An attempt to develop a falsification key for the Guilford-Martin Personnel Inventory

324 74 3MB

English Pages 83

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

An attempt to develop a falsification key for the Guilford-Martin Personnel Inventory

Citation preview

AN ATTEMPT TO DEVELOP A FALSIFICATION KEY FOR THE GUILFORD-MARTIN PERSONNEL INVENTORY

A Thesis Presented to . the Faculty of the Department of Psychology The University of Southern California

In Partial Fulfillment of the Requirements for the Degree Master of Arts

by Peter Richard Ekstrom June 1950

UMI Number: EP63984

All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion.

Dissertation Publishing

UMI EP63984 Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code

ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 4 8 1 0 6 - 1346

Ps T his thesis, w ritten by ..... under the guidance of h.XB-r F a c u lty C om m ittee, and approved by a l l its members, has been presented to and accepted by the C ouncil on G raduate Study and Research in p a r t ia l f u l f i l l ­ ment of the requirements f o r the degree of ..............M.STM...OT...ARTS............

.................H. J.DSUEL3l.sIrA.... Dean D ate.

Ma.3L.1950._______

Faculty Committee

TABLE OF CONTENTS CHAPTER I.

II.

PAGE

INTRODUCTION

...........

1

The statement of the p r o b l e m ...............

2

Outline of the t h e s i s .....................

2

Importance of this s t u d y ...................

2

REVIEW OF THE L I T E R A T U R E ......................

4

Literature pertaining to the ability of individuals to falsify self-inventory tests

.

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

4

Literature pertaining to the prevention or correction of falsification on selfinventory t e s t s ....................... Summary III.

.

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

14

METHODS AND P R O C E D U R E S ........................

16

The measuring device .......................

16

Constructing a falsification key ...........

17

The s u b j e c t s ..............................

17

The method u s e d .........................

17

Applying the falsification k e y s ...........

IV.

8

22

The s u b j e c t s ..........

22

The method u s e d ...................

2^

S u m m a r y ....................................

24

RESULTS AND D I S C U S S I O N ........................

2J

ill CHAPTER

PAGE Results and discussion

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

27

The nature of the lie k e y s ................

27

Reliabilities of the liek e y s ............

28

The honesty i n d e x e s .....................

20

Correlations ..............................

32

Attempts to increase prediction of the c r i t e r i o n .............................. Summary V.

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

SUMMARY AND C O N C L U S I O N S .....................

.

38 46 48

B I B L I O G R A P H Y .........................

52

A P P E N D I X ...................

54

LIST OP TABLES TABLE I.

PAGE CR*s Used as Cutting Points to Separate Easy-to-Influence and Hard-to-Influence I t e m s ................................

II. III.

Item Division for Estimating Reliabilities

.

29

Reliabilities Estimated by the Spearman-Brown Prophesy Formula

IV.

27

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

31

Means and Standard Deviations of Honesty Indexes, Factor Co, and Criterion for 287 F o r e m e n ....................................

V.

34

Product-Moment Correlation Coefficients between Honesty Indexes and Factor Co.

Significance of Correlations (N=287) VI.

...

35

Product-Moraent Correlation Coefficients between the Honesty Indexes and Supervisory Ratings.

Significance of Correlations

(N=287)............... VII.

Means, Standard Deviations, and PEARSON r*s for Corrected Cooperativeness Scores

VIII.

56

...

40

Means and Standard Deviations of Easy-toInfluence and Hard-to-Influence Co Responses

42

V

TABLE IX.

PAGE Product-Moment Correlations for Easy-toInfluence Responses (E), Hard-to-Influence Responses (H) with Co and the Criterion

(N=28t ) ......... X.

45

Proportion of College Students Answering Yes, Ho, and ? to Each Item of the GuilfordMartin Personnel Inventory under Experi­ mental Conditions I and II.

Critical Ratios

of the Differences ( N = 1 5 ^ ) ............

.

69

CHAPTER I INTRODUCTION The use of pencil and paper personality inventories as employee screening devices has met with much criticism. A good deal of this criticism has been leveled at these measuring instruments because of their inherent weakness of depending upon the testee to evaluate his own feelings and personality traits. The fact that these convenient tests are invalidated when used in a competitive employment situation because of conscious or unconscious falsification has been recognized by many psychologists.

