THE \FALIDAfiQN (3? THE MECHKEAN SfATE M~SCALES
W [Ti-i COLiEGE‘. FRESHMEN

Thesis {tar 1‘2‘28 Bearea a? P31. 1).
M§CHEGAN S‘a‘AT‘c‘ UNWERSE‘EY'
fia‘aafy' 3.: Hayden
”i953

  
 
 
 
   
 
 
 
  
 
 
 
  
    
  
 
  

 
 

LIBRARY

Michigan State
University

This is to certifg that the
thesis entitled

0N OF THE MlCHIGAN STATE M-SCALES

THE VALIDATI
WITH COLLEGE FRESHMEN

presented bg '

Mary L. Hayden

owards tulfillment

has been accepted t
of the requiremen

Education

ts for

M— deqree in

   

Major professor

Dam August 8, 1963

 

  

L I B R A R Y
Michigan State
University

   
    

 

 

ABSTRACT

THE VALIDATION OF THE MICHIGAN STATE M-SCALES
WITH COLLEGE FRESHMEN

by Mary L. Hayden

This study was concerned with(l) the predictive
efficiency of an objective measure of academic motiva-
tion (M-Scales, Form C) when applied to a general col-
lege population, and(2) the improvement in the precision
of prediction of academic achievement (GPA) attained by
the addition of the M-Scales to an aptitude predictor
(CQT-Total).

The sample, 330 males and 367 females, was selected
from the total population of first-quarter freshmen at
Michigan State University in 1962. An academic aptitude
measure (CQT), academic motivational measure (M-Scales),
and academic achievement measure (GPA) were obtained for
each student.

The analyses were carried out independently for the
male and the female samples and invfilved the following:
(D reliability estimates were calculated using‘Hoyt's
analysis of variance technique;(2) Pearsonian correla-
tions were computed to assess the relationship of the

M-Scales to academic achievement;(3) multiple correlations

 

 

 

 

 

 

Mary L. Hayden

were computed to assess the improvement in prediction
of GPA attained by the addition of the M—Scales to the
CQT.

For the male sample, a significant, low-positive
relationship was found between the G801, WRL, M—Total
scores and the grade point criterion. There was no
Significant relationship between the male HTI scores
and academic achievement. There was no significant
relationship between the sub- and total M-Scales
scores and academic achievement (GPA) for the female
sample.

The "t" testsof significance from zero for the
M-Scales beta weights in predicting GPA were not
significant. Addition of the M-Scales scores to the
CQT measure in a multiple regression equation did not
result in an increase in precision of prediction of
the grade point criterion.

It was concluded that at this stage of deve10p4
ment the experimental M-Scales have little utility

for college pOpulations.

.. .I‘IHWIW' It ".x ‘1'

I,’

tr.
loiq‘l

 

 

 

THE VALIDATION OF THE MICHIGAN STATE M—SCALES

WITH COLLEGE FRESHMEN
By

Mary E? Hayden

A THESIS

Submitted to
Michigan State University
in partial fulifllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
College of Education

1963

 

 

 

 

 

 

 

 

ACKNOWLEDGMENT

The writer wishes to acknowledge the assistance
of the following people, as well as others, who con-
tributed to the progress of the dissertation.

To Dr. William W. Farquhar, committee chairman,
for his continued interest and assistance in the com-
pletion of this dissertation.

To Dr. Bill L. Kell who contributed much to the
progress of my doctoral program and professional pre-
paration.

To Dr. Carl Gross and Dr. Bufford Stefflre for
serving on the guidance committee.

To Dr. Arvo Juola of the Evaluation Services for
making the subjects of the study available for research
purposes.

To Margie Williams, James De Jong, and John Gordon
for their assistance in the statistical analysis.

To my collegues, Dorothy Frayer and Rod Hilsinger,
for their assistance in the data collection.

And lastly, to the freshman students of Michigan
State University, without whose cooperation this study

would not have been possible.

11

 

 

 

 

 

TABLE OF CONTENTS

ACKNOWLEDGMENTS.
LIST OF TABLES
LIST OF FIGURES.
Chapter
I. THE PROBLEM .

The Statement of the Problem

Importance of the Problem . . .
Rationale of the Study .
The Research Hypotheses . .
Organization of the Study . . .

II. REVIEW OF THE LITERATURE.

Special Curriculum Groups
Freshmen Students .
Discrepant Achievers
Discussion of Selected Studies.
Summary

III. DESIGN AND METHODOLOGY

Population
Instrumentation.

Data Collection Procedures
Analysis Procedures

Null Hypotheses.

Summary .

IV. ANALYSIS OF DATA

Reliability Analysis
Correlational Analysis
Multiple Correlation .
Summary .

V. SUMMARY, CONCLUSIONS, AND IMPLICATIONS

Conclusions .
Discussion of Results. .
Implications for Further Research.

111

Page
ii

vi

N] O\O\#I\JI—‘ H

 

 

 

 

 

 

 

Chapter Page

BIBLIOGRAPHY............65
APPENDIX. 0 n o n o o e o o o o I O 67

iv

 

 

O

..»....: .W¥.u W.‘te
.91 31".
,\ J,‘ r

e

. 4r,

 

 

 

 

 

 

 

U.)

LIST OF TABLES

Page

Summary of Hoyt's Analysis of Variance
Reliability Estimates for the M-Scales . 37

Intercorrelations Among Achievement,
Aptitude, and M-Scales . . . . . . #0

Two Variable Estimates of Grade Point
Average Achievement Criterion with
Corresponding Beta Weights and Multiple

Correlations. . . . . . . . . . 44

 

 

LIST OF FIGURES

Figure Page

1. Distribution of GSCI Raw Scores for

Males (N=50) . . . . . . . . . . 46
2. Distribution of WRL Raw Scores for

Males (N=50) . . . . . . . . . . 47
3. Distribution of HTI Raw Scores for

Males (N=50) . . . . . . . . . . 48
4. Distribution of M-Total Scores for

M8195 (N=50) o e e a o e o o o 0 1+9
5. Distribution of GSCI Raw Scores for

Females (N=50) l o n e o a a o a 50
6. Distribution of WRL Raw Scores for

Females (N=50) . . . . . . . . . 51
7. Distribution of HTI Raw Scores for

Females (N=50) . . . . . . . . . 52

8. Distribution of M-Total Scores for
Females (N=50) . . . . . . . . . 53

vi

 

 

 

 

 

J. .., ,
m . ...-,... a v... ..
.. u ) use 7.
I 7...- a... ,.WHHU.3._.lm...g{.. ..
..4,.!..f. .gfiw/u . a

i ,. 4 . 1..

a . :
. . G. . . .
t I l

 

 

 

CHAPTER I

THE PROBLEM

During the past decade an increasing volume of
educational research has been directed to the study
of academic achievement of college students. Fishman
and Pasanella list a total of 70 studies, published
and unpublished, completed by 44 colleges which used
both non-intellective and intellective characteristics
as predictors of college achievement. These studies
constitute 12 per cent of the total of 580 college
admission-selection studies completed in the period

1949—1959.1

The Statement of the Problem

 

While the major effort of such research has
been focused upon the relationship of intellective
characteristics to academic achievement, an increasing
number of researchers have investigated the relation—
ship of non—intellective characteristics to academic
achievement. In contrast to research on the intellec—

tive characteristics of college students, research on

 

1J. A. Fishman and A. K. Pasanella, "College
Admission-Selection Studies," Rev. Educ. Research,
30 (1960), pp. 298-310. *-

 

 

 

 

 

Ill" !
v
.
_
n
I.
_
x. . a u. ... . . O . a 4» . . r
_\ z . . t .e . .,e .. .d h. .. v . I . _.I . .i . .
a ll 11. Me I.” . «Time, ‘4. _... Ta 7. .. u . ..
. - .. _ : i .. ..u._w...2.l, .. . _ . . .e.
. « .. 3 ~u. .1. .. l . i
I .. o L.. t I. . . kl. .. I. 0..
. t s

 

 

the non-intellective characteristics has long been

confronted with a number of methodological problems.

Perhaps most serious of the problems is the
lack of theoretically—based objective instruments
which have been constructed specifically for use in
assessing the relationship of motivational factors
to achievement in an academic setting. In the absence
of such instrumentation, the practice has been to use
objective and projective personality measures developed
primarily for other purposes.

The Michigan State M-Scales, an objective re-
search measure of academic motivation, was recently
developed and published by Farquhar and associates.2
The purpose of the present research is to determine
the predictive validity of the Michigan State M-Scales,
Form C, in the academic achievement of a college popu-
lation. Sample items of-the‘M-Scales may«be found
in the Appendix.

Importance of the Problem
College and university admissions personnel have

traditionally used past achievement records and intel-

 

2W. W. Farquhar, A Comprehensive Study of the
Motivational Factors Underl in Achievement of Eleventh
Grade RI h School Students, Research Project No. 836
{8358}, Supported 5y tHe U. S. Office of Education, in
cooperation with Michigan State University, 1959.

 

 

 

all!!!
TIT“!

 
 

3

lective criteria in their selection procedures. In
addition to their use as admissions criteria, these
indices of achievement and aptitude are being more
widely used in the educational guidance of students
with the rapid influx of students coming from hetero-
geneous educational backgrounds.

