THE MMES PERSONALITY INVENTGRY
Thesis for the Degree of M. A
MICfHGAN STATE UNIVERSITY
IAMES J. MOREARTY, JR.
1 9 6. 9

 

ABSTRACT
THE JAMES PERSONALITY INVENTORI
A MULTIPLE CHOICE TEST OF THE NEED FOR.ACHIEVEMENT
By

James J. Moriarty, Jr.

The James Personality Inventory (JPI) was an attempt
to develop a multiple choice measure of n Ach. The
advantage of a multiple choice test is that it is easier
to administer and score than the McClelland TAT measure
of n Ach.

An 8? item True/False test (JPI #1) was created from
the n Ach literature. It was administered with the TAT to
30 Ss in Research Session #1. The correlation between
the TAT and JPI #1 scores was .18 (NS). The internal
consistency of the test was high. .77.

Four subscales, JPI #2, #Za, #3. and #h, were
empirically generated from this data. These scales were
tested on the data of Research Session #1. JPI #2 scores
correlated .73 with the TAT scores. The internal consistency
of the scale was .70. JPI #2a correlated .6u with the
TAT and showed an internal consistency of .65; JPI #3
and JPI #u did not correlate significantly with the TAT
criterion but the internal consistency of the scales was
high, .67 and .88, respectively.

Research Session #2 was an attempt to cross-validate

the above scales. 9n Ss were administered the TAT and the

87 item JPI. The correlation between the TAT and JPI #1
scores was .19 (NS). The internal consistency of JPI #1
remained high, .76. The correlation between the TAT and
JP1 #2 scores was significant and somewhat higher, .25
(p< .02). However, the internal consistency of this
scale dropped to .32. JPI #Za correlated with the TAT
.22 (P<.03) and had a higher internal consistency, .bz9.
JPI #3 correlated with the TAT .21 (p< .01») with an
internal consistency of .ho. The correlation between
the TAT and JPI ## was .15 (NS). The internal consistency
remained high, .76.

The results are consistent with other attempts
(Nehrabian, 1968) to create a multiple choice measure of
n Ach. The development of an alternate measure of n Ach
may not be the fault of the multiple choice testing, but
rather the impracticality of using the TAT as the validating
criterion. The unreliability of the McClelland TAT renders
it an unstable criterion and consequently attenuates the
size of validation correlations. A research strategy that
employs a clear behavioral criterion of n Ach would prove
more fruitful.

Evidence from Mehrabian‘s and the author's work, as
well as internal consistency measures on the TAT, suggest
that n Ach is a multidimensional factor. Future attempts

to construct a multiple choice test of n Ach should be

developed on the basis of a multidimensional personality

model of the high achiever.

 

 

THE JAMES PERSONALITY INVENTORY
A MULTIPLE CHOICE TEST OF THE NEED FOR ACHIEVEMENT
By

James J. Moriarty, Jr.

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF ARTS
Department of Psychology
1969

In memory of Aunt Mamie,

always a model of achievement for me

11

ACKNOWLEDGEMENTS

I wish to gratefully acknowledge the professional
and moral support that I have received from Professor
Uleman who has worked with me on every phase of this
research. His open posture informed me that the success
or failure of the research was my own, but his insight and
experience were always present. An appreciation for his
approach has made me a better student and teacher.

My appreciation of Professor Aronoff is as much for
his shared views of psychology as it is for his criticisms.
and suggestions to my research. His depth view of behavior
is one that has forced me to reconsider a number of issues
in social science. Time will tell where such a perspective
will lead me.

Professor Hanley's protean understanding of issues,
academic and otherwise, has always provided a critical
conscience for me. I sometimes suspect that he understands
my motivations better than I. His advice on statistical
methodology has been invaluable to the creation of this
thesis.

Jerry Gilmore's counsel with computer programming has
greatly facilitated the progress of my work as well as
given me an appreciation of the age in which I live. Bob
Gurney's advice with statistics has proved invaluable and
enriching. And finally, Dianne Wilhelm's clerical help
and friendship have made the tedious work of the thesis
more enjoyable than it would have been alone.

111

TABLE OF CONTENTS
Page

LIST OF TABLESO'OO0.0.0.000.000000000000000...0‘000007-11

LIST OF APPENDICES ...............................V111

Chapter
I " INTRODUCTIONOOOOOOIOOOOOOOOOOOOOOOOOOOOOOOO‘OO 1

The Problems Developing a Multiple ChoiCe
Test or n ACh ooeooeooeoeo'oooooooeoooocooooo

strategy or the Research ..........‘.........
Significance of the Achievement Motive .....
Characteristics of the High Achiever .......

Definition Of 11 ACh 00300000000006.9000

1
2
3
L.
5
Portrait of the High Achiever ......... 5
Measuring Need for Achievement ............. 8
The n Ach TAT ......................... 8
Psychometric Disadvantages of the TAT . 9
Alternate Measures of n Ach.....;.......10
Rationale of a Multiple ChoiCe Measure
of n Ach .............s................ 12
II - DESIGN AND PROCEDURE OF THE RESEARCH ...,.., 13
Strategy of Item Creation and Selection .... 13
Creation of the Item Pool ............. 13

The Problem of Item Ambiguity .... IA

Reduction Of Item P001 0000000000000... 15

Research Session #1: Empirical Selection
and Keying Of Items cocooeeocecooeoocoooceoe 16

Procedure of Research Session #1 ...... 16

iv

Page

Subjects ...........................
Administration of TAT and JPI ......
Results of Research Session #1 ..........
TAT and JPI #1 .........e...........

Criteria for Item Selection on JPI #2,
#23, #3, 8‘ #4 cocoon-coco05000000000000...

Construction of JPI #2 ..........8.......
Construction of JPI #2a .................
Construction of JPI #3 ..................
Construction of JPI #h ..................
Rationale of Research Session #1 ........
Research Session #2: Cross-Validation Session
Procedure of Research Session #2 ........
Subjects .....£.....................
Administration of TAT and JPI ......
.Results of Research Session #2 ..........
TAT and JPI #1 .....................

TAT and JPI #2 .........:...........

TAT and JPI #2a ....................

TAT and JPI #3 .....................

TAT and JPI #u .....................

III - DISCUSSION 000......OOOOOOOOOOOOOOOOOOOOOOOO.

Unreliability of the TAT as a Criterion of

n AOh o-ecoooecooococococo00050000000000.0000..-

Relationship Between the TAT and JPI,......{..

11 A011 as a Multi-Dimens 101181 Factor 0 o o o e o c o o o

16
16
18
18

18
19
21
22

23

25
25
25
25
25
25
25
26
27
27

28

28
31
35

Page
BIBLIOGRAPHY 0.00000000000000000000000COOOQOOOOOO 37

APPENDIX A 0.0.0.000...COO...CCOOOOOOOOOOOOO'OOOOOQLFZ
”PMDIX B O...00.0.0.0...000......00.00.000.00000u9

vi

LIST OF TABLES

Table Page

1 JPI Correlations with the TAT and Summary Test
8133131513103 eooeooceooeoeoeeooeoeoco0000000000 20

2 Percentage Agreement in Scoring Achievement
Imagery and Rank Order Correlations between,
n Ach Scores Obtained by Novice Coders/Experts
and. Moriarty/Experts one...occ‘occooooooeeooo.op51

vii

 

 

LIST OF APPENDICES

Appendix .. Page
A COMPfiI‘iSOH Of Item Keys 000.00.000.00... “’2

B Reliability of TAT Scoring by Moriarty. 49

viii

THE JAMES PERSONALITY INVENTORY
by

James J. Moriarty, Jr.

I - INTRODUCTION

EMW'WEWWMQInM

This research was an attempt to develop a multiple
choice measure of the need for achievement (n Ach). The
James Personality Inventory (JPI) was constructed to be a
reliable and easy-to-score test that would identify the
same Ss that McClelland's (1953) projective measure of
n Ach identifies.

There are several psychometric and pragmatic dis-
advantages to McClelland's Thematic Apperception Test (TAT).
Briefly, the problems with the TAT are: (l) the administra-
tion and scoring of the TAT require considerable training:
(2) the scoring of the TAT is time consuming: (3) the
TAT has low test-retest reliability.