Steinmetz^ was among the first to

test the fakeability of self-inventories.

Bernreuter

summarized the weaknesses of self-inventories when he wrote: There appear to be three main criticisms of the use of self-ratings in estimating personality traits. Briefly they are: first, that scores on such tests are due merely to chance; second, that they are rendered invalid by a tendency on the part of subjects to answer in such a manner that their responses will meet with social approval, and third, that they are rendered invalid by desire on the part of subjects to be other than they are.

^ H. C. Steinmetz, “Measuring Ability to Fake Occu­ pational Interest,” Journal of Applied Psychology, 16:123-30, 1932. o Robert G. Bernreuter, “Validity of the Personality Inventory,” The Personnel Journal, 11:303-86, April, 1933.

2 In spite of the studies and comments referring to this weakness, there have been relatively few studies devoted primarily to any method of correcting or allowing for the falsification that exists. The statement of the problem.

The main objective of

this study was to develop a new key for the Guilford-Martin Personnel Inventory that would measure the degree of honesty with which the test was taken. The second purpose was to determine the usefulness of such a device as a means of improving prediction of employee success from the Guilford-Martin Personnel Inventory. Outline of the thesis.

In the next chapter the

history of falsification or honesty keys will be discussed, thus laying the groundwork for the present study.

Chapter

Three will discuss the methods and procedures used in developing the new key and also the method used in applying the new key to 287 foremen to determine its usefulness. In Chapter Four the results of the analysis of the data will be presented, along with the conclusions that are warranted from these data.

Chapter Five is a summary of the study,

results, and conclusions. Importance of this study.

The selection of the best

man available for a given job has long been a goal of those

5 doing job placement.

As yet, there has been little work done

to make pencil and paper tests of personality or any selfevaluating scheme useful in accomplishing this end.

The

problem of proper employee selection affects more than just the employer in an economic.manner.

When the best person

available for a given job is selected, the employee as well as the employer has benefited.

CHAPTER II REVIEW OF THE LITERATURE Since Woodworth constructed the first self-inventory of personality in 1917* the psychological literature has been overflowing with various types of personality tests. In the great majority of these tests the author states that frankness on the part of the subject has to be assumed.

In

a competitive situation where individuals may believe it to be to their advantage to make a good showing, this assumption is not warranted. Compared to the total number of self-inventory tests available there are only a very few studies reported that make an attempt to quantify the amount of falsification possible on a test or to develop means of correcting for this error once it has occurred.

This chapter will review the

literature available on two topics.

The first deals with

those studies that demonstrate the ability of individuals to simulate the behavior of other persons in so far as a self­ inventory score is concerned.

The second section pertains

to the literature dealing with methods of detecting and correcting for any faking that may occur. Literature pertaining to the ability of individ­ uals to falsify self-Inventory tests.

The first study to

5 be published that demonstrated the ease with which self­ inventories may be falsified was made by Steinmetz-^ in 1932. He gave the Strong Vocational Interest Blank under two conditions to forty-six students. given under standard conditions.

The Strong test was first After a six-week interval

the same subjects were asked to fill out the inventory in a manner which they thought would qualify them as male school-teacher administrators.

As a result of this pro­

cedure Steinmetz states: Students are able intentionally to distort their scores on an interest blank, and to succeed in quali­ fying well for an occupation chosen at random, so far as they are concerned, despite a low average initial predilection. Steinmetz observed that those individuals who were most unlike school administrators under standard conditions were able to change their scores the most.

The only other

variable related to the amount of falsification possible was specific knowledge about the job of school administrator. In a study by Kelly, Miles, and Terman^ nineteen men and thirty-three women junior college students were adminis­ tered the Stanford test of masculinity-femininity under

^ Steinmetz, loc. cit. A Ibid., p. 129. ^ E. L. Kelly, C. C. Miles, and C. M. Terman, "Ability to Influence One’s Score on a Pencil-and-Paper Test of Per­ sonality," Character and Personality, 4:5^8-64, 1934.

6 standard conditions.

One week after this initial testing

the group was split into two sections.

One section was

asked to make themselves appear extremely masculine regard­ less of their true sex.

The second section was asked to make

themselves appear as feminine as they could.

Two weeks after

the original testing the instructions were reversed for the two sections.