Although considerable refinement of academic
aptitude instrumentation has led to some improvement
in the precision of prediction, a good part of the
variance in achievement of college students is still
unaccounted for. Borow agrees with other investiga-
tors that the greater part of the variance among
college students in academic achievement is unexplained
by the predictive indices in use. He concludes that a
consequent and considerable margin for error exists in
the selection process.3

The validity of the assumption that intellective
criteria (aptitude tests) alone are effective predic-
tors of college achievement has been Opened to question
with the appearance of the conceptual framework of
over- and under-achievement and the findings of studies
focused upon college-level discrepant achievers. Non-

intellective characteristics, particularly that of

3H. Borow, "Current Problems in the Prediction
Of College Performance," J. Amer. Assoc. Coll. Registr.,

22 (1946), pp. 14-26.

 

 

academic motivation, are receiving greater emphasis

in empirical studies of academic achievement and
being cited as determining variables in the academic
achievement of college students.

Attempts to reduce the unexplained variance
in the prediction of academic achievement by means
of academic motivation are directly dependent upon
the availability of instruments which have validity
for such purposes. The value of such an instrument
would derive from increased precision in selection
of students for college entrance and more efficient

educational-vocational counseling practices.

Rationale of thew§tudy

Farquhar and associates have defined academic
motivation as "a combination of forces which initiate,
direct, and sustain behavior toward a scholarly goal."“
Motivation is conceptualized as a variety of forces
which can take the form of multi—factor causality.

The authors of the Michigan State M-Scales
hypothesized that academic motivation is a personality

complex or syndrom which consists of the following

 

uWilliam W. Farquhar, Motivation Factors Related

to Academic Achievement: Final Report of COOperative
Research Project No. 846, College of Education (East

Lansing, Michigan: Michigan State University, 1963),

P- 9-

 

 

 

     

non-intellectual variables: (1) need for academic
achievement, (2) self concept, (3) academic person—
alities factors.

Evidence that the above personality variables
can be measured and objectively quantified has already
been demonstrated in the research project centered
upon the relationship of academic motivation to the
academic achievement of eleventh grade students.

The Michigan State M-Scales significantly discriminated
between statistically defined (two-stage regression
technique) discrepant achievers on such variables.5

For the purposes of the present study, academic
motivation, a non-intellective factor, is being con-
sidered as having value in reducing the unexplained
variance of achievement prediction at the college
level. Academic motivation, as measured by the Michi-
gan State M-Scales, has been demonstrated as having a
positive relationship to academic achievement for
special groups in a high school population. It is
prOposed in this study to determine the predictive
validity of the M—Scales for a general pOpulation of

college freshmen.

 

. 5Farquhar, A Comprehensive Study of The Motivational
Factors underlying_Achievement of Eleventh Grade
High School Students. op. cit. (1959).

.
1.. '._‘ _..' 1"- ‘-
'..nuu\u‘j:-. “W

 

it'l‘ ‘,'
”Eh-i. 1:, J

I I ~, ,' abh‘fllui!‘ ft

 

I
.
(
v
.-
. .

 

 

 

 

 

 

 

The Research Hypotheses

The research hypotheses to be tested in the

study are:

I. An objective assessment of academic moti-
vation will have validity in estimating
first quarter grade point averages for
college freshmen.

II. An objective assessment of academic moti-
vation, when added to a measurement of
academic aptitude, will increase the pre-
cision of estimating first quarter grade
point averages for college freshmen.

Organization of the Study

The organization of the dissertation is as
follows: In Chapter II a review of the research
related to the investigation is presented together
with a consideration of the limitations of such
research. The general design of the study, including
a description of the construction of the Michigan
State M-Scales, sample selection, instrumentation,
data collection procedures, and the techniques used
in analyzing the data, is presented in Chapter III.
The analysis of the data is presented in Chapter IV.

The summary, conclusions, and implications for fur-

ther research are presented in Chapter V.

 

 

 

 

 

 

 

 

CHAPTER II

REVIEW OF THE LITERATURE

A review of the literature on the relationship
of non-intellective variables and college achievement
for the period 1952-1962 gives evidence of the variety
of personality instruments used in this research area.
The studies will be presented, according to the type
of college population investigated, together with a

discussion of the limitations of the studies.

Special Curriculum Groups

 

Klugh and Bendig (1955), Mitchell (1961), Weiss,
Wertheimer, and Groesbeck (1959) investigated the rev
lationship between non—intellective variables and
academic achievement in special curriculum groups. A
review of the three studies, including the investiga—
tor's statement of purpose, sample, design, and findings,

is presented followed by a critical analysis.

Klugh and Bendig

 

Klugh and Bendig investigated the relationship
of non-intellectual variables as measured by the Tay-

lor Manifest Anxiety Scale and the Hr (Honor Point

7

 

 

 

Ratio) Scale of the California Personality Inventory
to academic achievement.1

Sample.--One hundred eighty-four men and women
students enrolled in introductory psychology classes
at the University of Pittsburgh during the fall se-
mester of 1954-1955.

Design.--The Taylor Manifest Anxiety (TMA), Hr
scale, and American Council on Education Psychologi-
cal Examination (ACE) were administered to all sub—
jects. Raw scores on the three tests were converted
to stanines on the basis of previous group norms.
Intercorrelations were computed betweeen the variables
and multiple correlations were calculated between the
independent variables and GPA.

Findings.—-A correlation of .29 (significant at

 

.01 level) was obtained between ACE and Hr scores. A
correlation of -.29 (significant at .01 level) was ob-
tained between the Hr and TMA scores. The correlations
of ACE and Hr scores with GPA were: ACE, .62, Hr, .32
(both significant at .01 level).

Comparisons of the single order and multiple
correlations indicated that a combination of the ACE
and Hr scales is a better predictor of GPA than ACE

alone (significant at .05 level). Adding the Taylor

1H. A. Klugh, and A. W. Bendig, "The Manifest
Anxiety and ACE Scales and College Achievement,
J;Consult. Psych., 19 (1955), p. 487.

 

 

 

 

 

 

 

 

 

 

 

Manifest Anxiety Scale to the ACE, Hr combination did

significantly (.05 level) increase the multiple cor-
relation with GPA.

Critical Analysis.--The author did not include
an adequate description of the method of sampling
selection used nor an adequate description of the
sample (i.e., whether freshman, sophomore, junior,
or senior status). The addition of the two person—
ality measures (TMA and Hr Scales) to an aptitude
predictor was reported as significantly increasing
the prediction of GPA but the size of the resultant
multiple correlation was not indicated. The crite-

rion, GPA, was not adequately described.

Mitchell?

Mitchell investigated the correlates of dif—
ferent measures of achievement motivation and con—
ducted an analysis of the factorial dimensions of
achievement motivation.2

Sample.-—One hundred thirty-one female stu-
dents in the teacher training curriculum at the Uni-
versity of Texas. All subjects were members of the

investigator's sections in elementary educational

psychology.

 

2J. v. Mitchell, "An Analysis of the Factorial
Dimensions of the Achievement Motivation Construct,"
J;_Educ. Psych., 52 (1961), pp. 179-187.

 

 

lO

Design.--The grade point average for the semester
in which the study was conducted and the scores on the
American Council on Education Psychological Examination
(ACE) were used to compute a discrepancy score which
was used as the criterion variable for achievement
motivation.

Several measures of achievement related attitudes,
including the McClelland Test of Achievement Motivation
(MTAM), were administered to all subjects. Intercor—
relations were computed between the measures of achieve-
ment motivation, ACE, discrepancy score, and GPA.

Findings.--Scores on the MTAM correlated .18

 

(significant at .05 level) with the discrepancy score.
A correlation of -.06 (non-significant) was obtained
between MTAM scores and GPA. A negative correlation

of -.23 (significant at .01 level) was obtained between
the MTAM and ACE.

Critical Analysis.--The rationale for the choice

 

of the criterion (discrepancy score) was not explicitly
stated. The method of sampling selection used and the

description of the sample was not adequate.

Weiss, Wertheimer,gand Groesbeck
Weiss, Wertheimer, and Groesbeck investigated

the relationship between n Achievement scores of the

 

 

 

 

 

11

Edwards Personal Preference Schedule (EPPS) and
McClelland Picture Story (MPS) scores and academic
achievement.3

Sample.--Sixty male students at the University
of Colorado. All subjects were in introductory psy—
chology classes. _

Design.—-The two personality instruments (EPPS
and MPS) were administered to all subjects. The cri-
terion was the three semester cumulative grade point
average for the freshman year. Intercorrelations were
computed between the two personality tests and GPA.

Findings.—-The correlation between n Achievement
scores of the EPPS and GPA was .42 (significant at .05
level). The correlation between MPS scores and GPA
was .34 (significant at .05 level).

Critical Analysis.—-Evidence of rater reliability
on the scoring of the McClelland Test of Achievement
Motivation protocols was not offered. On the basis of
other evidence, the omission of a reliability estimate

of scorer consistency considerably limits the conclu—

 

3P. Weiss, M. Wertheimer, B. Groesbeck, "Achieve-
ment, Academic Aptitude and College Grades, Educ, and
Psych. Mst. , 19 (1959), pp. 663— 666.