The nature of multiple choice testing obviates
problems of scorer training as well as facilitates the
speed of test scoring, particularly if computer time is
available. The problem of establishing satisfactory test-
retest reliability is more complex, but ease of administra-

tion and scoring of multiple choice tests facilitates

sound test construction.

Wflmw a

Essentially two types of experimental design may be
employed to select and validate items for a multiple
choice personality inventory. The first method is that of
motive arousal and subsequent item selection. Items are
administered to one group of Ss under neutral conditions,
and to another group under motive arousal conditions.
Items which differentiate groups are selected for further
validation studies. This method of item selection is
analogous to the original methodology that McClelland
(1953) employed to create the scoring system for the TAT
measure of n Ach.

The second method of item selection is that of
correlating items against a criterion and selecting
those items which correlate with the criterion. This
was the experimental design employed in this study
because a consistently reliable method of arousing n Ach
is not available (Klinger, 1966), and multiple choice
items have been traditionally viewed as resistant to
situational manipulation.

The basic strategy of the research was to create
multiple choice items with face validity from the n Ach
literature, and then establish their concurrent validity
against a n Ach criterion. Hence if an item correlated

with scores on the TAT it was selected for inclusion

in the JPI,

An attempt was also made to build a scale of
internally consistent items on the assumption that n
Ach is essentially a unidimensional personality character-
istic. That is, SS high in n Ach would respond to

multiple choice items in a consistent and similar manner.

W 8- Riflemh mm

The initial work of McClelland and his associates
at a Naval base in the 1940's has led to the uncovering
of the human motive termed "need for achievement".
Research on the achievement motive constitutes a
significant body of literature in the field of social
psychology. Major research projects have been carried
out at Wesleyan University (McClelland, Atkinson, Clark,
and Lowell, 1953), the University of Michigan (Atkinson,
1958), Harvard University (McClelland, 1961) as well
as various non—academic research centers (e.g., Krumbolz
and Farquhar, U.S. Air Force, 1957; Behavior Research
Service, 1960).

Researchers have linked the motive to such varied
behaviors as academic achievement (e.g., Rosen, 1956;
Robinson, 196h), verbal memory tasks (e.g., Karolchuck
and Worell, 1956; Sampson, 1963), parent-child relations
(e.g., Winterbottom, 1958), and cognitive styles
(e.g., French, 1955).

In a provocative clumination of his work on the

achievement motive, Thg_Agni§11ng,§gg1§11,(1961), McClelland

4

applied psychological reasoning and research procedures
to an analysis of economic development of civilizations.
Since n Ach was considered to be a major psychological
motivation of the entrepreneur, it was hypothesized that
societies that encouraged the development of such
personalities would prosper economically. In effect,
McClelland was examining under a psychological cover one
of the bases of western civilization, the Protestant
work ethnic (cf., Weber's hypothesis on the relation of
Protestantism to economic growth).

Despite such widespread research on the high achiever,
there is still no method of identifying this personality
type which is easy to administer and score. The present
TAT scoring system has psychometric and pragmatic
deficiencies. The development of an improved measure

of n Ach is the subject of this thesis.
Wmmm

We will draw a composite portrait of the high
achiever to give the reader a conception of the personality
type that the TAT and JPI are attempting to identify. More
detailed reviews of the research and theory of the motive
are contained in Brown's S931a1_£§xghglggy (1965, Pp. R23-M76)
and Byrne's An W (.129, W (1966, Pp. 284-338).
Heckhausen's The ARM 9:: the AW £23.11: (1967)
contains the most detailed accounting of research and

theory; it provides an indispensible introduction m)

researchers interested in this field of research.

Pailnitlan.9£.n.ésh

Need for achievement is defined as the desire to
do well or to compete with a standard of excellence.
The Achievement Goal is success in competition with
some standard of excellence: engagement in competitive
activity, other than pure agression, where winning or
doing as well or better than someone else is the primary
concern; and the subjective satisfaction in succeeding

(Atkinson, 1968).

Paxirait.2£.ihe High Achiezer

Empirical evidence has shown that people who are
high in n Ach work harder at laboratory tasks; learn
faster; do better work in high school with I-Q- partialled
out; and seem to do their best work when it counts for
their record and not when other special incentives are
introduced such as pressure from the outside to do well,
money prizes, or time off from work (McClelland at al., 1953).

High achievers have realistic aspiration levels,
whereas low achievers have relatively low or unrealisti-
cally high aspiration levels (Atkinson, 1964). They are
resistant to social pressure, choose experts over friends
as work partners, tend to be more active in college and
community affairs, like risky occupations, perform better
under longer odds, and choose moderate risks over either

safe or speculative ones (McClelland :1 al., 1953).

High achievers are better able to delay gratification
than low achievers (Mischel, 1961). They are less
dogmatic and less neurotic than low achievers. They

have a tendency'to desire social interaction with high
status targets (Nehrabian, 1969). They enjoy traveling
and in general are future-oriented. They conceive of
time in metaphors of a flying or galloping beast hection)
as opposed to a vast peaceful ocean (static) (Knapp and
Garbutt, 1958).

High achievers have been indulged less by their
parents during their childhood than low achievers
(McClelland, 1961). They come from families in which
there is stress on early self-reliance and mastery
(Winterbottom, 1958). And finally, they are drawn to
and do better at entrepreneurial occupations (McClelland, 1961).

The high achiever is likely to be active and
successful in business enterprises because business
involves risk, personal responsibility (possibilities of
credit for success), and measurable results. These are
characteristics that attract people with high n Ach.
Several studies support the hypothesis that men with high
n Ach are predisposed to enter business rather than the
professions (Behavior Research Service, 1960; Veroff
.2j,al., 1960). The n Ach scores of business managers,
representative of the entrepreneurial class, were

Significantly higher than the n Ach scores of professionals,

representative of men of equal abilities and background.
Such findings have led Brown to suggest that whereas
McClelland originally attempted to find a generalized
measure of achievement motivation, "his measure has
turned out to be chiefly a measure of motivation for
business or economic achievement."1

This lrge collection of research from various
psychological laboratories gives us a composite portrait
of a person with high n Ach as someone who is concerned
about accomplishment, or who wants to do well at what
he undertakes; he is energetic: and he is predisposed
toward innovations, toward working at tasks which are
not safe and traditional but that involve a moderate
degree of risk--perhaps only then he can feel subjective
satisfaction from succeeding.

The initial problem of the research was to create
a pool of items that would "tap", or identify SS with,
the above described personality characteristics. The

details of the creation of the item pool are contained

in the "Design and Procedure" chapter.

1
Roger Brown. £1.13 a W. New York: The
Free Press, 1965. p. 473.

8
Wm 191-211“ mi

ThenAEhTAT

Supported by Freud's original work on dreams,
years of psychoanalytic experience, and the clincal
success of Murray's Thematic Apperception Test, McClelland
decided that an excellent place to look for indicators
of the achievement motive was in fantasy (McClelland,
£3 21.. 1953). McClelland at al. developed a variation
of Murray's TAT based "on the notion from experimental
studies of animals that motives could be experimentally
aroused and their intensity controlled by manipulating
arousal conditions."1 A four picture series of cards
depicting characters involved in potentially achievement
oriented situations are administered to a S. For example,
one of the n Ach cards pictures a student sitting at his
desk with an open book in front of him. Whereas a
subject with low n Ach would describe the boy as "day
dreaming" about his girl friend or some non-achievement
oriented matter, a subject with high n Ach would describe
the student as dreaming about how great a physician he
will be one day, the great scientific discoveries he will
make, etc. The subject is not describing or perceiving
what is in the picture but rather projecting his own

personality, including his n Ach, into the picture.

*‘—_ ;

1D.C. McClelland, J.w. Atkinson, R.A. Clark, and

E.L. Lowell, The aghiezementjmgjlxg. New York: Appleton-
Century-Crofts, 1953. p. 3.

9

the subject's score on n Ach is the simple count of the
number of achievement related images in the story he
has written.

Smith and Feld (1958) reviewed the published
research that used the n Ach system of content analysis.
Their survey revealed that interjudge scoring reliabilities
of .66, .89, .95, .89, .91, .80, .96, and .82 and score-
rescore reliabilities of .95, .93. .95, .88, and .94
have been reported for n Ach. This degree of inter-
scorer reliability and score—rescore reliability is

satisfactory for research purposes.