All subjects took the test under three

conditions. The results of this study show that apparent mascu­ line and feminine traits can be presented by either sex when told to falsify their answers.

The men were able to achieve

a M-F score more feminine than that received by the average women under standard conditions.

Similarly., women were able

to appear more masculine than the average male.

The ability

to shift one*s score did not appear to be related to intelligence, age, or scholastic achievement. Hansen

using the Humm-Wadsworth Temperament Scale

had a group of college students take the test under standard conditions.

Those subjects who were classified as the "poor-

risk” group because of a low "Normal11 score or because of an elevated Hysteroid or Paranoid component were asked to take

6

Alice Viola Hansen, "An Experimental Study of the Ability to Simulate Humm-Wadsworth Temperament Performance," (unpublished Master1s thesis, The University of Southern California, Los Angeles, 1942), pp. 1-44.

7 the test a second time.

The instructions given this group

were to answer the question as if they invariably followed the Golden Rule.

Hansen summarizes her results as follows:

In general, the ability to simulate a desired HummWadsworth Temperament performance has been demonstrated. Although this does not invalidate the scale when used for self-analysis or for vocational guidance/ it does make it impossible to consider its results valid when it is used for hiring or placing employees. The per­ sonality test needed for employment work is one which will test the honesty with which the subject has taken the test as well as the components of his t e m p e r a m e n t .7 Kimber

8

gave the California Test of Personality to

389 beginning psychology students.

They were asked to

simulate the profile of a happy and well adjusted student. Two weeks later they were given the test again under stand­ ard conditions.

In addition to these data Kimber secured

Army Alpha scores on his subjects as a measure of intelli­ gence.

The conclusions drawn from this study indicate

that (l) individuals have Insight into the meaning of items of a personality test; (2) individuals differ greatly in the degree of insight which they possess; their susceptibility to insight;

(3 ) items vary in

(4) intelligence bears

little relationship to insight.

7 Ibid., p. in. ° J. A. Kimber, "The Insight of College Students into the Items of a Personality Test,11 (unpublished Doctorfs dissertation, The University of Southern California, Los Angeles, 19^5)> PP* 1-186.

8 Now that it has been established that the self­ inventory type of test is not immune to faking, the question arises as to what can be done to prevent or correct for this error. Literature pertaining to the prevention or correction of falsification on self-inventory tests. Hathaway

9

Meehl and

discuss eight possible solutions to either elimi­

nate or correct for falsification on pencil and paper tests. They are: 1.

The test instructions may contain special appeals’

to the persons taking the test asking them to be frank. 2.

The items of the test are disguised so that the

significance of a given response is not obvious. 3.

Half of the items are stated in a negative manner

so that all keyed items are not true and good. 4.

The subjects are kept anonymous.

5.

The subjects are asked to sort cards with printed

statements on them into various stacks representing the acceptance or rejection of the statement.

It is felt that

this gives the subject a greater degree of freedom than leaving a written record of his responses.

9. P. E. Meehl and S. R. Hathaway, "The K factor as a suppressor variable in the MMPI,” Journal of Applied Psychology, 30:525-64, 1946.

9 6.

A number of questions are included in the test to

indicate the readiness of the individual to confide his true feelings. 7.

This can be used as an index of frankness. Items may be repeated in a test to check the

consistency with which they are answered in the same way. 8.

Opportunities are presented for answering items

in a very favorable way, but in a way which could almost certainly not be true.

If too many items are answered

favorably it casts doubt on the validity of the remaining items. Items 1, 2, 3* and 5 are preventive measures.

Such

procedures may help, but they give no assurance that falsi­ fication will not occur. industrial situation.

Item 4 would be useless in an

Items 6, 7* and 8 give evidence that

faking has occurred and give an attempt at quantifying the extent of falsification.

On this subject Hosenzweig stated:

Astute phraseology in the instructions and questions of the test have sometimes been resorted to, but such expedients are rarely very effective. Might it not be more effective to recognize at the outset that such tests have certain limitations that can never be completely circumvented and then go on to the measurement of these limiting factors themselves, thus obtaining information by which a correction may be applied to the subject*s

answers?^

10

S. Hosenzweig, "A Suggestion for Making Verbal Personality Tests More Valid,” Psychological Review, 41:400, 1934.

10 Humm and Wadsworth‘S

were among the first investiga­

tors of the self-inventory to lay great stress upon the problem of detecting non-cooperation and distortion of responses when evaluating a particular profile score.

