult 


Please  CIRCLE  ONE  number  in  each  item  after  you  have  responded  to 
the  corresponding  item  in  the  expectations  questionnaire. 


(1) 

5 

4 

3 

2 

1 

(ii) 

5 

4 

3 

2 

1 

(2) 

5 

4 

3 

2 

1 

(12) 

5 

4 

3 

2 

1 

(3) 

5 

4 

3 

2 

1 

(13) 

5 

4 

3 

2 

1 

(4) 

5 

4 

3 

2 

1 

(14) 

5 

4 

3 

2 

1 

(5) 

5 

4 

3 

2 

1 

(15) 

5 

4 

3 

2 

1 

(6) 

5 

4 

3 

2 

1 

(16) 

5 

4 

3 

2 

1 

(7) 

5 

4 

3 

2 

1 

(17) 

5 

4 

3 

2 

1 

(8) 

5 

4 

3 

2 

1 

(18) 

5 

4 

3 

2 

1 

(9) 

5 

4 

3 

2 

1 

(19) 

5 

4 

3 

2 

1 

(10) 

5 

4 

3 

2 

1 

(20) 

5 

4 

3 

2 

1 

Next  Page  Please 


' 

-  i  -  ■  ! 


.  ....  .  :: 

X'-  : 

■ 

■  2 

■ 

- 

(Very  Difficult 


5  4  3  2  1  Not  At  All  Difficult) 


(21) 

5 

4 

3 

2 

1 

(39) 

5 

4 

3 

2 

1 

(22) 

5 

4 

3 

2 

1 

(40) 

5 

4 

3 

2 

1 

(23) 

5 

4 

3 

2 

1 

(41) 

5 

4 

3 

2 

1 

(24) 

5 

4 

3 

2 

1 

(42) 

5 

4 

3 

2 

1 

(25) 

5 

4 

3 

2 

1 

(43) 

5 

4 

3 

2 

1 

(26) 

5 

4 

3 

2 

1 

(44) 

5 

4 

3 

2 

1 

(27) 

5 

4 

3 

2 

1 

(45) 

5 

4 

3 

2 

1 

(28) 

5 

4 

3 

2 

1 

(46) 

5 

4 

3 

2 

1 

(29) 

5 

4 

3 

2 

1 

(47) 

5 

4 

3 

2 

1 

(30) 

5 

4 

3 

2 

1 

(46) 

5 

4 

3 

2 

1 

(31) 

5 

4 

3 

2 

1 

(49) 

5 

4 

3 

2 

1 

(32) 

5 

4 

3 

2 

1 

(50) 

5 

4 

3 

2 

1 

(33) 

5 

4 

3 

2 

1 

(51) 

5 

4 

3 

2 

1 

(34) 

5 

4 

3 

2 

1 

(52) 

5 

4 

3 

2 

1 

(35) 

5 

4 

3 

2 

1 

(53) 

5 

4 

3 

2 

1 

(36) 

5 

4 

3 

2 

1 

(54) 

5 

4 

3 

2 

1 

(37) 

5 

4 

3 

2 

1 

(55) 

5 

4 

3 

2 

1 

(38) 

5 

4 

3 

2 

1 

(56) 

5 

4 

3 

2 

1 

Next  Page  Please 


. 

v  •  • 

. 

. 

. 

, 

' 

(56) 

(57) 

(58) 

(59) 

(60) 
(61) 
(62) 

(63) 

(64) 

(65) 

(66) 

(67) 

(68) 

(69) 

(70) 

(71) 


-  4  - 


(Very  Difficult  3_ 


321  Not  At  all  Difficult) 


5  4  3  2  1 


(72) 


5  4  3  2  1 


5  4  3  2  1 


(73) 


5  4  3  2  1 


5  4  3  2  1 


(74) 


5  4  3  2  1 


5  4  3  2  1 


(75) 


5  4  3  2  1 


5  4  3  2  1 


(76) 


5  4  5  2  1 


5  4  3  2  1 


(77)  54321 


5  4  3  2  1 


(78) 


5  4  3  2  1 


5  4  3  2  1 


(79)  54321 


5  4  3  2  1 


(80) 


5  4  3  2  1 


5  4  3  2  1 


5  4  3  2  1 


5  4  3  2  1 


5  4  3  2  1 
5  4  3  2  1 
5  4  3  2  1 
5  4  3  2  1 


230 


-  -  .  . . . . 


; 

. 

. 

r: 

. 