Esxahametria Disadxantasas.a£.the.TAT.
McClelland's TAT has the psychometric disadvantage

of low rest-retest reliability. Product-moment co-
efficients of equivalence-stability for three to five
week intervals have ranged from .22 on college students
(Lowell, 1950) to .6A on high school students (Morgan, 1953).
Although Haber and Alpert (1958) have found a somewhat
satisfactory correlation of .5A, with two parallel sets
of pictures, the majority of reported reliability tests
of the TAT are in the range of .26 (Krumbolz and Farquhar,
1957) and .22 (Birney, 1959). Such low test-retest
reliability casts severe doubt on the stability as well
as the possible validity of the measure.

At least one researcher (Heckhausen, 1967) has

argued that whereas the test is not suitable for

10

psychodiagnostic purposes, its degree of discrimination
is sufficient for experimental research. Consequently
a large number of the n Ach studies employ only extreme

scorers in their subject pool.

Alternatei’iaammsnfnésh

Possible substitute tests of achievement such as
self-report measures (e.g., De Charms, Morrison, Reitman,
and McClelland, 1955: the measure of achievement in the
Edwards Personal Preference Schedule, Edwards, 1959),
or more structured versions of the TAT (e.g., Iowa
Picture Interpretation Test, Hurley, 1955; French, 1958)
do not correlate with the TAT measure (e.g., DeCharms,
Morrison, Reitman, and McClelland, 1955; Williams, 1955;
Marlowe, 1959; Atkinson and Litwin, 1960; or Barnette,
1961). Hence for this and other reasons they are
rarely employed in studies of achievement.

Heckhausen (1963) has developed a six picture TAT
measure of n Ach using "Hope of Success" and ”Fear of
Failure" as two dimensions of n Ach. McClelland's measure
correlates between .MO and .70 with ("Hope of Success"

+ "Fear of Failure"). Unfortunately, the Heckhausen
measure also suffers from problems of administration and
scorer training as well as time consuming protocol scoring.

Hehrabian (1968) undertook construction of a
multiple choice measure of n Ach simultaneous to the

initiation of this research project. He constructed

11

separate male and female scales of "The Tendency to

Achieve." Each scale consists of 34 items which are

designed to discriminate high versus low achievers. For

example, the difference in time perspective of high

versus low achievers was considered. High achievers

have a more distant future perspective than low achievers

(e.g., Hicks and Epley, 1960). Mehrabian attempted to

tap this perspective by the following item, "I think

more of the future than of the present and past." (True)

The reliability of the test is good. A ten week

test-retest yielded a product-moment correlation
coefficient of .78 for the male scale and a coefficient
of .72 for the female scale. The TAT correlated .28
(N=110, p(.01) with the male achievement scale and
.17 (N=111, p<.05) with the female scale. Further
investigations of the validity of hehrabian's scale
indicate that it is a useful measure of n Ach. Following
an actual course examination, male high achievers who
were identified by the Hehrabian scale showed a stronger
Zeigarnik effect (i.e., recall a greater percentage of
the failed items relative to the percentage of passed
items) than male low achievers (Weiner, Johnson, and
Mehrabian, 1968). The Hehrabian scale correlated .64
(N=110, p<.01) with Rotter's (1966) internal-external
control scale, indicating that high achievers perceive

themselves as having a greater degree of control over

12

the events which influence their lives (Nehrabian, 1968).
These findings are consistent with evidence which

differentiates high from low achievers.

Relianals Di a.fisl£inle.Qheiae.Eeasars.2£.n.éah

Briefly stated, the problem explored in this thesis
is as follows: despite the fact that there is widespread
research on the achievement motive, the TAT method of
identifying high achievers has some serious disadvantages.
The JPI was developed as an alternate test of n Ach
because a multiple choice test is easy to administer and
score. Since the TAT is the primary means of measuring
n Ach, it was necessary to employ it as the criterion
of n Ach. The primary objective was to establish the
JPI'S validity against the criterion of the TAT. Hence
scores on the JPI should correlate highly with scores

on the TAT.

II — DESIGN AND PROCEDURE OF THE RESEARCH

Spratfigy of Item Crgapjgn and Selection

The strategy of the project was to review n Ach
research and theory and create an item pool from which
the JPI could be developed. Items were created that
attempted to tap the cluster of interrelated character-
istics which distinguish high achievers from low
achievers (cf., Characteristics of High Achiever, p.4).
The initial keying of items was done on the basis of
face validity. Hence agreement or disagreement with an
item was indicative of a behavioral disposition which
has been found to differentiate high versus low achievers.
Final keying of items was determined by the empirical
performance of the item. If an item distinguished
between low and high scorers on the TAT it was retained

and keyed in the appropriate direction.

WC 8 Himaltemfiaal

300 items were created from an extensive review of
the n Ach literature. For example, "I prefer moderate
risks ( 3 out of 6) to either safe (5 out of 6) or
speculative (1 out of 6) risks." (Item #65). The item
was keyed "True" because n Ach theory suggests that SS
high in n Ach prefer moderate risks to extremely
speculative or safe risks (Atkinson, 1964). This item
was eliminated from the final scale because it did not
help to identify Ss who were consistently scoring high

or low on the TAT measure.

-A

14

' consistent

The following item was keyed H?alse,'
with n Ach theory, and retained throughout the study
because it distinguished between Ss who were scoring
high and low in the TAT criterion. "It is more important
to be socially acceptable than to be a success." (item #68)

The Emblem 5212 Item W: The problem of
item ambiguity raises three issues. (1) Item ambiguity
increases the probability that acquiescence sets will
influence item response. (2) Item ambiguity disguises
the variable that the test is measuring. (3) Item
ambiguity expands the opportunity for projection to
occur.

Ambiguity in items on the JPI was considered
undesirable. This approach was employed in order to
avoid acqmdescent response sets. "Scores on personality
tests will be influenced by acquiescent tendencies only
to the degree to which the test contains items which are
difficult or ambiguous and thus provide Opportunities
for acquiescent responses to occur."1 If an item is
somewhat unclear, so that the S is in doubt about the
appropriate response, or if he is indifferent about

taking the test and is not willing to "figure out the

meaning" of the item, response sets become important.

 

1Samuel Messick and John Ross, M

want in
W and Emman- New York! John Wiley. 1962. p. 106.

15

Hence the items were made clear in meaning, brief in
wording. This was designed to keep ambiguity to a
minimum and reduce the Operation of acquiescent response
sets.

It appears that the 33 were unaware of the person-
ality variable being measured. Only 4% of the Ss
guessed with any accuracy the purpose of the testing
sessions. Hence it was not necessary to employ
ambiguity in items to disguise the item intent.

Although item ambiguity would provide an increased
opportunity for projection to occur, this factor was
considered less desirable than avoiding the problem
of acquiescent response sets that item ambiguity produces.

For the above reasons, ambiguity in items was kept to a

minimum.

mammal

The 300 items were divided arbitrarily into two
pools of 150 items. The item pools were administered
to two groups of 75 males each. Items were eliminated
against three criteria. (1) If 95% or more of the S
answered ”True" or "False" to the item it was dropped.
The item did not discriminate between SS. (2) Only one
item was retained if several items seemed to be asking

the same question and had approximately the same index

of difficulty.1 (3) Items were eliminated that SS

1Index of difficulty is the proportion of the total
group who got the item wrong.

l6

consistently reported as ambiguous or too difficult to
understand. The 86 retained items constituted JPI #1
(cf., Appendix A for JPI #1). According to the fi,pzigzl
method of keying, 44 items were keyed "True"; 43 items

were keyed "False."
Easeereh.§eseien.11! Emeiriee1.§e1eeiien.end Iexine.ef Items

Preeednre e£.Reeeareh Seseien.£1

l§nhjegts. The SS were 30 male undergraduate
volunteers who were enrolled in Introductory Psychology
courses at Michigan State University. Course credit
was given to 83 who participated in experiments. Two
data gathering sessions, October 3 and 5, 1967, were
necessary to obtain the sample. Data was also gathered
on 44 females but was not analyzed because the primary
purpose of the research was to construct a male test of
n Ach.1 The mean age of the 33 was 18.8, S.D., 1.5. The
educational level was 13.7.