The

method used is to count the number of “no** responses given to the keyed items on the Humm-Wadsworth Temperament Scale. Inasmuch as J6 per cent of the scored items of the test are obviously suggestive of abnormality when replied to in the affirmative, the “no-count*1 is to some extent a measure of the s u bjects tendency to avoid, consciously or otherwise, the saying of "bad” things about himself when taking the test.

If the “no-count** of an individual is excessive, it

follows that the subject has responded in a defensive manner. The validity of such a test profile would be doubtful.

The

procedure described by these authors would give an indica­ tion that cheating had been done on the test, but it would not give any solution as to how this error should be corrected.

The only method available would be to readmin-

ister the test to the subjects telling them that the “yes” and “n o ” answers should be approximately equally divided. This readministration method makes a large percentage of the

■S D. G. Humm and G. W. Wadsworth, “The HummWadsworth Temperament Scale,** American Journal of Psychiatry,

92:163-200, 1935*

11 biased test acceptable.

12

The basis for the present study is an investigation performed by Ruch.

13

In a comprehensive study of the

Bernreuter Personality Inventory, Ruch developed a new method of evaluating the degree of faking that persons indulge in when taking a self-inventory test. Ruch stated that the main purpose of his study was to develop a scoring stencil that would reveal the extent to which scores on this test had been influenced by the subject. Two hundred and forty-five male elementary psychology students were given the Bernreuter test under standard conditions.

After they had completed the form, they were

asked to fill out another form with the following instruc­ tions : Imagine that you are applying for a position as a salesman. Your showing on this test will decide whether or not you get the job. You know the characteristics of a good salesman. See if you can answer these questions as a good salesman would, whether you really feel that way or not.1^ The standard and influenced blanks were scored for extroversion-introversion.

The group taking the test under

12 Hansen, pp. pit., ,p. 43. 1^ P. L. Ruch, flA Technique for Detecting Attempts to Fake Performance on the Self-Inventory Type of Personality Test,” Studies in Personality (New York: McGraw-Hill Book Company, Inc., 1942), pp. 229-34. Ibid.j p. 52.

12 standard conditions had a median score corresponding to the college median score on Bernreuter*s norm.

The median of

the influenced group fell at the 98th percentile for extroversion.

Ruch states:

The next phase of the investigation consisted in building a scoring stencil to reveal faking after the fact. . . . The argument is rather simple. If answers to items on a test like the Bernreuter can be faked at all, the chances are that some are easier to fake than others. Therefore, it should be possible to give each item a weight to represent the extent to which it can be faked by the average college student. This was done by tabulating the frequency of each answer to each question for the standard condition and for the influenced condition. These frequencies were converted into per­ centages, and an ’’honesty 11 weight was assigned to each reply according to the magnitude of the critical ratio of the difference between the frequency of the reply in the honest and in the influenced condition.15 By this method an "honesty score" could be obtained for anyone taking the Bernreuter test.

In checking the new

scoring testing technique Ruch found that he could differ­ entiate all introverted persons who were attempting to appear extroverted.

However, the key was not as efficient in

detecting extroverted individuals who attempted to make them­ selves more like a salesman. A technique very similar to the one proposed by Ruch was used by Goldstein

i££* °it

16

to detect malingering in the army.



H. Goldstein, f,A Malingering Key for Mental Tests Psychological Bulletin, 42:104-18, 1945.

13 An intelligence test was given to three groups of men: (l) a good group, which had passed the examination;

(2) a

failure group that had failed the examination; and {3) & malingerer group composed of men screened by the army psychologist who had the capacity to pass the examination but who had failed intentionally in order to escape mili­ tary service. Goldstein did an item analysis of the data received from these three groups.

Those items that were passed by

more failures than malingerers were considered easy; those items that were passed by more malingerers than failures were considered hard.

By basing a key on the differences

obtained Goldstein was able to differentiate 75 to 90 per cent of test failures as non-malingerers.

The staff was

then in a position to concentrate their efforts upon the remaining 10 to 25 per cent of the men as possible malin­ gerers, thereby saving much effort and time.