 

 

 

 

 

12

sions of the study.“ Evidence of use of the consistency

scale of the EPPS was not given.5

Freshmen Students

 

Published studies of research centered on the
relationship of non-intellective variables to college
achievement with a general college population are at
a minimum. Frick's study (1955) represents the one

published study in the area in the period 1952-1962.

arses

Frick investigated whether the addition of per—
sonality variables in a battery would improve prediction
of grade point average of a relatively homogeneous popu-

lation over that afforded by an aptitude test alone.6

 

Y. _T_

Farquhar and Krumboltz obtained a retest relia-
bility of .26 after a 9 week interval for the McClelland
TeSt of Achievement Motivation. "The low reliability
would cast doubt on the stability as well as the possible
validity of the instrument,” see J. Krumboltz and W. W.
Farquhar,"Reliability and Validity of the n-Achievement
Test," J. Consult. Psych., 21 (1957), pp. 226-228.

 

51n the EPPS test manual, Edwards advises against
using profiles which have Consistency scores lower than
9. See A. L. Edwards, Edwards Personal Preference
Schedule Manual (New York: Psychological Corporation,

1959). p- 161

6J. W. Frick, "Improving the Prediction of Academic
Achievement by Use of the MMPI," J. Appl. Psych., 39

(1955). pp. 49-52.

 

 

 

 

 

 

 

 

 

13

Sample.—-Two hundred sixty-seven women at the
University of California, who had completed two semes-
ters of the freshman year in 1951 and 1952. In order
to be included in the sample, each subject's validating
scales on the Minnesota Multiphasis Inventory (MMPI)
had to be within acceptable limits for research purposes.

Design.——The two instruments (American Council
on Education Psychological Examination (ACE) and MMPI)
were administered to all subjects at the time of matri-
culation. Pearsonian coefficients of correlation were
computed between the variables and first year GPA. A
multiple regression model was used to determine the
increase irlprediction<af GPA attained by addition of
MMPI scales to a measure of aptitude (ACE).

Findings.--A11 scales of the MMPI were negatively

 

correlated with ACE scores. The Paranoid (Pa) scale
of the MMPI correlated .13 (significant at .05 level)
with GPA. ACE scores correlated .48 (significant at
.01 level) with GPA.

The coefficient of multiple correlation derived
by combining the beta weights for all 8 scales (Hs, D,
Hy, Pd, Pa, Pt, Sc, Ma) of the MMPI and the ACE was
.64. The coefficient of determination when GPA is
predicted from ACE scores alone was .23; the coeffi-

cient of multiple determination was equal to .41.

 

 

14

Critical Analysis.--The author used the MMPI

 

scales on the basis that good personality adjustment,
as indicated by the MMPI, would be reflected in grade
point average. No research hypothesis is generated to
indicate the expected findings. The criteria for in-
dicating that the sample is a relatively homogeneous
group are not explicitly stated, nor was the propor- E

tion of subjects eliminated from the sample because of

low validating scores on the MMPI given. A test of
significance for the beta weights of the ACE and MMPI

was not included.

Discrepant Achievers

 

Gebhart and Hoyt (1958), Krug (1959), Merrill
and Murphy (1959), and Morgan (1952) investigated
personality differences between some combination of

students defined as high, average, or low achievers.

Gebhart and Hoyt

 

Gebhart and Hoyt investigated some personality
correlates of over- and under—achievement.7

Sample.-—Two hundred and forty freshmen selected
on the basis of predicted grade point average from the

population of male freshmen enrolled in the Schools of

 

7G. G. Gebhart, and D. P. Hoyt, "Personality
Needs of Under- and Over-Achieving Freshmen," J. Appl.

Psych., A2 (1958), pp. 125-128.

 

 

 

 

 

15

Engineering and Architecture or Arts and Letters at
Kansas State College in the fall of 1956-1957.

Each of the two school groups was subdivided
into three ability levels on the basis of predicted
GPA. The Pre—Engineering Ability Test was used in
making the predictions for Ss in the College of
Engineering and Architecture; the American Council
on Education Psychological Examination (ACE) was used
in making the predictions for Ss in the College of
Arts and Letters.

A further subdivision of subjects was made into
over- and under—achievers on the basis of achieved
versus predicted grade point averages for the first
semester. These procedures resulted in 12 groups
(2 schools, 3 ability levels, 2 levels of achievement).

Design.--A factorial design was employed and
an analysis of variance technique used on each of the
16 variables of the Edwards Personal Preference
Schedule (EPPS). The personality instrument had been
administered to all subjects at the time of matricu-
lation. First semester grades were the criterion of
achievement.

Findings.--Over-achievers scored significantly
higher than under—achievers on the Achievement (.001
level), Order, Intraception, and Consistency scales

(all at .05 level) of the EPPS. Under—achievers

 

l6

scored significantly higher than over—achievers on
the Nurturance (.001 level), Affiliation (.05 level),
and Change (.01 level) scales of the EPPS.

Critical Analysis.——The author did not explicitly

 

indicate the empirical or theoretical rationale for
comparing groups of discrepant achievers from two
different schools within the college. A technical
criticism derives from the use of different bases of
prediction (i.e., Pre-Engineering Ability Test for Ss
in the School of Engineering and Architecture and
American Council on Education Psychological Examination
for Ss in the School of Arts and Letters) for the two
groups at the first semester level without including
statistics on the first semester attrition rate due
to changes of major. The inclusion of Ss in the Low
Ability range with Consistency scores less than 9
(mean of the group on the Consistency scale was 10.3
with a standard deviation of 2.1) would appear ques—

tionable in light of Edwards' manual recommendations.

Krug
Krug investigated the relationship between

Edwards‘Personal Preference Schedule scores and over-

 

8 J
_ In the EPPS test manual, Edwards advises against
dsing profiles which have Consistency scores lower than
9. See Edwards, op. cit., p. 16

 

 

 

 

1?

and under-achievement in engineering students at
Carnegie Tech.9

Sample.——Two samples, each consisting of 120
Ss, were drawn from the population of 411 freshmen
who entered the College of Engineering and Science
in September, 1956. Two predictions of grade point
average were made for each subject: (1) the per—
formance—based prediction was made on the basis of
achievement scores of the College Entrance Board
Examinations and high school standings; (2) the ap—
titude-based prediction was made on the basis of
Verbal and Math scores from the College Board Scholas-
tic Aptitude Test.

A student was assigned to the over-achievement
group if his first year grade point average was above
the predicted GPA; to the underachievement group if
the average was below that predicted. In each of the
two samples there were 20 subjects at each of the 3
levels (high, normal, low) of expected performance
for both over- and under—achievement groups.

Design.—-A factorial design was used (2 pre-
diction bases, 3 predicted achievement levels). An
analysis of variance technique was used on each of

the 15 scales of the EPPS.

 

9R. E. Krug, "Over— and Underachievement and
and the Edwards Personal Preference Schedule," g;

Appl. Psych., 43 (1959), pp. 133-136.

 

 

 

 

 

 

l8

Findings.--In the aptitude-based sample, over—
achievers scored significantly higher than under—
achievers on the Achievement (.001 level), Order
(.05 level), and Endurance (.01 level) scales of the
EPPS. Under-achievers scored significantly higher
than over-achievers on Affiliation (.01 level) and
Heterosexuality (.05 level) scales of the EPPS.

In the performance—based sample, the null
hypothesis of no difference between groups of over—
and under—achievers on the scores of the EPPS was
accepted for all scales except that of Achievement.
The over—achievers scored significantly higher than
under-achievers on the Achievement (.05) scale of the
EPPS.

Critical Analysis.——The author did not report

 

using a test of significance for the discrepancy be-
tween achieved and predicted GPA. In the discussion
of results, significant interaction (between groups
and levels) was reported for Deference, Succorance,
and Endurance scales of the EPPS in the aptitude-
based group but the level of significance was not
indicated. Evidence of use of the consistency scale

of the EPPS was not given.10

 

loEdwards, op. cit.

 

 

 

 

l9

Merrill and Murphy

Merrill and Murphy investigated personality
factors as measured by the Edwards Personal Preference
Schedule (EPPS) which may discriminate between a group
of college students with low predicted academic achieve—
ment who seem to be overachieving and a group of similar
students who seem to be performing as predicted and thus
are failing.11

Sample.--The sample was selected from the popula—
tion of 300 freshmen admitted to the University of Utah
in the fall quarter, 1955—1956, whose predicted grade
point average was 1.50 or below. A multiple correlation
formula (Jex, 1949, University of Utah) which uses high
school grades and achievement test scores was used in
prediction of grade point average.

The over-achieving sample (N, 49) consisted of
those students who had attained a grade point average
of 2.00 or above after one quarter of college work.
The achieving—as-expected sample (N, 52) consisted
of those students who had attained a grade point
average of 1.00 or below after one quarter of college

work.

 

11
R. M. Merrill, and D. T. Murphy, "Personality

Factors and Academic Achievement in College," J;
Couns. Psych., 6 (1959), pp. 207—210.

 

 

‘mms.

