A Comparison of Two Attitude Scaling Techniques

49 Pages·2.078 MB·English

by  Niven, Jarold R

PURDUE UNIVERSITY
THIS IS TO CERTIFY THAT THE THESIS PREPARED UNDER MY SUPERVISION
Jar old H, Niven
A Comparison of Two Attitude Scaling Techniques
ENTITLED ____________________________
COMPLIES WITH THE UNIVERSITY REGULATIONS ON GRADUATION THESES
AND IS APPROVED BY ME AS FULFILLING THIS PART OF THE REQUIREMENTS
FOR THE DEGREE OF
Doctor of Philosophy
Professob in Charge of Thesis
y Head of School or Department
19^~/
TO THE LIBRARIAN
THIS THESIS IS NOT TO BE REGARDED AS CONFIDENTIAL.
'C' yv/-
PROFESS OH HT CHARGE
REGISTRAR FORM 10—7.47—1M

A COMPARISON OF TWO ATTITUDE SCALING TECHNIQUES
A Thesis
Submitted to the Faculty
of
Purdue University
by
Jarold R, Niven
In Partial Fulfillment of the
Requirements for the Degree
of
Doctor of Philosophy
June, 1951
ProQuest Number: 27714294
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
uest
ProQuest 27714294
Published by ProQuest LLC (2019). Copyright of the Dissertation is held by the Author.
All rights reserved.
This work is protected against unauthorized copying under Title 17, United States Code
Microform Edition © ProQuest LLC.
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106 - 1346
ACKNOWLEDGEMENTS
I am deeply indebted to Dr. C. H. Lawshe, not only for the
guidance and encouragement furnished me in the preparation of this
thesis, but also for his generous aid and direction thruout my
graduate work.
Appreciation is also due the members of my advisory committee.
Professors E. J. McCormick, N. 0. Kephart, K. S. Davenport and
W. V. Owen for their helpful suggestions and assistance.
I am grateful to Mr. Ivan Willis, Vice President of Industrial
Relations, Mr. Ernest Reed, Manager of Education and Training and
Mr. Wendell Wood, General Supervisor of Research and Testing, all of
the General Office, International Harvester Company, and to Mr. Paul
Johnson, Works Manager, Mr. H. A. Torgerson, Assistant Works Manager,
and Mr. Guy Myers, Training Director of the Louisville Works, Interna­
tional Harvester Company, whose cooperation and assistance made this
study possible.
ABSTRACT
Niven, Jarold R*, A COMPARISON OF TWO ATTITUDE SCALING TECHNIQUES,
June, 1951, 39 pages, 6 tables, 14 titles in the bibliography, appendix,
A research problem in applied psychology which makes a statistical ev­
aluation of the Reciprocal Averages Scaling Technique and Guttman1 s
Cornell Scale Analysis,
The objective quantification of attitude measurement has been a
problem which has been attacked with numerous methods. It was the pur­
pose of this study to investigate two of these methods.
Two hundred and ninety-five manufacturing supervisors were admin­
istered an attitude questionnaire consisting of 29 items. The population
was divided into a primary group and a hold out group. Both scaling
techniques were applied to the primary group and tried out in the hold
out group.
Twelve items were selected by the Reciprocal Averages Technique and
ten by the Cornell Technique, nine of which were common. Lower bound
reliability coefficients for the two methods ranged from ,78 to ,84 and
while higher for the Reciprocal Averages Technique were not significantly
different. The obtained reliability coefficients did not differ signi­
ficantly from the primary to the hold out groups. Total scores for the
two methods correlated .93 in the primary group and .89 in the hold out
group.
The Cornell Technique resulted in an eleven point scale and the
Reciprocal Averages Technique provided a 56 point scale which might
ii
prove to be more dis criminative *
The items and weights resulting from the Reciprocal Averages
Technique were applied to a second population of 243 supervisors who
were identified as to the divisions in which they worked. An analysis
of variance of mean scores for the various divisions provided an F ratio
significant beyond the 1% level of confidence.
iii
TABLE OF CONTENTS
PAGE
INTRODUCTION 1
HISTORY AND BACKGROUND 1
PROCEDURE 3
Questionnaire 3
Population 4
Reciprocal Averages Technique 4
Cornell Scale Analysis Technique 7
Comparison of Methods 11
RESULTS 12
Reciprocal Averages Technique 12
Cornell Scale Analysis Technique 12
Comparison of Methods 13
Comparison of Administration Effort 14
APPLICATION 14
SUMMARY AND CONCLUSIONS 15
BIBLIOGRAPHY 17
APPENDIX A
Further Analysis of Items hy the Reciprocal Averages
Technique 19
APPENDIX B
Tables 22
APPENDIX C
Questionnaire 29
APPENDIX D
Code Data for Punching IBM Cards 38
iv
LIST OF TABLES
TABLE PAGE
1« Fourth Approximation Scale Values Established by the
First Analysis of the Reciprocal Averages Technique 22
2, Coefficients of Reproducibility, Plus Percentage Ratios,
Item Response Combinations and Item Cutting Points
Obtained by the Cornell Scale Analysis Technique 23
3# Mean Scores by Division for the 243 Supervisors Admin­
istered the Second Questionnaire 24
4, Analysis of Variance of Mean Scores by Divisions for
the 243 Supervisors Administered the Second Questionnaire 25
5. Third Approximation Scale Values Established by the
Second Analysis of the Reciprocal Averages Technique 26
6* Fifth Approximation Scale Values Established by the
Third Analysis of the Reciprocal Averages Technique 27
v
A COMPARISON OF TWO ATTITUDE SCALING TECHNIQUES
The selection of items for the measurement of attitudes and the
objective quantification of such measurement has been a problem which
has been subject to considerable research and investigation. In recent
years, several statistical methods have been developed for the purpose
of selecting and scaling attitude measuring items. It is the purpose
of this study to investigate and compare two of those methods, the
Reciprocal Averages Scaling Technique and Guttman*s Cornell Scale
Analysis.
HISTORY AND BACKGROUND
The early work of attitude scaling was done by Thurstone (14)
who used the method of equal appearing intervals. Scale values are
determined by finding the median values of items as they are placed on
a continuum by a number of judges. The interquartile range of judgments
is used as a measure of ambiguity. Items are chosen in terms of equal
spacing on the continuum and the least amounts of ambiguity.
Likert (11) departs from this method in that items are assigned
predetermined weights ranging from 0 to 4> are administered to a trial
population, and final selection is on a basis of internal consistency.
The technique developed by Guttman (6,8) is used to investigate a
set of items in order to ascertain whether or not the items as such are
attributes of a universe of qualitative data or content. To the extent
that the items are unidimensional, it is possible to formulate a scale
1