Goldstein’s

study indicates that the method devised by Ruch is appli­ cable to more than personality testing. McKinley, Hathaway, and M e e h l , b e i n g aware of the shortcomings of the self-inventory type of tests, have developed a new scale for use with the Minnesota Multiphasic

J. C. McKinley, S. R. Hathaway, and P. E. Meehl, ,fThe MMPI: VI the K Scale,11 Journal of Consulting Psychology, 12:20-51, 19^8.

14 Personality Inventory.

The ftK scale” was developed by

studying the item response frequencies of diagnosed abnormals who had normal profiles on the MMPI.

It was assumed that

these abnormals obtained a normal profile because of a defensive attitude.

Their response frequencies, from an

item analysis, were compared with a group of normal subjects. The differentiating items were weighted so that a high score would be found among abnormals with a normal profile and a low score would be found in clinically normal persons having a deviant profile.

The authors evaluate the ”K scale” as

follows: In this operational sense it can be said that a high K score is indicative of a defensive attitude and a low score suggests unusual frankness or self criticality (”plus getting”). The extremes of defensiveness and plus-getting may be called ”faking good” and ”faking bad” r e s p e c t i v e l y . ^ Sumrnary.

A review of the literature pertaining to

the falsification of pencil and paper tests of personality has established the fact that the self-inventory type of test is extremely susceptible to falsification.

The error

due to falsification is at Its zenith when the individuals are in a job-getting frame of mind.

Whenever individuals

have been motivated to distort their true personality as measured by a test and assume some favorable traits that are

Ibid., pp. 20-21.

15 supposed to be needed for success in a field of endeavor, they have succeeded to an alarming degree.

The ability to

fake tests was found not to be related to age, intelligence, or scholastic achievement. Some possible means of preventing cheating were discussed with the general conclusion drawn that the pre­ ventive measures proposed are not adequate to prevent a person from faking if he so chooses.

The only alternative

then appears to be to detect cheating after it has occurred and to evaluate the magnitude of the error.

With this

knowledge, the true score of falsified profiles could be estimated.

The technique developed by Ruch that enables us

to evaluate this error was presented as the basis for this present study.

CHAPTER III METHODS AND PROCEDURES The first section of -this chapter is devoted to a description of the Guilford-Martin Personnel Inventory. Section II relates the methods and procedures used to obtain a falsification key for the Guilford-Martin test.

Section

III describes the procedures used in evaluating the new key. Section IV summarizes the methods and procedures used in this investigation.

I.

THE MEASURING DEVICE

The Guilford-Martin Personnel Inventory1^ was developed to accomplish two purposes.

The first of these

was to extend the list of factors measured by other person­ ality tests developed by the same authors to include those of cooperativeness, agreeableness, and objectivity.

The

second purpose was to develop a test capable of locating potential trouble-makers before their being placed on the pay-roll. The scales are defined by Guilford and Martin as: 19 J. P. Guilford and H. G. Martin, The GuilfordMartin Personnel Inventory, Manual of Directions and Norms (Beverly Hills, California: Sheridan Supply Company, 1943), 333 pp.

IT O-Qbjectivity. This is opposed to personal reference or a tendency to take things personally. Ag-Agreeableness. This is opposed to belligerence or a dominating disposition and an overreadiness to fight over trifles. Co-Cooperativeness. This is opposed to faultfinding or overcriticalness of people and things.^0 The reliabilities of these traits using the splithalf method and then estimating by the Spearman-Brown formula were .83 for 0, .80 for Ag, and .91 for Co.

The

intercorrelation of 0 - Ag is .64, 0 - Co is .55* and Ag Co is .65 .

II. 1.

CONSTRUCTING A FALSIFICATION KEY The subjects.

The subjects of this study were

154 male college students enrolled in elementary psychology. Their median age was 22 years with a range of 18 to 60 years. Most of the subjects were sophomores at the time of the testing although there were students from all four classes and several graduate students included in the sample. ffhe method used.

On October 24, 1949* the sub­

jects were requested to take part in an experimental study. After all of the material to be used had been distributed, the following instructions were read:

20

Ibid., p. 4.