- -._-.I . fi

 

 

20

Design.-—The Edwards Personal Preference
Schedule (EPPS) had been administered to all subjects
as part of a special course required for the popula—
tion from which the samples were drawn. The mean
scores of the 11133 on the 15 scales of the EPPS
were analyzed using the "t" test of significance
of differences between means.

Findings.--The over—achieving sample scored
significantly higher than the achieving—as expected
sample on Dominance (.01 level), Deference (.05
level), and Endurance(.05 level) scales of the EPPS.

The achieving-as—expected sample scored signi-
ficantly higher than the over—achieving sample on
Autonomy (.01 level), Affiliation (.05 level), and
Change (.05 level) scales of the EPPS.

Critical Analysis.-—The author did not generate
a research hypothesis which would indicate on which
scales the two groups might differ even though it would
appear that adequatetheoretical base is available for
directional decisions. The selection of the population
from which the two samples were drawn in 1955-1956 on
the basis of a prediction formula derived in 1949 would
seem indefensible. Evidence of use of the consistency

scales of the EPPS was not given.12

 

l2Edwards, op. cit.
1

 

 

 

 

21

Morgan investigated the relationship between
certain measured personality variables and the academic
achievement of achieving and nonachieving college
students of high ability.l3

Sample.-—One hundred and thirty—two male students
from the College of Science, Literature, and the Arts
at the University of Minnesota who had scored at 136
or above on the American Council of Education Psy-
chological Examination (ACE). 0n the basis of first
year grade point average, the sample was subdivided
into three groups (high, middle, low). In order to
obtain two clearly defined groups which could be
characterized as achievers and non-achievers, only
the achievers (high range, N= 40) and nonachievers
(low range, N: 30) were considered in the study.

Design.--Comparisons of the scores of achievers
and nonachievers were made on the Minnesota Multipha-
sis Personality Inventory (MMPI) and the Thematic
Apperception Test (TAT). The mean scores of the 70
subjects were analyzed using "t" tests of significance.

Findings.——Achievers scored significantly higher

 

than nonachievers on three of the special scales of the

MMPI: (l) Dominance, (2) Social Responsibility,

 

13H. M. Morgan, "A Psychometric Comparison of

Achieving and Nonachieving College Students of High
Ability," J. Consult. Psych., 16 (1952), pp. 292-298.

 

 

22

(3) Intellectual Efficiency (all significant at .01
level). Achievers scored significantly higher than
nonachievers on n, Achievement (.02 level) scores
derived from TAT stories.

Critical Analysis.-—A test of significance of
the difference between means of achieved GPA for the
two groups was not included. A reliability estimate
of the scoring of the TAT protocols by the investi-
gator was obtained by two independent judges rating

a random sample of protocols.

Discussion of Selected Studies

Evidence that a relationship does exist between
measured non-intellective characteristics and academic
achievement of college students has been demonstrated
in the review of research. The findings of the studies
must be viewed within the framework of a number of
limitations related to the design and instrumentation
used.

The limitations in design include (a) use of
non-random or poorly defined sampling methods, (b)
lack of cross—validation groups, (0) lack of opera-
tional definitions of the non-intellective variables
under investigation, (d) lack of reliability and

validity estimates of the non-intellective instrumen-

tation.

 

 

 

 

23

The major limitation of the instrumentation derives
from the use of instruments with sound theoretical base
but limited validity evidence, or evidence of predictive
validity but little theoretical base. The choice of
instruments is essentially between theory and a "shot-gun"
approach. What is needed is a validated instrument which
incorporates both theory-base and predictive validity.

In the following design chapter, a measure is described
which appears to have the requisite elements defined

above.

Summary

A review of the research on the relationship of
non-intellective variables and academic achievement has
been presented. The majority of published studied have
been conducted with two types of college population:

(a) special curriculum groups, and (b) discrepant achievers.
This factor limits generalization of the findings to a
general college population.

Gebhart and Hoyt, Krug, Merrill and Murphy in their
study of low aptitude students found that dominance
deference and endurance as measured by the Edwards
Personal Preference Schedule, were the characteristics
which differentiated between the achieving and the

vailing low aptitude student.

 

 

 

 

 

 

 

 

24

Klugh and Bendig, in their study of a Special cur-
riculum group, demonstrated a positive relationship
between the Hr scale of the California Personality
Inventory and academic achievement. Mitchell found a
negative relationship between n Achievement, as measured
by the McClelland Test of Achievement Motivation, and
academic achievement. Weiss, Wertheimer, and Groesbeck,
using the Edwards Personal Preference Schedule, demon—
strated a positive relationship between n Achievement and
academic motivation.

No studies assessing the relationship between
academic motivation and academic achievement in the
general college pOpulation have been attempted to date.
It was concluded that this factor derives from the non-

availability of apprOpriate instrumentation.

 

 

 

 

 

 

 

CHAPTER III

DESIGN AND METHODOLOGY

The design of the study is centered on the relation-
ship among the following variables: achievement scores,

motivational scores, and aptitude scores.

Population

 

The pOpulation consisted of 5075 first quarter
freshmen at Michigan State University in the Fall term,
1962.

Definition of the Sample

 

The sample consisted of all entering freshmen who
attended the first or second Orientation Clincs for
Freshmen during the weeks of June 21st or June 27th, 1962.
Those freshmen attending one of the two sessions were
used due to their availability for testing purposes under
the auspices of the Office of Evaluation Services of
Michigan State University. For the purposes of the study,
the two samples were dichotomized into sexes. The following
two criteria were set for inclusion in the study: (1)
completion of the motivational scales; (2) completion of

the first quarter at Michigan State University.

25

 

 

26

Sample Size Attained

 

Four female and three male subjects failed to meet
the first criterion; three female and five male subjects
failed to meet the second criterion resulting in a loss
of seven females and eight males. The number of subjects
in the final samples consisted of 330 males and 367 females.
The College Qualification Test mean for the male sample
was 142.47 (s. d. 21.67); the female mean was 129.12 (s.

d. 25.21). A comparison of the College Qualification
means with a measure of central tendency (the median) for
the freshman population indicated that the samples were

representative of the general freshman population.
Instrumentation

A score for each individual was obtained on the
following variables: (1) aptitude test score as measured
by the College Qualification Test; (2) motivational
scores as measured by the three sub—test scores and the
total score of the Michigan State M—Scales; (3) achieve—
ment score as measured by first quarter composite grade

point average.

College Qualification Test

 

For the purposes of the study, the Total Score of
the CQT was used as the aptitude measure. The CQT is
administered as part of a battery of aptitude tests during
the Orientation Clinics. Published by the Psychological

 

 

 

 

 

 

27

Corporation, the CQT is a multi-purpose battery designed

to serve colleges in their admission, placement, and
guidance procedures. The CQT consists of the following
subtests:

l. Verbal: A fifteen minute test of vocabulary, containing
50 synonyms and 25 antonyms questions.

2. Numerical: A thirty—five minute test containing 50
questions drawing on arithmetic, algebra, and geometry.

3. Information: A thirty minute test composed of 75
questions from the fields of science (physics, chemistry,
biology) and social studies (history, government, economics,
geography).

Scores on the three subtests are summed to yield the CQT

Total score.

Michigan State M—Scales

 

For the purposes of the study, the Michigan State
M-Scales were used as the objective motivational instrument.
The M-Scales were developed by Farquhar and group as part
of a comprehensive research project on the motivational
factors related to academic achievement in eleventh grade
students.1 The M-Scales were designed to discriminate
between high and low motivated students who were assumed

to be discrepant in academic achievement.

 

lFarquhar, Motivation Factors Related to Academic
Achievement, op. cit. (1959)

 

 

28

The original form of the M-Scales consisted of four
subtests. A revision of the scales for use with college
freshmen consisted essentially of combining the selected
items from one subtest (the Preferred Job Characteristics.
(PJCS) Scale) with another subtest (the Generalized
Situational Choice Inventory (GSCI). The PJCS consisted
of all possible combinations of eight items designed to
measure high motivation with eight items designed to measure
low motivation. Subsequent administration of these itesm
indicated a low tolerance on the part of the adolescents
for the repetition. The directions for administering the
scale were identical to the GSCI and the theory base was
also the same as that of the GSCI. Therefore, the most
discriminating items of the PJCS were added to the GSCI.

The revised form of the M-Scales, Form C, consists
of a total of 124 items for the males (GSCI=50 items,
WRL=48 items, HTI=26 items), and a total of 109 items

for the females (GSCI=36 items, WRL=48 items, HTI—25 items).

Generalized Situational Choice Inventory (GSCI)
In the develOpment of the GSCI, the authors adapted
and extended McClelland's hypotheses that N—achievement

is composed of: (a) long-term involvement, (b) unique

accomplishment, (0) competition with a standard of excellence.

The constructs were polarized by postulating a continuum

of achievement motivation with the low motivation of

[IF—Hf”.

 

29

N-achievement being opposite in composition from that
advocated by McClelland and associates. Furthermore,
the constructs were viewed as being related specifically
to the academic setting.2

A two hundred item forced—choice inventory describing
situations logically related to the extremes of the polar
dimension theory was developed. One alternate of each
dichotomized forced choice pair of items was related to the
high academic motivation pole and the other to the low
academic motivation pole. Students were instructed to
choose the alternate which they would most prefer. Responses
were scored either 0 or 1 with the high score in the
direction of high motivation.