,  : 

■ 

.  ?: 

APPENDIX  D 


QUESTIONNAIRE  RETURNS  BY  SCHOOLS 


2J2 


TABLE  XL 

QUESTIONNAIRE  RETURNS  BY  SCHOOLS 


School 

Code 

Number  of 
Teachers 

Returns 

School 

Code 

Number  of 
Teachers 

Returns 

21 

9 

9 

'  i 

43 

14 

14 

51 

9 

8 

6 

'14 

13 

15 

9 

7 

60 

14 

13 

5 

10 

10 

54 

14 

12 

29 

10 

10 

25 

14 

12 

51 

10 

10 

49 

15 

15 

42 

10 

10 

33 

15 

14 

57 

10 

10 

27 

15 

12 

48 

10 

9 

16 

15 

12 

55^ 

10 

8 

59* 

15 

8 

50 

10 

7 

4 

16 

16 

17* 

10 

6 

45 

16 

15 

8 

11 

11 

41 

16 

13 

18 

11 

11 

36 

17 

16 

19 

11 

11 

58* 

17 

6 

26 

11 

11 

62* 

17 

0 

28 

11 

11 

24 

18 

18 

32 

11 

11 

12 

18 

17 

37 

11 

11 

34 

18 

17 

53 

11 

11 

38 

19 

19 

46 

11 

10 

35 

19 

18 

22 

11 

10 

20 

19 

18 

30 

11 

10 

1 

20 

20 

44 

11 

8 

10 

20 

16 

7 

12 

12 

52 

20 

16 

9 

12 

12 

2 

21 

21 

23 

12 

12 

13 

21 

21 

40 

12 

12 

3 

22 

21 

39 

12 

10 

ll 

22 

19 

47 

13 

12 

14 

24 

24 

56, 

13 

9 

61 

13 

0 

Totals 

873 

765 

"X" 

These  schools  were  eliminated  from  the  study. 


;  -  ;  '  ,  .  r 

. 

; 

:  -  : 

.  .  :  ■  :  ;  •  .  : 

- 

* 

Y" 

. 

■ 

. 

, 

.. 

. 

:  :  :  : 

. 

.  .  -  ? 

APPENDIX  E 


NORMALIZED  DATA  BY  SCHOOLS 


234 


TABLE  XLI 

NORMALIZED  DATA  BY  SCHOOLS 


School  Considera- 
Code  tion 


Initiating  Ambi-  Consensus  Teacher 
Structure  valence  Principal 

Agreement 


1 

34 

62 

50 

48 

64 

2 

34 

65 

71 

50 

58 

3 

51 

42 

37 

41 

50 

4 

48 

52 

49 

41 

41 

5 

66 

52 

63 

48 

58 

6 

44 

49 

43 

44 

55 

7 

60 

45 

41 

41 

36 

8 

48 

41 

67 

59 

73 

9 

37 

49 

47 

52 

43 

10 

54 

47 

52 

48 

50 

11 

54 

69 

42 

50 

61 

12 

43 

33 

56 

41 

40 

13 

44 

37 

58 

62 

53 

14 

54 

45 

49 

63 

58 

15 

40 

33 

45 

54 

43 

16 

40 

26 

60 

52 

52 

19 

41 

52 

47 

52 

51 

19 

29 

37 

54 

62 

51 

20 

51 

55 

42 

73 

58 

21 

48 

60 

39 

44 

38 

22 

56 

52 

45 

46 

56 

23 

63 

45 

45 

41 

41 

24 

48 

41 

52 

38 

35 

25 

56 

55 

53 

56 

58 

26 

51 

52 

52 

52 

52 

27 

43 

58 

60 

35 

41 

28 

46 

55 

60 

54 

39 

29 

51 

60 

48 

38 

55 

30 

58 

62 

54 

56 

45 

31 

51 

42 

50 

70 

45 

32 

56 

55 

56 

63  ' 

62 

33 

51 

58 

32 

52 

50 

34 

37 

41 

36 

35 

48 

35 

40 

65 

47 

50 

55 

36 

38 

55 

43 

35 

37 

37 

66 

45 

35 

31 

48 

38 

47 

47 

39 

46 

47 

39 

58 

47 

26 

59 

48 

40 

58 

37 

58 

56 

45 

41 

51 

45 

63 

54 

51 

42 

54 

53 

71 

48 

67 

Continued  on  following  page 


••'.V  . 

'  •  ■  •  '  :  :  • 


-1  :  .. 

l 

' 

. 

' 

' 

. 

- 

■ 

1 

43 

44 

45 

46 

47 

48 

49 

51 

52 

53 

54 

55 

56 

57 


TABLE  XLI  (Continued) 


235 


Considera¬ 

tion 

Initiating 

Structure 

Ambi¬ 

valence 

Consensus  Teacher 

Principal 
Agreement 

29 

52 

52 

46 

53 

71 

65 

63 

26 

47 

71 

58 

41 

59 

69 

44 

49 

63 

38 

31 

48 

49 

32 

46 

45 

62 

65 

43 

54 

61 

48 

40 

54 

56 

64 

41 

37 

49 

52 

26 

62 

45 

52 

59 

41 

63 

74 

58 

43 

58 

60 

49 

57 

44 

64 

56 

58 

46 

56 

46 

51 

33 

57 

65 

53 

44 

49 

47 

50 

53 

46 

55 

39 

67 

47 

J.  . 

•  I 

' 

. 

: 

j  :  i  .  .  .  :  :  .