18 You have been given a copy of a standard personality inventory. We want to see how you analyze human reac­ tions. This is what we want you to do. Answer the questions in this inventory so as to make the best impression of yourself if you were applying for a job as an industrial supervisor. Make yourself look as good as possible. An industrial supervisor must beable to work with people; he must accept orders from his superiors; he must transmit orders to his subordinates; and, finally, he must be able to work effectively with others on the same level. The results of this inventory will not have any effect on your grade in this class; however, attendance for today will be taken from the answer sheets turned in. Remember, just answer the questions the way a good supervisor would, whether you yourself are that way or not. Be sure to answer every question. Go ahead. In order to emphasize and clarify the instructions the following was placed on the blackboard: A Good Industrial Supervisor A. B. C.

Works with people. Takes orders. Gives orders.

If you have to fake your answers thatrs 0. K.— make yourself look good. - Three weeks later the subjects were asked to parti­ cipate in the second half of the study.

At this time the

following instructions were read: As- you remember, you were given this same personality inventory about three weeks ago. At that time you were asked to answer the questions as a good industrial supervisor would answer whether you yourself were that way or not. This time we want you to answer as a poor industrial supervisor would answer. A person not qualified to be an industrial supervisor will not be able to work with people; he will resent orders given to him by his superiors; he will not be effective in transmitting orders to his subordinates; and, finally,

19 he will not be able to work effectively with others on the same level. The results of this inventory will not have any effect on your grade in this class; however, attendance for today will be taken from the answer sheets turned in. Remember, just answer the questions as a poor industrial supervisor would answer whether you yourself are that way or n o t . Be sure you answer every question. Again an outline of the instructions was placed on the blackboard.

This read: A Poor Industrial Supervisor A. B. C.

Is unable to work with people. Resents orders. Is Ineffective at giving orders.

If you have to fake your answers that’s 0. K.— make yourself look poor. During the t!fake good” condition, condition I, the subjects were not aware of the nature of the study.

They

were quite surprised to be asked to simulate a poor super­ visor under condition II.

The rapport of the subjects was

felt to be good and the results appeared to be the best available for these conditions. The above procedure netted 15^- subjects who had taken the Guilford-Martin Personnel Inventory under both of the conditions.

An item count was made of the 150 Items of the

test to determine the number of times each Item was answered “yes," “no,” or ”?" for each of the conditions. The next step was to compare the proportion of

20 subjects answering a given way to each item under both conditions.

For example, the proportion of subjects

answering ”yes” to item number one under condition I was compared with the proportion answering “yes” to the same item under condition II.

The ”n o ” and

proportions

were compared in a like manner. The differences obtained by comparing similar responses under condition I and condition II give a scale that does not have the property of equal units.

Therefore,

if the proportion of people answering item one “yes” under condition I were .90 and under condition II .95* this obtained difference of .05 is not identical with a differ­ ence obtained if the proportion had been .48 and .55 respectively.

For this reason the differences were expressed

as critical ratios. With the differences expressed as critical ratios it was then possible to determine which responses changed the most from condition I to condition II.

Those Items to

which a large proportion of the subjects answered a given way under condition I and to which a small proportion answered the same way under condition II were considered easy-to-influence items.

This change was expressed by a

comparatively large CH.

Similarly, a small CR denoted an

item which was hard to influence. In order to determine a cutting point between items

21 that were hard to influence and items that were easy to influence, three distributions were made.

These were

distributions of the critical ratios obtained by comparing "yes" responses -for condition I with "yes” responses for condition II, "no" responses for condition I with "no" responses for condition II, and I with

responses for condition

responses for condition II. The medians of these distributions were computed.

Those items with C R ’s below the median were the hardest to influence, and those above the median the easiest to influence. Those items appearing on the standard key of the Guilford-Martin test were assigned plus and minus values corresponding to easy-to-influence and hard-to-influence items respectively.

This new key, key A, was then ready

for use. A second key, key B, was then made. constructed in a fashion similar to key A.

This key was Key A used the

median of the distribution of critical ratios as the dividing point between easy-to-influence and hard-to-influence items.

Key B eliminated 'the middle third of the items and

used only the thirds on either extreme as easy-to-influence and hard-to-influence items.

The items on the standard key

that had critical ratios falling in the highest third were assigned pluses; those in the lowest third were assigned

22 minuses.

This key was then ready for use.

III.