Forty-five male and thirty female items remained
after cross-validation. Reliability estimates were obtain—
ed by Hoyt's analysis of variance technique. Reliability
estimates for the males ranged from .80 to .84 and for
females from .77 to .90. The correlation of the GSCI with
the grade point criterion was .50 for males and .32 for

females (significant at .01 level).

Word Rating List (WRL)

 

Following from the assumption that a student‘s self-

concept is a functionally limiting and facilitating factor

 

2
Farquhar, Motivation Factors Related, to Academic
Achievement, op. cit.(1963),

 

 

30

in academic achievement, a ll9-item instrument was develOped.

The items were developed by extracting descriptive words

and phrases from the self-concept literature and from a

review of personality, motivational, and intellectual

characteristics of students representing academic extremes.
Each student was asked to rate each of these concepts

or words on a four-point scale (never, sometimes, usually,

and always) as they thought their teachers would rate the

 

words in describing him as a student. Forty-eight male
and forty—eight female items, with thirty-five items in
common, remained after cross validation.

Reliability estimates were obtained using Hoyt's
analysis of variance technique. Reliability estimates
ranged from .91 to .93 for males and from .88 to .93 for
females. The correlation of the WRL with grade point
criterion was .51 for males and .42 for females (signifi—

cant .01 level).

Human Trait Inventory

 

Items which previous research (Altus, Gough, McQuary,
and Truax) had found to differentiate between discrepant
achievers were used to construct a l25-item personality
inventory. Students were asked to rate each item on a four-
point scale (never, sometimes, usually, always) as it
applied to themselves. A high score indicated reSponding

similar to over-achievers.

 

31

Twenty-six items remained after cross-validation
for the males and twenty-five items for the females. The
twenty-three most significantly discriminating items were
selected for Hoyt's analysis of variance reliability
estimates. For males the reliability coefficients ranged
from .68 to .80 and from .71 to .76 for females. The
correlation of the HTI with grade point criterion was .42

for males and .36 for females (significant at .01 level).

M-Scales Total

The four measures (GSCI, WRL, HTI, and PJCS) were
combined into one instrument and labeled the Michigan
State M—Scales. The Total scale contained 139 male and
136 female items. On a random sample of 240 students
from the original 4200, reliability estimates based on
Hoyt's analysis of variance technique were estimated to
be .94 for males and .93 for females. Correlation of the
total scale with the grade point criterion was .49 to .56
for males and .30 to .43 for females for validation and

cross-validation samples respectively.

Grade Point Average (GPA)

The criterion of college achievement was the composite
grade point average earned on the basis of grades for all
courses taken in the first term. Grade points at Michigan
State University are calculated on the following scale:

A=4 points, B=3 points, C=2 points, D=l point, F=O points.

 

 

 

 

 

32

The average grade point is computed by dividing total points

earned by the total credits carried for the term.

Data Collection Procedures

 

Test Administration

 

The two instruments (CQT and Michigan State M-Scales)
were administered to all subjects during their attendance
at the Orientation Clinics for Freshmen. Although the
M-Scales are not timed tests, the subjects were requested
at the beginning of the testing session to complete the
scales within a 75 minute period due to time limitations
imposed by the Office of Evaluation Services.

Administration of the tests was done in large groups
by proctors who were supplied with a copy of the test
booklet. Subjects were instructed to read the instructions
attached to the test booklet for each scale and to raise
their hand if they should have questions. The responses

were entered on IBM answer sheets for machine scoring.

Processing of the Data

 

The scores for each of the two instruments were
obtained from the Testing Services. The composite grade
point average was obtained from the Office of the Registrar
at the end of the fall quarter. The scores for each of the
six variables were punched on IBM cards for each subject

as the preliminary step to analyzing the data.

 

33

Analysis Procedures

 

Three major types of analyses were conducted on the
data: (1) estimates of Michigan State M—Scales reliability
(sub- and total scales); (2) correlations of the motivational
(M-Scales) and aptitude (CQT-Total) measures with grade

point average; (3) multiple correlational analyses.

Reliability Analysis

 

Internal consistency reliability estimates were
obtained for the three subtests (GSCI, WRL, HTI) and the
total test of the M—Scales by Hoyt's analysis of variance
method. A random sample of 50 males and 50 females was

used in computing seperate reliability estimates.

Correlational Analysis

The Pearson product-moment coefficient was used to
estimate the correlations of the motivational and aptitude
measures with the grade point criterion. Intercorrelations
were calculated between the sub- and total scales of the
M-Scales, CQT-Total, and GPA using the Kl6-M program on
Michigan State University's MISTIC digital computor.

A multiple correlation was calculated between GPA and
(l) GSCI and CQT, (2) WRL and CQT, (3) HTI and CQT, and
(4) M-Scales Total and CQT. Multiple correlation enables
an evaluation of the significance of adding the M-Scales
to a multi—variate relationship. A test of the significance
of the beta weights from zero in the regression equation

was performed.

   

34

It was planned to use an "F" test for the statistical
significance of additional variables in the multiple
regression equation if the beta weight tests were found

to be significantly different from zero.

Assumptions of Correlation Model

 

In deciding on the correlational model to use in a
study, the important question is which type of correlation
will yield the most accurate picture of the relationship
DuBois states:

When all variates are based on the kinds of scales
apprOpriate for computing correlations and when all
regressions are linear, there should be no difference

in conclusions whether correlational analysis or
analysis of covariance is used. Under these conditions,
the two approaches are merely different sides of

the same coin. However, when the number of

external variates is more than one or two, the use

of the matrix format appears to afford greater ease

in computation than the procedures described as
analysis of covariance.3

For the purpose of this study, the assumption is made

that the data is normal in distribution and has linearity.

Null Hypotheses

 

In the present design the relationships of the M—Scales
to GPA and CQT—Total are formulated into the following:
HOI There is no relationship between an objective
measure of motivation (the M-Scales) and academic

achievement (GPA).

 

3Phillip Du Bois, Multivariate Correlational Analysis
(New York: Harper and Brothers, 1957) p. 159-160.

 

 

 

 

35

Sub hypothesis A—-There is no relationship'between
the sub- and total scales of the
M—Scales and the CQT-Total.

Sub Hypothesis Be-There is no relationship among
the three sub-scales of the M—Scales.

H II There is no increase in the recision of prediction

of academic achievement (GPA attained by the addition
of an objective measure of motivation (M-Scales) to an
academic aptitude measure (CQT-Total).

The level of significance was set at the .05 level

0 .'-'lr'.
,.

for the correlational analysis and at the .01 level

luv--

 

for tests of significance of the bets weights.

Summary

Samples were drawn from the pOpulation of 5075
first quarter freshmen at Michigan State University in
the Fall Term, 1962. For the purpose of the study, the
samples were dichotomized into sexes. A design based on
correlation was used to test the empirical relationship
between academic motivation and academic achievement.
Product-moment and multiple correlational analyses were
conducted on the motivational, aptitude, and achievement
variables. The design included calculating and testing
the beta weights of the regression equations for the

significance of adding variates to multiple prediction.

 

 

 

 

 

CHAPTER IV

ANALYSIS OF DATA

Three major types of analysis were conducted on the
data: (1) estimates of test reliability of the M-Scales; .
(2) estimates of test validity of the M—Scales; (3)

 

estimates of the increase in precision of prediction of a
GPA attained by the addition of the M-Scales to an
vaptitude predictor (CQT—Total).

Reliability Analysis

 

Internal consistency reliability estimates were
obtained by Hoyt's analysis of variance technique for the
sub- and total scales of the M—Scales (see Table l).
Seperate reliability estimates were computed using a random
sample (n=50) for the males and the females. The
reliability coefficients for the male§_were .79 for the
GSCI, .91 for WRL, and .59 for the HTI. The total-scales
reliability coefficient for the males was .60. Reliability
estimates for the females were .55 for the GSCI, .86 for
the WRL, and .54 for the HTI. The total scales reliability
estimate was .55 for the females.

With the exception of the WRL for both sexes and the
GSCI for males, the reliability coefficients were not satis-

factory.

36

 

37

 

 

TABLE 1

SUMMARY OF HOYT'S ANALYSIS OF VARIANCE

RELIABILITY ESTIMATES FOR THE M—SCALES
Males Females
M-Scales (N=50) (N=50)
Generalized Situational .79 .55

Choice Inventory

Word Rating List .91 .86
Human Trait Inventory .59 .54
M—Scales Total .60 .55

 

 

 

 

 

 

 

 

38

Correlational Analysis

 

The purpose of the correlational analysis was to
determine the relation of the M—Scales to (l) the grade
point average criterion, (2) an academic aptitude measure
(CQT-Total), and to (3) determine the interrelationships of
the three sub- scales of the M-Scales.

Correlation analysis was conducted testing the
following general null hypothesis:

Null Hypothesis—-There is no relationship between an
objective measure of academic motivation
(the M-Scales) and academic achievement
(grade point average).