Adminietzeiien.ef.TAT and QEI. A screen projected
n Ach TAT, picture series 2, 8, 1, 7 (Atkinson, 1958),
was administered under neutral conditions (McClelland,

et_al., 1953). The story protocols were scored for n Ach.

;—— E
—— ‘—

1The subject call did not distinguish sex. It has
been the experience of this experimenter that it is easiest
to gather a sample when sex is not distinguished on
subject sign—up sheets.

17

by a trained rater1 whose percentage agreement in scoring
achievement imagery with expert scorers was in the mid
.90'8.

Following the administration of the TAT, the 87 item
JPI #1 was administered. SS selected one of four choices
to describe how they felt about an individual item.

TRUE: (1) This is very true of me

(describes me very well).
(2) This is often true of me
(describes me pretty well).
FALSE: (3) This does ngt,describe me very well.
(4) This does n9; describe me at all.

In the item analysis choices (1) and (2) were
treated as TRUE. Choices (3) and (4) were counted as
FALSE. The four choice selection was employed to make
it easier for the SS to answer the items. Repeated
data analysis showed that there was no significant
difference in item performance if the response was
scored in the four choice categories. Since the collapsed
categories proved easier to work with the two choice
categories were used in data analysis.

Three questions were printed at the end of the JPI.

Did you enjoy taking this test?

yes fiflé, neutral 33&. no Zé
What do you think was the purpose of the

"picture test?”
What do you think was the purpose of the JPI?

1The trained rater was the author. His scoring
proficiency is reported in Appendix B.

18

Only 3% reported with any accuracy the purpose of either
test.1

.eemliefi eifleeeemhéeesieefl

TAT,and‘££l,fil. The range of TAT scores on the 30
38 was -4 to 15. The mean was 3.1, S.D., 4.1. Cronbach's
(1951) coefficient alpha, an estimate of the internal
consistency of the TAT, was .23. All 87 items were used
in computing the JPI scores. The items were keyed
consistent with n Ach theory. Hereafter this §,p:1921
keying of items will be referred to as JPI #1. The
internal consistency of the scale, as measured by the
Kuder—Richardson Formula 20, was .77. The correlation
between the TAT and JPI #1 scores was .18(N=30, NS).
Table 1 on page 20 reports this and all subsequent
data.

mmmw O mgflfi.m.fl.é&

Four scales were developed from the data gathered
in Research Session #1. They were constructed on the

basis of four sets of criteria for item selection.

* 1*: __.__

llt appeared that the SS were unaware of the person-
ality dimension being measured and consequently were not
influenced by an artificially induced n Ach set. This may
be a problem with Hehrabian's test since it is named
The Resultant Achievement Motivation Scale (RAM). That
is, merely informing the S that he is taking a test that
is concerned with achievement may induce a mental set
that will affect his performance. The induced mental set
'was the basis for McClelland's original research strategy.

19

Items on JPI #2 were selected if an item showed a biserial
correlation of .20 or greater with the TAT scores.

Items were selected for JPI #2a, a subscale of JPI #2,

if the mean TAT score of the Ss who got the item right

was 1.5 higher than the mean TAT score of the Ss who
missed the item. Items on JPI #3 were selected from

their performance on a contingency table. If 75%

of the SS who scored above the median TAT score responded
to an item in a consistent manner, the item was selected.
Items on JPI #4 were selected on the basis of item

internal consistency.

Wail-”Ii:

The selection criterion for items on JPI #2 was a
biserial correlation1 of 1.20 or greater with the TAT
scores. That is, if an item correlated with the criterion
measure it was selected. For example, the item analysis
showed that item 5 had a biserial correlation of .42 with

the TAT scores. Hence it was selected.

 

—————

Two correlations are commonly used as an index to
decide whether to keep or discard an item. The first is
the biserial correlation. This is the correlation between
the 3's performance on an item (right or wrong) and his
test score. It assumes that the distribution of test
scores is normal and that there is some underlying normal
distribution of the right-wrong dichotomy. The point
biserial correlation is also a correlation between the
S's performance on an item (right or wrong) and test score.
It assumes that the test score distribution is normal and
that the division on item performance is a natural
dichotomy. The biserial correlation was used as the
criterion for item selection because the degree of agree-
ment or disagreement with an item was considered to be a
continuous normally distributed variable, i.e., Ss varied
in the degree an item described them.

20

Table 1 is a summary of the correlations and test

statistics of the two data gathering sessions.

Table 1.
JPI Correlations with the TAT and Summary Test Statistics

 

P—v

__ __ ___

Research Session #1 JPI #1 JPI #2 JPI #2a JPI #3 JPI #4

 

N = 30 87 43 27 37 59

items items items items items

Range 43-76 18-39 12-23 16-35 21-54

Mean & S.D. 57.6:8.2 26.e:u.9 18.233.u 22.2:u.5 4018.?

Internal Consistency

(Kuder-Rich. 20) .77 .70 .65 .67 . .88

TAT & JPI ** **

Correlations .18 .73 .64 .30 .12

Research Session #2

 

N = 94

Range 29-73 16-33 8-24 13-33 16-49
Mean & .S.D. 58:8 26.613.3 18.2i2.9 22.6:3.4 39.3:6.6
Internal Consistency
(Kuder-Bich. 20) .76 .32 .49 .40 .76
TAT & JPI * * ,
Correlations .19 .25 .22 .21 .15

* *‘I'

The critical validatin correlations are underlined.
Note that JPI #2, #2a, and 3 correlated significantly
with the TAT criterion. The internal consistency of

JPI #4 is high but the scale is not related to the TAT
measure 0

21

Item 74 showed a negative correlation of .81. It
was selected but its keying was reversed from the fi,pzlgzi
keying of False to True.

43 items were selected according to the biserial
correlation criterion. 26 items were keyed True and
17 items were keyed False (cf., Appendix A for items).
This scoring key was applied to the JPI data of Research
Session #1. The internal consistency of JPI #2 as
measured by the Kuder-Richardson Formula 20 was .70.

The correlation between the TAT and JPI #2 scores was
.73 (N=30, p< .01).

Since Scale #2 showed good internal consistency
and a very strong correlation with the Ss's TAT scores it
was decided to test its validity on another sample.

These data are reported in Research Session #2.

Wmfllfie

A 27 item subscale, JPI #2a (cf., Appendix A for
items), was derived from the 43 item JPI #2. An item
was selected if the mean TAT score of the SS who got
the item right was 1.5 higher than the mean TAT score
of the SS who missed the item.

27 items were selected according to this criterion.
16 items were keyed True: 11 items were keyed False.

This scoring key was applied to the JPI data of

22

Research Session #1. The internal consistency of JPI #2a
as measured by the Kuder-Richardson Formula 20 was .65.
The correlation between the TAT and JPI #2a scores
was .64 (N330, p<.01).

The good internal consistency and strong correlation
with the Ss's TAT scores indicated that JPI #2a should
be tested on another sample. These data are reported

in Research Session #2.

Wnfllfl

JPI #3 was constructed with items from a contingency
table. An item was selected if two criteria were met
in the item response pattern: (1)' 75% of the 83 who
scored above the median TAT score loaded in one direction
on a particular item; (2) 40% of the 33 who scored
below the median TAT score loaded in the opposite
direction on the item. The item was keyed in the direction
selected by Ss who scored high on the TAT.

For example, on item 78, 86% of the Ss who scored
above the median TAT score answered False. 43% of the Be
who scored below the median TAT score answered True.

The item was selected and keyed False.

37 items were selected according to this procedure
of item analysis. 22 items were keyed True and 15
items were keyed False (cf., Appendix A for items).

Scale #3 was applied to the JPI data of Research Session
#1. The internal consistency of JPI #3 was .67. The

23

correlation between the 3's TAT scores and their JPI #3
scores was .30 (N=30, NS).

Although JPI #3 did not correlate significantly
with the TAT criterion, the internal consistency of the
scale warranted testing it on another sample. These data

are reported in Research Session #2.