APPLYING THE FALSIFICATION KEYS

In order to evaluate the usefulness of the lie detection keys, they were applied to the results of a group of industrial foremen and supervisors. T*

subjects»

The subjects were 287 male foremen

and supervisors in charge of production at a large eastern 21 manufacturing plant. These men had taken a battery of tests, one of which was the Guilford-Martin Personnel Inventory.

All testing was done under standard conditions.

The men were informed that the results of the testing would not affect their Job status in any manner.

It was

explained that the standards set by them might be used to determine standards by which others could be selected for similar jobs.

There was virtually no opposition to the

testing program. Each subject was rated independently by three of his immediate superiors.

The ratings assigned were from 1 to 5>

5 being the highest attainable.

The raters were asked to

proportion their judgments into an approximate normal distribution.

21

The average of these three ratings was used

Name withheld by request.

23 as the criterion with which to compare the lie score. The method used.

The first step in evaluating

the lie keys was to score the tests of the 287 foremen using key A and key B.

Of the total number,of items answered* the

number- of those being easy-to-influence and hard-to-influence was recorded. The analysis from this point on was confined to the variable* cooperativeness (Co).

There were two reasons for

limiting the analysis. The first was that the results of a 22 study by Mackie using the same population showed the factor Co to be the only one on the Guilford-Martin Personnel Inventory to be essentially rectilinearly related to the criterion of supervisory success.

The second reason was that

it was felt that it was a better procedure to explore one factor thoroughly than to use the time available doing a partial study of the three factors. The next step in the analysis was to determine the reliabilities of the easy-to-influence and hard-to-influence scores for the 287 foremen. It was then necessary to find a method of utilizing the data received by the rescoring of the tests using key A

00

R. R. Mackie* "Norms and Validity of Sixteen Test Variables for Predicting Success of Foremen*” (unpublished M a s t e r ^ thesis. The University of Southern California* Los Angeles* 1948)* 58 pp.

24^ and key B. needed.

Some measure of the degree of falsification was

Three methods of combining the number of easy-to-

influence and hard-to-influence items answered by the subjects were tried in an attempt to develop a suitable lie scale.

The correlations between these lie scales and factor

Co were computed as were those between the lie scales and supervisory ratings.

As a final step of the investigation,

the relationships between the two components of the lie scales with the criterion and cooperativeness scores were analyzed.

IV.

SUMMARY

Chapter Three has described the methods and proce­ dures used in this investigation.

The procedure can be

outlined as follows: 1.

The Guilford-Martin Personnel Inventory xvas

administered to 154 college men under two conditions,

tffake

good” and ”fake poor.” 2.

The differences in the way items were answered

under the two conditions were quantified. 3.

Two lie keys were built by using the magnitude of

these differences as indicators of the influenceability of each item:

key A using all items appearing on the standard

scoring key, and key B using the extreme thirds. 4.

Two hundred and eighty-seven inventory blanks

25 filled out by industrial supervisors were rescored, using key A and key B. 5*

Honesty indexes were formulated to express the

degree of falsification. 6.

The interrelation of the honesty indexes,

supervisory ratings, and cooperativeness scores was explored.

CHAPTER IV RESULTS AND DISCUSSION In this chapter the results obtained by the statis­ tical analysis of the data are presented and discussed.

A

short summary of the major points is included at the end of of the chapter.

II.

RESULTS AND DISCUSSION

The nature of the lie keys.

The method used to

develop the lie keys was discussed in the preceding chapter. As has been stated* key A utilized the full distribution of critical ratios; key B used the extreme thirds.

Table I

represents the.cutting scores used to distinguish hard-toinfluence and easy-to-influence items. By using these values, the hard-to-influence and easy-to-influence items were determined for the standard scoring key of the Guilford-Martin test. of 66 possible

Key A had a total

responses for the factor Co.

Thirty-five

were easy-to-influence and 31 were hard-to-influence items. Key B had a total of 44 Co responses.

The breakdown of these

44 items was 23 easy-to-influence items and 21 hard-toinfluence Items.