Two subsidiary hypotheses were constructed to assess
(l) the relationship of the M—Scales to the CQT-Total and
(2) the interrelationships of the three sub-scales of the
M-Scales.

Sub Hypothesis A--There is no relationship between the sub-
and total scales of the M—Scales and
the CQT.

Sub Hypothesis B——There is no relationship among the three
sub-scales of the M—Scales.

All analyses were done seperately for each sex.

M-Scales Correlation Analysis

In Table 2 the sub- and total scales correlations

with the grade point average criterion are presented for the

 

 

 

 

 

 

 

 

   

39

male and the female samples. The sub- scales intercorrela-
tions and the sub- and total correlations with the CQT

are presented in the lower right corner of Table 2.

M-Scales Correlation with Grade Point Average

 

An inspection of Table 2.1 indicates that the
GSCI, the WRL, and the M—Total for malg§_are related
significantly with grade point average at the .05 level or
greater. The criterion correlations for males ranged
from .08 (HTI) to .19 (GSCI). The GSCI, WRL, HTI, and
M—Total for females are not significantly related with
grade point average. The criterion correlations ranged
from —.01 for the M-Total to .04 for the WRL.

The level of significance was set at .05 for the
validation of the sub- and total scales of the M—Scales
for the general pOpulation. Correlations of three of the
scales for the males (1. e., GSCI, WRL, and M—Total) were
low-order, positive, and all significant at the .05 level
or greater. Correlations of the sub- and total scales
for the females were not significant.

Except for the three tests mentioned above for males
(i. e., GSCI, WRL, and M-Total), the null hypothesis of no
relationship between the M-Scales and grade point criterion

was accepted for the general college pOpulation.

 

 

 

 

 

 

 

40

TABLE 2

INTERCORRELATIONS AMONG ACHIEVEMENT,
APTITUDE, AND M-SCALES

 

 

Measures GPA CQT-T GSCI WRL . HTI ‘M-T

 

Grade Point Average .59** .02 .04 .00 -.01

College Qualification
Test Total .48** .10 .01 -.03 .00

Generalized Situational

I
Choice Inventory .19** .24** l .24** .37** .60
Word Rating List .14* .14* : .37** .50** .77
Human Trait Inventory .08 -.03 I .34** .52** .76
M—Scales Total .17** .l8** : .70 .88 .68
l

 

*Significant at .05 level.

**Significant at .01 level.

Note: Dotted lines separate sub-scale intercorrelations in
lower right corner. Correlations above diagonal are
for females (n=367) and below the diagonal for males

n=330 .

 

 

 

 

 

 

41

Correlations of CQT—Total with M-Scales and
Grade Point'Average

 

 

Inspection of Table 2 indicates that for males the
GSCI (r=.24) and M—Total (r=.l8) were significantly related
with the CQT at the .01 level. The sub-scale correlations
with CQT for males ranged from -.03 (HTI) to .24 (GSCI).
The correlation of the CQT—Total with grade point criterion
was .48 (significant at the .01 level).

Again referring to Table 2 the GSCI, WRL, HTI, and
M-Total scores for females were not significantly related
to the CQT. The correlations ranged from -.O3 (HTI) to
.10 (GSCI). The correlation of the CQT-Total scores with
grade point criterion was .59 (significant at the .01 level).

The level of significance for assessing the relation—
ship between the sub— and total M—Scales with the CQT was
set at .05. Correlations of three of the scales for males
(GSCI, WRL, and M—Total) with the CQT were significant at
the .05 level or greater. Correlations of the sub- and
total scales for females with the CQT were not significant.

The null hypothesis of no relationship between the
M-Scales and academic aptitude was accepted for the general
college pOpulation with the exception of the three male

tests (1. e. GSCI, WRL, and M—Total).

M-Scale Intercorrelations

 

The sub-scale intercorrelations are also summarized

in Table 2. Significant correlations were found at the

 

 

42

.01 level between all sub—scales for both the male and
female samples. Correlation coefficients for the male
sample ranged from .34 to .52. For the female sample
the coefficients ranged from .24 to .50.

The level of significance for assessing the
interrelation of the sub-scales was set at the .05 level.
The null hypothesis of no relationship between the sub—

scales of the M—Scales was rejected for the general

college population.

Multiple Correlation

 

The purpose of the multiple correlation analysis
was to determine the improvement in precision of predic—
tion achieved by the addition of an objective measure of
academic motivation (the M-Scales) to an academic aptitude
measure (CQT-Total).
Multiple correlation analysis was conducted testing
the following null hypothesis:
Null Hypothesis--There is no improvement in the precision
of prediction of grade point average
attained by the addition of the M—Scales

to the CQT (Total).

Sub- and Total M-Scales Analysis

 

Each of the three sub-scales and the total M-Scales
were added seperately to the CQT Total in a regression

estimation of the grade point criterion. The beta weights

 

 

43

0f the resultant regression equations and the multiple
correlation coefficients are presented in Table 3 for

males and females. Regression analyses were conducted
independently for each sex.

The heaviest weighting of the sub- scales for 22122
was the GSCI (.1345), the weightings for the WRL (.0699)
and the HTI (.0894) were approximately equal. The
weighting for the M—Total for males was .8730. As would

 

be anticipated, for all four equations, the heaviest
weighting was for the CQT-Total.

The heaviest weighting of the sub- scales for
females was the WRL (.0340). Weightings for the GSCI
and the M-Total were -.0393 and —.7470 respectively.
As expected, the heaviest weighting for all four

equations was for the CQT.

Test of Significance of the Beta Weights

 

In testing any particular beta weight, three alterna—
tive findings can be hypothesized. (l) The beta weight
does not contribute to estimating the criterion ea
condition which is reflected in non-significant t-test
from zero. (2) The beta weight does contribute to the
criterion estimation (t-beta significant) but not to an
increase in precision (F test insignificant). (3) The
beta contributes (t—beta significant) by increasing the
precision (F-test significant).

 

 

 

 

 

44

TABLE 3

TWO VARIABLE ESTIMATES OF GRADE POINT AVERAGE
ACHIEVEMENT CRITERION WITH CORRESPONDING BETA
WEIGHTS AND MULTIPLE CORRELATIONS

 

Males (N=33o)

Females (N-357)

 

 

 

 

 

 

 

Multiple Multiple

Beta Weights R Beta Weights R
GSCIa .l345éGSCI) + -.O393EGSCI) +

4.6176 CQT) ** .49 5.9603 CQT) ** .59
WRL .O699(WRL) + .O340EWRL; +

4.720(CQT) ** .49 5.9156 CQT ** .59
HTI .0894(HTI) + .0188gHTI; +

4.840(CQT) ** .49 5.9259 CQT ** .59
MT .8730(MT) + —.7470(MT)) +

4.660(CQT ** .49 5.9223 CQT) ** .59

aGSCI Generalized Situational Choice Inventory;

**t—test

WRL
HTI

MT
CQT

II H II II II

Word Rating List;

Human Trait Inventory;
M-Scales Total;
College Qualification Test (Total).

for Beta Weights significant at .01 level.

 

45

The t-tests of significance of the beta weights of
the three sub— and total M-Scales were all negative for the
male and female samples (see Table 3). With these negative
findings, the F test for significance of adding the M-Scales
to the CQT in prediction of the grade point criterion was
not carried out.

The t-testsof significance of the beta weights of
the CQT were positive at the .01 level for both the male
and female samples. The CQT significantly contributes to
the estimation of grade point average.

The null hypothesis of no increase in the precision
of prediction of GPA by addition of the M-Scales to the

CQT was accepted for the general college population.

Further Exploration

In an attempt to trace further some of the possible
sources of the consistently negative findings of this
study, a random sample of 50 students for each sex was
drawn from the total group to whom the M—Scales were
administered. The frequency distribution of scores of these
students on the sub— and total M—Scales together with the
mean, standard deviation, the median, and the possible
score are presented in.Figures 1-8 for males and
females.