Wufllfl

The selection criterion for items in JPI #4 was
item internal consistency. If the biserial correlation
of an item was equal to or greater than .20 the item
was included. In this instance, the biserial correlation
is the correlation between the item's performance and
the total JPI #1 test score. A high biserial correlation
indicates that there is a tendency'for those who got the
item right to obtain high scores on the JPI,

If our assumption is correct that n Ach is a
unidimensional personality characteristic, and our
selection of items is tapping this dimension, we should
build an internally consistent test. For this reason,
items on JPI #4 were selected on the basis of internal
consistency.

59 items were selected according to the internal
consistency criterion. 32 items were keyed True and

27 items were keyed False (cf., Appendix A for items).

24

This scoring key was applied to the JPI data of Research
Session #1. The internal consistency of JPI #4 as
measured by the Kuder-Richardson Formula 20 was .88.

The correlation between the TAT and JPI #4 scores was
.12 (N=3o, NS).

Although JPI #4 did not correlate with the TAT
criterion, the strong internal consistency of the scale
made it worthwhile to attempt another validating
session with this set of items. These data are reported

in Research Session #2.

Ramdenala.erwfiaaaareh.§eesien.£1
The purpose of Research Session #1 was to select
and key'items empirically. It was also to build a scale
of items that was internally consistent on the assumption
that n Ach is a consistent behavioral predisposition.
The problem was to select a wide variety of items that
tap the personality syndrome described in the Introduction.
Research Session #1 produced four scales which were
internally consistent. JPI #2 and #2a correlated highly.
.73 and .64, with the criterion. Research Session #2
was an attempt to cross-validate them. Although JPI #3
and #4 didJnot correlate with the criterion, their
high internal consistency warranted testing them out

on another sample.

25
Beesemhmeieniz: Sues-WW

W Research Sessim #2

SthaQLa. The Ss were 94 males drawn from the
same population as the Ss in Research Session #1. Two
data gathering sessions, November 27 and 29, 1967, were
necessary to gather a large sample. Data was also
gathered on 123 females but was not analyzed. The mean
age of the 83 was 18.6, S.D., 1.8. The educational
level was 13.5.

Adminismnen 9.: TAT and £21. The test adminis-
tration procedure of Research Session #1 was followed.
57% of the Ss reported that they enjoyed taking the
tests; 30% were neutral; 13% did not enjoy the testing.
5% reported accurately that they knew what the test was

measuring.

.Rasnlts.e£.Raeeareh.§saeien.#2

TATjand iglflfil, The range of TAT scores on the
94 83 was -4 to 16. The mean was 5.35, S.D., 5.1.
Cronbach's coefficient alpha was .26. The internal
consistency of the scale was .76. The correlation of
the TAT and JPI #1 scores was .19 (N=94, NS). These
results are similar to the data of Research Session #1.

TA: and £21,#2, JPI #2, the 43 item scale
constructed with items that correlated with the TAT

criterion, yielded an internal consistency of only .32.

26

This is a drop in internal consistency from .70.
The TAT and JPI #2 scores correlated .25 (N=94, p(.02).
This was the critical correlation for validating the JPI,
There was no significant difference between this
correlation and Mehrabian's (1968) validating correlation
between the RAM and TAT.
253 and LP; #23. JPI #2a, the 27 item subscale of
JPI #2, yielded a slightly lower validating correlation
'with the TAT, .22 (N=94, p<(.03). The internal consistency
of the subscale was significantly higher. .49.1
The moderate size of these coefficients of internal
consistency may indicate that n Ach is not a unidimensional

personality characteristic. The significance of this

finding is considered in the Discussion.

 

1The Spearman and Brown formula (adapted from
Guilford, 1954, P. 391) estimates that the internal
consistency of this subscale would be .60 if the same
kind of items could be obtained to raise the length
of the scale to 43 items.

 

 

i3
rhn = .60 =
li-(nmflrt.c 1+133-2231.49)
27 27
Inn = estimated internal consistency of the lengthened
teSto

n = increase in test length

internal consistency of original test

27

TAT.and g2;,£3‘ JPI #3, the 37 item scale developed
from the contingency table, yielded an internal consistency
of .40. This is a drop from .67. The TAT and JPI #3
scores correlated .21 (N=94, p<.04). Once again the
size of the validating correlation was low. Although
the validating correlations and internal consistency
of JPI #2 and #3 are in the same range, the higher
validating correlation of JPI #2 indicates that it is
a better testing instrument.

TAT and g2; fig. The internal consistency of JPI #4,
the 59 item scale constructed on the basis of item
internal consistency, remained high, .76. The validating
correlation, however, did not reach significance,

.15 (N=94). The scale is obviously a measure of some
consistent response pattern, but it does not validate
against the TAT criterion of n Ach. Further research

is needed to determine what it is measuring.

28

III — DISCUSSION

In the discussion, three points are considered!
(1) the unreliability of the TAT as a criterion of n Ach;
(2) the strength of the relationship between the TAT

and JPI;

(3) n Ach as a multi-dimensional personality characteristic.

Wmmfliasammumninéah

The weak validating correlations of the present
research, along with the small TAT validating correlation
of Mehrabian's work, suggest to this author the im-
practicality of developing a multiple ohOice measure of
n Ach with the TAT as the sole criterion for n Ach. The
researcher cannot be sure that the low validity correlations
that he obtains are due to the lack of validity of his
measure or the low reliability of the TAT. As Guilford
points out, "We should obtain a very erroneous idea of
how well we are doing with a selection test or composite
score if the reliability of the criterion measure were
only .30, which can bery well happen, and if no correction
for attenuation were made. If the uncorrected validity
coefficient rxy were .20, the correlation corrected for

errors in the criterion would be .36, or almost doubled."1

lJ-P. Guilford. mm mm: <2nd Ed.)
New York: McGraw-Hill, 195“. P. #01.

29

This is precisely the case with the TAT. Its
internal consistency and test-retest reliability
coefficients are only in the range of .26 (e.g.. Cronbach's
alpha = .23, .26; test-retest coefficients = .22,
Birney. 1959s .26, Krumbolz and Farquhar, 1957). If we
apply the formula that Guilford (195h, P. #01) suggests
for oneway correction for attenuation to the JPI data,

we find that the validating correlation rises from

 

 

.25 to .h9,
r .2
xy 5
rm =3 __ , 014'9 =
ryy .26
Figure 1. Formula for One-Way Correction for
Attenuation.
rxw = correlation with correction in Y only (TAT)
rxy = obtained correlation between X & Y (TAT & JPI#2)
ryy = reliability coefficient of Y (TAT)

The correction for attenuation informs us of the
relationship between the two variables 1: the TM were to
have perfect reliability. Such corrected correlations are
of little practical value since they cannot be used in
prediction. However, correlation coefficients corrected
for attenuation are of theoretical importance in the
analysis of relationships, in that allowance can be made
for variable errors of measurement, i.e., errors of
measurement tend to reduce or attenuate the correlation

between variables.

30

The formula for complete correction for attenuation
(Guilford, 195”, P. #00) indicates that there may
actually be a strong relation between the two tests.

r 025
‘5' .88 =

____, O
jfrx;_r;;t' ,[:26 .32—'

Figure 2. Formula for Complete Correction for
Attenuation.

rvw =

 

 

rvw = correlation between true components of X & Y
(TAT & JPI #2)

r1y = obtained correlation between X & Y

rxxvryy a reliability coefficients of X & Y

Once again such corrected correlations are of little
practical value, but they give us an idea of what may be
the true relationship between variables if the variables
had high reliability. Low reliabilities limit the size of
any validating correlations.

We must reserve Judgement on the strength of JPI #2
because of the unreliability of the TAT as a criterion of
n Ach. Because of the limitations of poor TAT reliability,
researchers should be careful to employ a different n Ach
criterion in future attempts to develop alternate

measures of n Ach.

31
Whimmwmlfl

Although there is a statistically significant
relationship between the TAT and JPI #2, the relationship
is too small, on the evidence presented here, to consider
JPI #2 as a valid substitute measure of n Ach. Further
validating experiments are necessary to establish the
validity of the JPI. In view of Hehrabian's work
(1968, 1968a, 1969), such validity studies may prove
fruitful.

Hehrabian undertook construction of a multiple
choice measure of n Ach simultaneous to the initiation
of this research project. His approach was similar, that
is, creation of items from the n Ach literature, and
partial validation against the TAT criterion. His
results were markedly similar. Mehrabian's Resultant
Achievement Motivation Scale (RAM) correlated with the
TAT .28 (N=llO, p< .01), while JPI #2 correlated with
the TAT .25 (N=9u, p<.02).

In addition to using the TAT as a validating
criterion, Mehrabian employed various laboratory
problem solving tasks as n Ach criteria. For example,
Weiner, Johnson and Mehrabian (1968) used the RAM
scale in a study in which a greater Zeigarnik effect

was hypothesized for high achievers than for low achievers.

33 who scored high on the TAT-TAQ index (an individual's

32

z-score on the TAT minus his z-score on the Test Anxiety
Questionnaire, Handler and Sarason, 1952) showed a
Ziegarnik effect in 15 out of 20 cases, whereas 33 who
scored high on the RAM showed a Ziegarnik in 17 out of
21 cases. The difference between the proportion of high
versus low achieving Ss who recalled a greater percentage
of incompleted than completed questions approaches
significance (p<.10) when the TATzTAQ is used as a
measure of achievement and is significant (p( .05)
when the RAM is used.

Law (1968) investigated problem solving behaviors
of high and low achievers categorized according to
short versions of the RAM. He used match-stick problems
similar to those originally used by Wertheimer (19h5).
Groups of three to four male subjects were presented
with 11 match-stick problems and allowed 15 minutes to
solve them, after they had seen a sample solution and
were told, ”We would like to know how many you can solve
in 15 minutes.” The product-moment correlation of
achievement scores and number of problems attempted
was .38 (N=57. p<.01). Thus, high achievers tend to
persevere and attempt more problems and also solve more
problems than low achievers. Furthermore, for low
achievers Law found a .lll correlation (N=35. P< .05)
between the degree of preference for returning to

participate in a similar experiment and the subjects'

33

subjective experience of success in solving the problems.
However, for high achievers, he found no such significant
correlation (r = .17. N=35. p(.10).

Shapira (1968) hypothesized that the difference
in the performance of high versus low achievers is less
in a group achievement situation than in an individual
achievement situation. Her dependent variable was
performance on the Digit Symbol substitution subscale
of the Wechsler Adult Intelligence Scale (Wechsler, 1955)
Performance scores were the number of correct substitutions
in one minute. In the individual achievement condition,
an emphasis was placed on scores obtained by each
individual on the Digit Symbol test. Ss were told that
the test carried much weight in the determination of
intelligence level. In the group achievement condition,
an emphasis was placed on scores obtained by the entire
group. Ss did not write their names on the tests and
were told that the experimenter was interested in the
-intelligence level of UCLA students whom they happened
to represent.

The 83 were classified according to the RAM as
high (low) achievers for above (below) median scores.
Shapira found that the only significant effect was due
to the high versus low achiever condition which indicates

that irrespective of instructions, high achievers perform

better than low achievers.

3h

In addition to the findings reported thus far, there
is also some correlational data (Mehrabian, 1968)
relating the RAM to other scales. With an N of 110,
the RAM correlated with Handler and Sarason’s (1952)
Test Anxiety Questionnaire -.16 (p (.05): .29 with
the TAT-TAG). (p(.01): .6u with Rotter's (1966) internal-
external control scale (p<.01): and -.18 with the
Crowne and Marlowe (1960) social desirability scale
(p<.05).

Mehrabian (1969) reports further correlational
data with a revised version of the RAM which consists
of 26 items (the revised scale correlated .9# with
the 3% item RAM). The revised RAM correlated .20
(p<.05) with the Bass (1967) Task Orientation Scale.
.62 (p<.05) with the Jackson (1967) Achievement Scale:
«24 (p<.05) with the Liverant (1958) Social Love and
Affection Scale: .23 (p<.05) with the Guilford and
Zimmerman (19149) Sociability Scale. -.I+o (p<.05)
with the Eysenck and Eysenck (1963) Neuroticism Scale;
-.25 (p<.05) with Rokeach's (1960) Dogmatism Scale:
.56 with the Cattell and Eber (1957) Shy-venturesome
Scale, with the high achievers scoring on the more
venturesome end. And finally, the RAM correlated
.2“ (p(.05) with the Crowne and Marlowe (1960) social
desirability scale. This correlation is in the opposite
direction from an earlier study (Mehrabian, 1968).

35

With the exception of the contradictory findings with
the Crowne-Marlowe scale, these correlations are con-

sistent with n Ach research and theory.
.n.énh.as.a.flnltieDimsnsianal.Eaaiar

One of the questions raised by this research was
whether or not n Ach should be considered as a uni-
dimensional or multidimensional personality factor.

An analysis of the TAT data of Research Sessions #1
and #2 indicated that n Ach is a multidimensional
factor. Cronbach‘s (1951) coefficient alpha on the TAT
was only .23 (N=30) and .26 (N=9n) on the samples
tested. This is an analysis of variance approach to
estimating internal consistency that is analogous to the
Kuder-Richardson Formula 20. The low internal consistency
of the TAT indicates that several factors are operating
which contribute to an S's overall TAT score.

Mehrabian started with the assumption that n Ach
was a heterogeneous personality dimension. This
provided the rationale underlying the construction of
the RAM. ”Two factor analyses of the male scale yield 11
factors and the first factor accounts for no more than
15% of the total variance. The results of the factor
analyses therefore indicate that the male scale is

heterogeneous in content."1 If we accept the RAM as

 

1Albert Mehrabian. An 511311813 9.12 W
P 11“ , Englewood Cliffs, N.J.: Prentice-Hall, 1968a.

36

as a valid measure of n Ach, n Ach must be considered
as a multidimensional factor.

The low coefficient alpha of the TAT, the results
of Mehrabian's factor analysis, and the low internal
consistency of JPI #2 are evidence that n Ach is a
multidimensional personality factor. Future attempts
to construct multiple choice measures of n Ach should

consider this heterogeneous conception of n Ach.

BIBLIOGRAPHY

BIBLIOGRAPHY

Atkinson. John W. Manna mm._mmn. and 32mm.
Princeton, N.J.s Van Nostrand, 1958.

Atkinson, J.W., and Litwin, G.H. Achievement motive and
test anxiety conceived as motive to approach
success and motive to avoid failure. J

.52qu of
Abnormal and Social 25131191921. 1960. 6.0. 52~63.

A’ckinson. John No All Won 1:9. Insulation. Princeton
N.J.a Van Nostrand, 196B,

Barnette, W.L. A structured and.semi-structured
achievement measure applied to a college sample.

WMWW.
1961, 21, 6u7-656.

Bass, B.M. Social behavior and the orientation

inventory! a review. WW. 1967.
25, 260-292.

Behavioral Research Service. H9141:,2attg:ns,g£_man§ggzs
.and.£2221311§1§. New York: General Electric Co., 1960.

Birney, 3.0. The reliability of the achievement motive.

Journal of Abnormal and M W. 1959.
55. 266-267.

Brown, Roger. Social nazghglggx. New York: The Free Press,

1965.
Byrne. Donn. An W In W. Enslewood
NeJcl Prentice-Hall, 196 c

Cattell, ReBe, and Eber, H.W. S

.EQIEQDQJIIIHIEEEQZ
.Qnaatlgnnaizg. Champaign, Illinois: The Institute
for Personality and Ability Testing. 1957.

Cronbach, L.J. Coefficient alpha and the internal structure

or teStSO W! 19519 15.! 297-33“-

Crowne, D.P. and Marlowe, D.C. A new scale of social
desirability independent of psychopathology. '
£93m]. of 92mm; W. 1960. 25:. 318-351;.

.De Charms, R.. Morrison, W., Reitman, W.. and McClelland, D.C.
Behavioral correlates of directly and indirectly
measured achievement motivation. In D.C. McClelland

(Edo). mains in Milan. New York: Appleton.
19550 PP. #1h—h23.

37

Edwards. A-Lo Em 221521121 W schedule-
New York: Psychological Corporation, 195%.

French, Elizabeth G. Interrelation among some measures
of rigidity under stress and nonstress conditions.

mummsmnlzmmmx. 1955.
51, 1114-118.

French, Elizabeth C. Development of a measure of complex
motivation. In J.W. Atkinson, 1958. Pp. 2u2-2u8.