All distributions for both sexes on the alb-and total
scales are slightly negatively skewed. The distribution

of scores on the HTI (male and female) was the most

 

46

Aomu osoom oHnHmmomv
Aomu zv nose: goo moaoom gem Home ao coaosoaacmam--. H .wam

 

 

whoom
5 3 7.6 3.4321 87.6
14 3 3 33 222
E o
ll‘m
L2
00. H: u shape: I: ma
wm. m u soapmfi>om pampcmpm
ma. mm H ewe:
L

 

sqoefqns

 

 

 

 

 

 

 

47

Aw:n whoow oHnHmmomv
AoT zv noHo: sou noaoom 3mm Hm: mo coHosoHaano--. m .me

 

 

 

 

 

 

 

 

 

oaoom
87654321098765.4321987. 5.43210
44.44444144333é3333333222 6.2222220
l-m
nlfioa
[-ma
00 H: n coHpoz
Ho. w u coHooH>om cementum
wH. mm H coo:

 

sqoefqns

 

 

 
 

 

flow u choom oHnHmmomv
Homnzv ooHoz woo nohoom zom Hem do coHospHAonHm--.m .mHm

 

 

 

 

 

whoow
65432109876
22222221111
0
m ..m m.
q
E
P .
0
Q.
S
.1- OH
oo.mm u coHooz
:m.m n COHpMH>oQ pamocmpm .I \
mm.mm u coo: fl UH

 

 

 

 

 

 

 

49

Aama u ohoom oHnHmmomv
Aom u zv moam: sch monoom Hmuoauz Mo COHuannpmHmuu.: .mHm

 

 

oaoom
4
immnmwwmwwmwmmm%%w%%%%%mwwwm%%w%&mm
lllllllllllllll. . .O
5% . _ :1 : Pr
1: m
LFOH
1-3
oo.NOHu GmHooz
om.m n coHpoH>om egosdoam
mm.mm n Cam:

 

saoefqns

 

 

 

 

50

Amm u opoom oHnmeomv
Homnzv nonEom pom nonoom Home do coHcsoHaemHm--.m .mHm

.u.
3

whoom

1098765432
22222

 

omuam
OH.mN

II II II

32
3333222

 

 

cmHooz
soauwfi>om ohmcdmpm
Com:

0

S
n
0..
Pr
8
O
1
S

- OH

- ma

 

 

 
 

 

51

Am: n whoom manammomv
Aomuzv mwamsmm pom mmnoom Hmz mo coausnahpmaans.m .me

FD u. 1,9..; 0
.4 n».4.4 u.n»

 

F EE PE,

 

 

 

 

oo.mm n Cmaomz
OH.m H COHDGH>®Q UMNUCMUW
NH . NM H Emmi

 

sqoefqns

Amm u whoom mHnHmmomv
Aomuzv mmamEmm pom mmpoom Hem mo soapspfippmfim--.p .wfim

 

 

 

 

 

 

 

 

 

whoom
“4321098765“.
22222111111 0
Ill.m
S
n
q
p
e
O
2 I 3
5 rl S
:ll.0H
om.Hm n cmficmz ma
mm.m u coapmfi>mm upmccmpm
mm.om u cam:

 

 

Amoa n mmoom mHnHmmomv
Aomnzv mmHmem mom mmnoom Hmuoenz mo soapsnfinpmfimuu.w .wam

whoom
0,9unlfo :zh» ?.2 11 O 9,9u 7.Au .4 32 1; O o/Qu 7L0 874
m9oiv999qu9988n08 8 87;:7777

 

_E

 

E EEHE M,

 

 

 

 

53

00.0w H QMHUQS
mH.OH u Coapmfi>mm thUCmpm
00.0w n cam: 11!

 

 

L(\
H sqoafqns

O

 

 

 

 

 

54

negatively skewed as well as characterized by the greatest
restriction of score range. The HTI is also the shortest
of the fliree sub—scales.

The distributions did not depart sufficiently from
normality to account for the low correlations between
the M-Scales and the grade point criterion. Therefore,
the correlations must be interpreted as an accurate
reflection of the utility of the M-Scales for this college

sample.

Summary

The reliability of the sub- and total tests of
the M-Scales was estimated using Hoyt's analysis of variance
technique on a random sample of males and females (N=50).
The reliability coefficients for the males were .79 (GSCI),
.91 (WRL), .59 (HTI), and .60 (total). The reliability
coefficients for the females were .55 (GSCI), .86 (WRL),
.54 (HTI), and .55 (total).

The validity of the sub— and total scales of the
M—Scales was estimated by correlation and regression
analyses independently for the male (N=330) and the female
(N=367) samples. Two of the male sub-tests (GSCI and WRL)
and the total M-Scales correlated significantly with the
grade point criterion. All of the correlations of the
male sub-scales were low—order and positive. A11 correla—
tion coefficients were in the zero range for the female

sample.

 

 

 

(it w .-“iii i-|&|..ouls

..
11 {it [likfl I i win

 

 

. h ‘
IF , in. D .F’tll {Jr}. _ r b» .|.I * v-’ .

. .. V _ . . I
I r {I‘lrivlillrif 1' WOLF I‘uFllfikil . . \ 1 I3. cf”?! {5' .Viw ‘

 

55

On the basis of the above findings, it has been
demonstrated that there is no significant relationship
between an objective measure of academic motivation
(M-Scales) and academic achievement (GPA) for a general
pOpulation of college freshmen.

As anticipated from the above findings, the regres-
sion analysis and the resultant multiple correlations and
beta weights indicated that there is no improvement in
precision of prediction of grade point average attained by
the addition of the M—Scales to a measure of academic

aptitude (CQT—Total).

 

 

 

 

 

 

CHAPTER V

SUMMARY, CONCLUSIONS, AND IMPLICATIONS

The major problem of this study was to determine
(1) whether an objective measure of academic motivation
(the M-Scales) had a significant positive relationship to
the academic achievement of first-quarter freshmen and (2)
whether there was a significant increase in precision of
prediction of academic achievement (GPA) attained by the
addition of the M-Scales to an academic aptitude measure
(CQT-Total).

For the purposes of the study, the sample was selected
from the general population of first-quarter freshmen at
Michigan State University and dichotomized into sexes.

A comparison of the sample means of the CQT with a measure
of central tendency (median) for the male and female
freshmen population indicated that the samples were
similar in academic aptitude to the general college pop-
ulation.

An academic aptitude measure (CQT-Total), academic
motivational measure (M—Scales), and academic achievement
(GPA) were obtained for each student. The analyses were
carried out independently for the male and female sample

56

 

 

 

 

 

57

and involved the following: (1) reliability estimates
were calculated using Hoyt's analysis of variance
technique; (2) Pearsonian correlations were computed to
assess the relationship of the M—Scales to academic achieve-
ment; (3) multiple correlations were calculated to assess
the increase in precision of prediction of academic achieve-
ment attained by the addition of the M-Scales to the CQT.
With the exception of the Hoyt's analysis of variance, all
statistical analyses were performed on the total sample
(males N=330, females, N=367) using the K16M program with
Michigan State University‘s MISTIC digital computor.
Separate reliability estimates were computed using a
random sample (N=50) for the males and the females. The
reliability coefficients for the males ranged from .59
(HTI) to .91 (WRI). For the females, the range of reliabil—
ity coefficients was from .54 (HTI) to .86 (WRL). With
the exception of the WRL (r=.91) and the GSCI (r=.79)
for males and the WRL (r=.86) for females, the reliability
coefficients fell below a satisfactory level.
The frequency distributions ofscores for both sexes
on the sub--and total M—Scales were characterized by a
slight negative skewness. The HTI, the least reliable of
the sub—scales for both sexes, displayed the sharpest
restriction in range of scores as well as the highest

degree of negative skewness.

 

 

 

 

 

 

 

58

A Significant, low-positive relationship was found be-
tween the malg GSCI, WEI” M-Total scores and the grade point
criterion. The correlation coefficients for males ranged
from .08 (HTI) to .19 (GSCI). For the females, no signifi-
cant relationship between M-Scales scores and the grade point
criterion was demonstrated; all of the correlations were in
the zero range.

Intercorrelations of the sub-scales were significant
for both sexes ranging from .34 to .52 for the male sample and
from .24 to .50 for the female sample. Correlations of the
malg_GSCI (r=.24), WRL (r=.lA), and M-Total (r=.18)scores were
significantly related to the CQT scores. For the females,
the correlations between the M-Scales sub- and total scores
with the CQT scores were in the zero range.

The "t" tests ofsignificance from zero for the M-Scales
beta weights in predicting GPA were not significant. Addition
of the M-Scales scores to the CQT measure in a multiple regres-
sion equation did not result in an increase in precision of
prediction of the grade point criterion.

The null hypothesis of no relationship between the M-
Scales scores and grade point criterion was accepted for the
general college population with the exception of three male
scales (i.e., GSCI, WRL, M—Total).

The null hypothesis of no relationship between the M-
Scales and the CQT scores was accepted for the general college
population with the exception of three male scales (i.e.,

GSCI, WRL, M—Total).

 

 

 

 

 

 

59

The null hypothesis of no interrelationship among the
sub— M-Scales scores was rejected for the general college
population. All correlations were significantly different
from zero at the .01 level or higher.

The null hypothesis of no improvement in prediction of
grade point average achieved by the addition of the M-Scales

to the CQT was accepted for the general college population.

Conclusions

On the basis of these findings, it is concluded that
for the obtained sample:
1. There is a significant, low—positive relationship between
the GSCI, WRL, M-Total scores and academic achievement (GPA)
for a male college population.
2. There is no significant relationship between the HTI
scores and academic achievement (GPA) for a male college
population.
3. There is no significant relationship between the sub-
and total M-Scales scores and academic achievement (GPA)
for a female college population.
A. There is a significant relationship between the GSCI,
WRL, M-Total scores and academic aptitude (CQT) for a male
college population.
5. There is no significant relationship between the HTI
scores and academic aptitude (CQT) for a male college

population.

 

6O

6. There is no significant relationship between the
sub- and total M-Scales scores and academic aptitude
(CQT) for a female college population.

7. The correlation between the GPA and CQT scores for
males (.49) and for females (.59) is not significantly
increased by adding either the sub— or total M-Scales

scores to the estimates.

Discussion of Results

 

Because the study was centered upon the validation
of the M-Scales, the results of the study will be dis-
cussed from two aspects: (1) the reliability of the M—
Scales, and (2) the relationship of the M-Scales to GPA.