Guilford, J. Po and Zimmerman, W.S. Th: W
W W' manual of We and
lni:zn::&_£lgnfi. Beverly Hills, California:
Sheridan Supply Co., 19h9.

Guilford, J.P. W maniacs. (2nd Ed.)
New York: McGraw-Hill, l95u.

Haber, R.N., and Alpert, R. The role of situation and
picture cues in projective measurement of the
achievement motive. 1958. In J.W. Atkinson,
1958, Pp. 6hh~663.

Hays. William L. Statistics for W- New York'
Holt, Rinehart & Winston, 1963.

Heckhausen, Heinz. Hoffnung und furcht in der
leistungsmotivation. Heisenheim/Glan: Rain,
1963. Reported in Heckhausen, 196?.

Heckhausen. Heinz. Wm 62 Mon.

New York: Academic Press, 19 7.

Hurley, J.R. The iowa picture interpretation test: a
multiple choice variation of the TAT, lgnrnél 91

W W0 19559 .19.: 372-376.

Jackson. D-N-- flannel for the W meant: form.
London, Canada: Department of Psychology, the
University of Western Ontario, 1967.

Karolchuck, Patricia, and Worell, L.. Achievement
motivation and learning.

Journal of Abnormal
and .6916]. W 1956. 66. No. 1., 291-308.
Klinger, Eric. Need achievement as a motivational

construct. WM. 1966. 66.
NO. 4, 291“308.

38

Knapp, 8.3., and Garbutt, J.T. Time imagery and the

achievement motive. m 9: W,
1958, 26, h26-h3h.

Krumboltz, J.D., and Farquhar, W.W. Reliability and
validity of the n achievement test. J a

of consulting new. 1957. 21. 22 -228.

Law, Robert. Problem solving behaviors of high and low
achievers. Unpublished work, U.C.L.A. Reported
in A.M. Mehrabian, 1968.

Liverant, 3. The use of rotter's social learning theory

in developing a personality inventory. Eaxghglnglgfil
W. 1958. 12. (2. Whole No. 455),

Lowell, E.L. A methodological study of projectively
measured achievement motivation. Unpublished
master's thesis, Wesleyan University, 1950.

McClelland, D.C., Atkinson, J.W., Clark, R.A., and
Lowell. E-L- The W32 nonn- New York!
Appleton-Century-Crofts, 1953.

Mc<31e11end. D-C- The achieving mien. New York!
Free Press, 1961.

McClelland, D.C. Achievement drive and economic growth.
In D.C. McClelland (Ede). In:

IQQLfi Q:
We Princeton, NQJ03 Van NOStrand,
19 4- Pp. 16-h5. Reprinted from.§§zxgzn
M We ’40! N00 (4" July 19620 PP. 99-112.

Mandler, 6.. and Sarason, S.B. A study of anxiety and

learning. Journal of A and. 42.61.13 e
W. 1952. 6.2. 1 6-173.

Marlowe, D. Relationships among direct and indirect
measures of the achievement motive and overt

behav1or. Jamal 91 W W.
1959. 23. 329-332.

Mehrabian, A.M. Male and female scales of the tendency
to achieve. E and P

Measurement... 19 8. 26. No. 2. 493-502.

Mehrabian. Albert M. An We 61 muslin
Englewood Cliffs, N.J.s Prentice-Hall, 1968a.

Mehrabian, A.M. Measures of achieving tendencies.

Educational and W 14.2mm. 1969

in press.

39

Messick, Samuel, and Ross, John. .fleasnzmmgn;_yn
W and 69.2mm. New York! Jolrm Wiley.1962.

Mischel. W. Delay of gratification, need for achieve-
ment, acqudescence in another culture. J

_annfil
91 W M Wu 19619 62! 5’4‘3“5520

Morgan, H.H. Measuring achievement motivation with
picture interpretation. ignzn61H2£,QQn§nltlng
W. 19530 170 289‘2920

Ricks, D., and Epley, D. Foresight and hindsight in the
TAT. Paper read at Eastern Psychological
Association, N.Y., April, 1960.

Robinson, W.P. The achievement motive, academic success,
and intelligence test score. ,Britigh J

m
Social W 23161191651._1964. 5t. 98-103.

Rokeach. 14. The men nosed mind. New York!
Basic Books, 19 0.

Rosen, B.C. The achievement syndrome. A

W
W Bauer. 1956. 21. 203-211.

Rotter, J.B. Generalized expectancies for internal
versus external control of reinforcement.

WM. 1966.. .812 (1. "hole “0- 609)-
Sampson, E.E. Achievement in conflict. .lgn:nal.2£

W. 1963. 31. 510-516.

Shaffer, Laurance F. P

WWII:
.nazghglggz. (5th Ed. New York: Teacher's College
Press, 1967.

Shapira, A. The effect of shared responsibility on the
problem performance of high and low achievers.
Unpublished manuscript, U.C.L.A., 1967. Reported
in A.M. Mehrabian, 1968.

Smith, Charles P., and Feld, Sheila. An evaluation of the
objectivity of the method of content analysis.
1958. In J.W. Atkinson, 1958. Pp. 23h-2hl.

Smith, Charles P., and Feld, Sheila, How to learn the
method of content analysis.for n achievement,
n affliation, and n power. 1958. In J.W.
Atkinson, 1958a. Pp 685-818.

Tyler. Lcone E. W measurements. Englewood
Cliffs, N.J.a Prentice-Hall, 1963.

U0

Veroff, J., Atkinson, J.W., Feld, Sheila, and Gurin, G.
The use of thematic apperception to assess
motivation in a nationwide interview study.
Psychological Monographs, 1960, 7#, No. 12
(Whole No. #99).

Weiner, B., Johnson, P.B., and Mehrabian, A.M. Achievement
motivation and the recall of exam questions.

.6nrn_1J a of Educational 25121161631. 1968. 62.
Pp. 181-185.
Wechsler. D. Kennels: 66.12.11: mamas-me 5661:.
New York: Psychological Corporation, 1955.
Wertheimer M. Ezgfingfiiyg thinking. New York: Harper,
19fi5.
Williams, J.E. Mode of failure, interference tendencies,
and achievement imagery. m at Abnormal and.
M Wt 1955: 5.19 573-580.

Winterbottom, Marina R. The relation of need for
achievement to learning experiences in inde-
pendence and mastery. 1958. In J.W. Atkinson,
19580 Pp. “53'n780

ul

APPENDICES

APPENDIX A

COMPARISON OF ITEM KEYS

Appendix A reports individual item keying for purposes

of keying comparison.

JPI #1
JPI #2

87 item,glnzigzl,scaleg an True & #3 False.
#3 item scale; 26 True & 17 False.

JPI #2a = 27 item subscale of JPI #2: 16 True & 11 False.
JPI #3 = 37 item scale: 22 True & 15 False.
JPI #4 = 59 item scale; 32 True & 27 False.

Key:
#2 #2a #3 #u

f f
t t
t t
/
t f

,1.

/.

2.
3.

#2

Item

I believe that it is necessary
for Man to master his physical
and social environments.

I like to be whimsical.

All I want out of life in the
way of a career is a secure,

not too difficult job, with
enough pay to afford a nice car
and eventually a home of my own.

My parents expected me to make
friends on my own at an early age.

Even when I like the "underdog"
in a contest I generally do no;
bet on him.

I generally show improvement

in task performance rather than
performance at the same or
slower rate.

My parents did.nnt stress
independent accomplishment
when I was a child.

I have a better memory for tasks
that I have left incompleted
than for tasks that I have
completed.

“3

Key' , .
#2 #2a #3 #4
t 3/9.
10.
t 11.
t t [.12.
t 13.
r t /. 1L».
f 15.
t t 16.
, 17.
t ,. 18.
t t t t 19.
t t t 20.
f f _ 21.
r r r 22.

Item

My parents consider me to be
a highly motivated person.

I prefer bright reds to
subdued blues in my color
choices.

I enjoy traveling more than
my friends do.

In choosing work partners I
prefer successful strangers
to unsuccessful friends.

At times I feel that I am
unconsciously tryirg to please
my mOthere

I do ngt_consider myself
independent in thought.

I like to think of time as
"a large revolving wheel."
(a metaphor)

I am a member of an upward-
mobile family.

I am.ngt concerned over
accomplishment.

If my parents told me to
stop seeing a friend of my
own sex, I would see that
friend anyway.

I get the greatest satis-
faction out of playing horse-
shoes when I stand a moderate
distance from the ring rather
than far away.

Competition and success should
not be viewed as the "great
criminals” of the American
Society as a number of critics
have claimed.

Parents should no; expect
a lot of their children.

I am.ng§ energetic.

Uh

Key!
#2 #2a #3 #h

#1

t t t t t 23.
r r ,./ 21..
t t t t 25.
r t 26.
f r r 27.
r r 28.
t t 29.
f f ,, 30.
t t t , 31.
t t t r t 32.
t t t , 33.
r r , 3h.
t t t f 35.

Item

I think that everyone should
become a professional man
if he is capable.

I am.ngt more productive in
my work than the average
person.

I believe that a man should
always compete with a
"standard of excellence."

I like to think of time as
”a vast expanse of sky."
(a metaphor)

I do,ngt like to volunteer
for psychological experiments.

The prestige of my job makes
little difference to me.

,_,.
My mother developed indepen-
dence and feelings of trying
for self-mastery in me as a

youngster.

When choosing to perform a
task I tend to choose friends
instead of experts.

I am a future-oriented person.

The values that I hold are
primarily individualistic.

When I have left a task
incompleted I have a great
desire to go back and
finish it.

My parents put many
restrictions on my behavior
as a youngster.

I enjoy organizing groups
and committees.

#1

Keys

#2 #2a
f

f f

t

f f

t t

f f

t t

#3

H

45

36.

-37.

38-

39.

40.

41.

“’2.

43.
#4.

U5.

#6.

47.

U8.
#9.

Item

Parents should not expect
a lot of their children.

I like to perform routine
tasks.

Even when one gets married
his main loyalty still
belongs to his mother and
father.

Planning only makes a person
unhappy since your plans
hardly ever work out anyway.

I am oriented toward performing
tasks well for their own sake
rather than to please others.

I make more demands on myself
than others make on me.

I do net like to engage in
competitive activity.

I have a domineering father.

I like to volunteer for
psychological experiments
even when credit is not being
given for it.

I usually work at tasks just
as hard whether or not I am
going to receive a reward.

I do not consider myself
independent in action.

I think that when subjects

take psychological questionnaires
they should try hard to please
the psychologist.

I do not have a fear of failure.

I like to think of time as
"bird in flight." (a metaphor)

V
#2a #3

Item

I do no; work harder than
most people.

Much of my effort has been
to compete and strive for
success.

I do not like a good argument.

Parents should reward their
children with hugs and kisses.

I like jobs involving
responsibility for initiating
decisions.

I would rather get 310 right
now than have to wait a month
to get $30 later.

I tend to value money not so
much for its own sake but
rather because it is a concrete
measure of my success.

My mother stressed indepen-
dence training in our family.

I am confident and optimistic
about what I can accomplish.

I am a past-oriented person.

My life has an activistic
orientation.

When the time comes for a
boy to take a job, he should
stay near his parents even
if it means giving up a good
job.

When I have left a task
incompleted which I know will

be easy to finish I have great
desire to go back and complete it.

Key: ‘
#1 #2 #2a #3 #u
t t

f f f
t

t t t

f f f f

f f f f
t

f f
t t t t
t

f f t f
f f f

47

63.

6h.

65.

66.

6?.

68.

69.
70.

71.
/.72.

73-

3274.
75.
76.

J 770

Item

I am a "man of action”
as opposed to a "man of
tranquility."

If I have invested a great
deal of time and energy in

a project and find that I
have been making a funda-
mental error, I generally do
ngt take the time to go back
to the beginning and start
over again.

I prefer moderate risks

(3 out of 6) to either safe
(5 out of 6) or speculative
(1 out of 6) risks.

I feel Tthat I conform less
than do some of my peers.

I believe that independence
is no; that important a virtue.

It is more important to be
socially acceptable than
to be a success.

I do not consider myself
an active person.

I like to take calculated
TISkSe

I do,ngt like to innovate.

I work hardest when I know
that my personal efforts will
make a difference in the
outcome of a project.

Parents should teach their
children to be self-reliant.

I do get like hard work.
I do not like to gamble.

I consider myself to be a
person of originality.

I like to engage in under--
takings where there is a
challenge.

Keys

#2 #2a #3

f f f

f f

t t
f

f f

t

t t t
f

t

#1.

NB

. 78.

79.

80.

81.

82.

83.

84,

85.

86.

, 87.

Item

My father tends to be
dogmatic.

As a child my mother often
rewarded me with "hugs and
kisses" when I tried my best.

Usually I will ngj,volunteer
for a difficult task.

I prefer extreme risks
(1 out of 6) to moderate risks
(3 out of 6).

I am uncomfortable when I
know exactly how well I am
performing on a task.

My father developed indepen-
dence and feelings of trying
for self-mastery in me as a
youngster.

I like to think of time as
”marching feet.” (a metaphor)

I dolngt like to perform
somewhat difficult tasks.

Others consider me to be high
in need for achievement.

I consider myself to be high
in need for achievement.

APPENDIX B
RELIABILITY OF TAT SCORING BY MORIARTY

Appendix B reports the reliability of TAT protocol
scoring by Moriarty. Moriarty learned the method of
n Ach scoring by studying the manual "How to Learn the
Method of Content Analysis for n Achievement, n Affliation,
and n Power" (Smith and Feld, 1958). Smith and Feld
report that "approximately three out of four persons
who complete the training procedure outlined here can be
expected to be ready to score protocols for research
purposes with very little additional practice; Inter-
judge reliability in the .90's is, of course, desirable,
but a number of useful studies have been conducted with
scoring reliabilities in the .80's."1 Moriarty's
percentage agreement in scoring achievement imagery
against the expert coding was in the mid 90's. This
is more than adequate for research scoring.2

Two criteria must be met to become a reliable
scorer. (1) There must be a high percentage agreement

on the prescence of motive-related imagery between the

——

1Charles P Smith and Sheila Feld, How to Learn the
method of content analysis for n achievement, n affliation,
and n power. 1958. In J.W. Atkinson, 1958. P. 691.

2"Most important is to be sure you can code reliably.
Check against expert coding in,book. More than adequate
means r a .90 or better.” Personal correspondence with
Atkinson.

“9

50

novice and the expert. This is "the single most important
scoring decision to be made."1 (2) There must be a
high rank order correlation on total score assigned to
each story between the novice scorer and the experts.
Moriarty achieved a mean percentage agreement in
scoring achievement imagery with the experts of .9153
his median agreement was .96. The norms presented in
the manual between novice scorers and experts for practice
story sets B-D were a mean of .856 and a median of .8h.
Moriarty's mean rank order correlation with the experts
on total score of the sample stories was .87; his median
rank order correlation was .91. The norms presented in
the manual between novice scorers and experts for the
rank order correlations on total scores were a mean of
.78 and a median of .785. Thus Moriarty's scoring was
well above that of the novice scorers in the manual and
correlated more than adequately with the expert scorers.
Table 2 (adapted from Atkinson, 1958, P. 238)
reports the percentage agreement in scoring achievement
imagery and the rank order correlations between n Ach
scores obtained by novice scorers and experts. It also
reports Moriarty’s percentage agreement and rank order

correlations with the experts on the sample scoring stories.

1

Charles P. Smith and Sheila Feld, An evaluation
of the objectivity of the method of content analysis. 1958.
In J.W. Atkinson, 1958. PP. ZBLZulc

Table 2.
Percentage Agreement in Scoring Achievement Imagery
and Rank Order Correlations between n Ach Scores Obtained

by Novice Coders/Experts and Moriarty/Experts

Subject Reliability Story Set Mean Median
Index B C D E F

Median of Z .82 .91 .8h * * .856 .au

Novice Scorers Rho .785 .805 .755 * * .78 .785

Moriarty % .96 1.00 .88 .7U#.96 .915 .96

Rho 097 .91 079 072 095 087 091

* = data not available

ER
fl

percentage agreement on the presence of motive-
related imagery between novice scorers and experts

Rho = rank order correlations on total score assigned
each story between novice scorers and experts

51

cats ‘98:}

HICHIGRN STATE UNIV.

 

32||||3|||3|333

LIBRQRIE