With the exception of the WRL and the GSCI for
males and the WRL for females, the reliability estimates
were not satisfactory for either the male or the female
samples. The distribution of scores on the WRL on which the
possible scores for both sexes are similar (n248) was
comparable for both samples. An interesting finding was
that the degree of negative skewness was higher for the
females than for the males on the WRL.

The WRL purports to measure the student's self
concept as a functionally limiting and facilitating factor
in academic achievement. Despite the higher male academic
aptitude (CQT mean=142.47, standard deviation~21.67), the
females with a lower academic aptitude (CQT mean-129.12,

 

 

 

 

 

61

Standard deviation 25.21) achieve at the same level as the
males. The female mean GPA was 2.26 with a standard
deviation of .73 compared to the male mean GPA of 2.17
with a standard deviation of .76. In light of the skewness
evidence cited above, it would appear that the overall
tendency of the self concept is in the direction of the
higher achieving females grade point average.

The HTI was the least reliable of the sub—scales
for both samples. The shortest of the sub-scales (n=26
for males, n=25 for females), the frequency distribution
of scores on the HTI was the most negatively skewed for
both sexes. A possible explanation for the high degree of
skewness, as well as the restriction in the range of scores,
may be the test-taking attitude of the college students
comprising the samples. The test content of the HTI
(items which attempt to assess the student's fantasy life,
level of anxiety, and compulsivity) may have activated the
defense mechanisms of these s0phisticated subjects in the
form of denial.

A consistent finding on all of the scales for both
sexes which warrants discussion is the restricted range
of scores. The frequency distribution of scores for a
random sample (N=50) of males and females on the M-Scales
(sub- and total may be found on pages 46—53cn?Chapter

IV).

 

 

 

 

 

 

62

A possible explanation for such restriction may be
the response "sets" of a college population. In this
context, "set" refers to the tendency of an individual to
respond in such a manner as to give a "good" impression
of himself.1 The restriction of range of scores in the
direction of "high" would tend to suggest that a response
set was operating to depress the reliability estimates for
the HTI and, conversely, the expanded range of scores found
for the WRL tended to inflate the reliability estimate for
both sexes.

A significant, low-positive relationship was demon-
strated between the male GSCI scores and the male WRL scores
and the grade point criterion. Although of low-order
(GSCI, r=.l9, WRL, r=.14), this finding might tend to
substantiate the applicability of the theoretical basis
of the two scales for a male college population.

The consistently non-significant, zero order relation-
ship between all female scores on the M—Scales and the
criterion implies that the dimensions of female academic
motivation at the college level are not congruent with

the male dimensions.

 

lAllen L. Edwards, "The Relationship Between the
Judged Desirability of a Trait and the Probability That
It Will be Endorsed," Journal of Applied Psychology, 37

(1953). pp. 90-93.

 

 

 

 

 

 

63

On the basis of the findings of this study, the
utility of the M-Scales for decision making at the college
level is extremely limited. In the Farquhar study (1963)2
with eleventh grade students, the M-Scales did correctly
classify those students who did or did not identify with
the educational environment. However, with a college
population a different emphasis in the theoretical founda-
tion might be indicated with a resultant difference in
item content. Particularly would this latter point be
an important consideration for the females where it
appears that the procedures employed with the Farquhar
high school sample did not generalize to the college

level.

Implications for Further Research

 

1. Construct a new objective measure of academic
motivation for a college population using a
methodology similar to that of the M-Scales
construction but a different theoretical emphasis.

2. Conduct a longitudinal study to determine if the
relationship between the M-Scales and GPA remains
constant for the male and the female samples of this

study.

 

2Farquhar, Motivation Factors Related to Academic
Achievement (1963), op. cit.

 

 

 

 

..\..o)

.. v.
{1-}

E

 

. 4
try.

Milli-Vb...

 

 

64

Particularly conduct a study with the female
sample of this study to determine if the relation-
ship between the WRL and the GPA remains constant
at the end of one year of college work.

Conduct a study with the male sample to determine
if there is a relationship between scores on the 5‘
GSCI and the WRL and the selected educational—

vocational objective.

 

Investigate the self concept dimensions of the

W
male and female college student as these dimensions ‘

are related to academic achievement.

 

 

 

 

 

 

 

BIBLIOGRAPHY

Borow, H. "Current Problems in the Prediction of College
Performance,” J. Amer. Assoc, Coll. Registr., 22.

(1946), 14-26.

 

Edwards, Allen L, Manual for the Edwards Personal Pref-
erence Schedule. New York: Psychological Corpor-
ation, 1959.

 

 

. "The Relationship Between the Judged Desirability
of a Trait and the Probability That It Will Be
Endorsed," Journal of Applied Psychology, 37 (1953),

90-93.

Farquhar, William W. A Comprehensive Study of the
Motivational Factors UnderlyingAchievement of
Eleventh Grade High SChool Students. ResearCh
Project No. 846(8458). Supported by the U. S.
Office of Education, in c00peration with Michigan
State University, 1959.

 

 

 

 

Motivation Factors Related to Academic Achieve-
ment: Final Report of COOperative Research Project
No. 846, College of Education. East Lansing, Michigan:
Michigan State University, 1963.

 

Fishman, Josuha and Pasanella Ann K. "College Admission-
Selection Studies," Review of Educ. Research, 30
(1960), 298—306.

 

Frick, J. W. HImproving the Prediction of Academic
Achievement by Use of the MMPI," Journal of Applied

Psycholoex. 39 (1955). 49-52.

Gebhart, C. Gary and Hoyt, Donald P. "Personality Needs
of Under- and Over—Achieving Freshmen," Journal of
Applied Psychology, 42 (1958), 125—128.

 

 

 

Klugh, Henry E. and Bendig, A. W. "The Manifest Anxiety
and ACE Scales and College Achievement," Journal of
of Consulting Psychology, 19 (1955), 487.

 

 

Krug, Robert E. "Over and Underachievement and the Edwards
Personal Preference Schedule," Journal of Applied

Psycholosy, A3 (1959). 133-136.
65

 

 

 

 

 

 

 

 

66

lCrumboltz, John D. and Farquhar, William W. "Reliability
and Validity of the n Achievement Test," Journal of
Consulting Psychology, 21 (1957), 226—228.

Merrill, Reed M. and Murphy, Daniel T. "Personality
Factors and Academic Achievement in College," Journal
of Counseling Psychology, 6 (1959), 207-210.

Mitchell, James V. Jr. "An Analysis of the Factorial
Dimensions of the Achievement Motivation Construct,"

Journal Of Educational Psychology, 52 (1961), 179—
187.

Morgan, Henry H. "A Psychometric Comparison of Achieving
and Nonachieving College Students of High Ability,
Journal of Consulting Psychology, 16 (1952), 292—298.

Weiss, Peter, Wertheimer, Michael, and Groesbeck, Byron.

”Achievement, Academic Aptitude, and College Grades "

Educational and Psychological Measurement, 19 (1959),
663-666.

 

APPENDIX

SAMPLE ITEMS FROM THE
MICHIGAN STATE M-SCALES

57

 

68

GENERALIZED SITUATIONAL CHOICE INVENTORY

I would prefer to:

11 a) Develop a new product which may or may not be good,

or
b) Make a product as good as the best one available.

12 a Receive money for my good grades, or
Be allowed to take any course I wanted because

of good grades.

Be successful in finishing a job, or
Finish a job.

13

14 a Get excellent grades because I have a great deal
of ability or

)

b) Get average grades because I have average ability.
)
)

 

15 a Be graded at the end of a course with the
possibility of making an "A", or
b Get a "C" at the beginning of a course along
with everyone else.
U)
WORD RATING LIST 2 a,
H H a)
a 9 r1 %
w m m m
TEACHERS FEEL THAT I AM: 2’, E 3 E.
z a: D <
84. studious l 2 3 4
85. different 1 2 3 4
86. discontented l 2 3 4
87. flighty l 2 3 4
88. responsible l 2 3 4
89. original 1 2 3 4
90. consistent I 2 3 4
91. intellignet 1 2 3 4
92. distractible l 2 3 4
93. in—the-know l 2 3 4
8
E >:
HUMAN TRAIT INVENTORY L. :j g a
o m m m
> E :3 3
(D O U) H
134. I work things out by myself rather than 2 W =3 <
have a friend show me how. 1 2 3 4
135. I have been quite independent and
free from family rule. 1 2 3 4

 

 

 

 

137.

69

When I have an opinion, I stand up
for it.

It is difficult for me to keep
interested in most of my school
subjects.

I have difficulty working under
strict rules and regulations.

F‘Never

m Sometimes

U)Usually

z-Always

 

 

 

 

 

 

 

 

 

   

 

 

Rosa TE 0

‘ _.-_
74—1

‘gt ‘3‘!“

., r“
9 fl;:~\‘ '5

"Emma

N

r,“

 

m
ii‘h

 

 

 

 

 

 
    
  

   

 

STATE U

0

1293

 

 

 

 

 

 

 

..."W WWI)IVIHLWIBIITIIAIITIIES
3 3 0 8 5 0 5 19