It is interesting to note that there were

34 items of the total 143 keyed responses on the GuilfordMartin Personnel Inventory which were influenced in the

27

TABLE I C H ls USED AS CUTTING POINTS TO SEPARATE EASYTO-INFLUENCE AND HARD-TO-INFLUENCE ITEMS

Key A

Yes

No

Key B

Easy-toinf luence items

Hard-toinfluence items

Easy-toinf luence items

Hard-toinfluence items

12.22 and above

Below 12.22

Above

Below

17.05

8.68

11.10

Below 11.10

Above

15.90

Below 7.84

1.56

Below

Above

Below

and above

1.56

1.91

1.10

and above

28 wrong direction.

These items were responded to by a major­

ity of the college population in a manner opposed to the "correct11 response given by the standard scoring key.

There

were 11 such responses keyed for factor 0, 10 keyed for Ag, and 13 for Co. influence.

All but one of these items were hard-to-

The discussion from this point on has been

limited to the keys for factor Co.

Keys and some data for

factors 0 and Ag are appended to this study. Reliabilities of the lie keys.

The reliabilities of

the lie keys were found by the split-half method and esti­ mated by using the Spearman-Brown formula:*^ r l{ ^

___ *

Easily influenced items on pages one and three of the test were compared with those on pages two and four.

The same

comparison was made of hard-to-influence items.

Table II

represents the total number of hard- and easy-to-influence items appearing on pages one and three as compared to those on pages two and four. From Table II it is obvious that the requirement of comparing equal halves of a distribution has not been met completely.

However, any deviation from an equal split will

23

J. P. Guilford, Fundamental Statistics in Psychology and Education (New York: McGraw-Hill Book Company, Inc.,

19IT2T7'p. 275.

29

TABLE II ITEM DIVISION FOR ESTIMATING RELIABILITIES

Key A Easy-toinfluence items

Key B

Hard-toinfluenee items

Easy-toinfluence items

Hard-toinfluence items

Pages 1 & 3

14

18

6

12

Pages 2 & 4

21

15

17

9

Totals

55

51

23

21

30 tend to decrease the obtained reliability.

Therefore* the

estimates made of reliability err on the safe side by estimating lower reliabilities than are actually character­ istic of the data. Table III presents the reliabilities of the lie keys as estimated by the Spearman-Brown prophecy formula.

In

the light of the error mentioned above* these reliabilities seem to be sufficiently high. The honesty indexes.

An index of honesty on frank­

ness was needed that would not be limited by the number of questions answered on the Personnel Inventory. Honesty Index I was defined as two times the number of keyed hard-to-influence items answered* divided by the number of easy-to-influence items plus the number of hardto-influence items.

The ratio was then multiplied by 100.

The formula for this index is expressed as follows: Honesty Index I « 2 x hard-to-influence items x 100__________ easy-to-influence + hard-to-influence items In a situation where the hard-to-influence items' are equal in number to the easy-to-influence items* an indivi­ dual should receive an index of 100 if he is absolutely honest.

The lower this index is* the greater is the falsi­

fication that has been indulged in by the subject. Honesty Index II was a simple difference score. number of keyed hard-to-influence items answered was

The

31

TABLE III RELIABILITIES ESTIMATED BY THE SPEARMAN-BROWN PROPHESY FORMULA

Key A

00 o

Hard-toinfluence items



Easy-toinfluence items

.72

Key B

•73

.72

52 subtracted from the number of keyed easy-to-influence items answered*

The smaller this difference, the more frank was

the individual taking the test. Honesty Index III was developed in an attempt to circumvent the shortcomings of Honesty Indexes I and II. The third index was a ratio score.

Honesty Index III was

computed as follows: 1.

Compute the lowest possible difference score that

each individual could obtain considering the number of keyed responses that were answered.

For key A, if a person

answered a total of 40 keyed responses, his lowest difference score would equal 9 - 51 = -22.

This assumes he answered

correctly every hard-to-influence item possible (5l). 2.

Compute the highest possible difference score

for each individual.

In the above example using key A and

a total of 40 responses, the highest score obtainable would equal 35 - 5 = 30.

This assumes that the subject answered

correctly every easy-to-influence item appearing on key A (35). 5.

Find the range within which the individual must

vary when he gives a certain number of responses.

In the

example given above, this range is between a plus 50 and a minus 22. 4. scale.

This equals 52 units. Find the position of the individual along this

If the individual in the example cited had a

53 difference score of nine points found by Honesty Index II, he would now be placed at a position of 31 points up from the bottom of the scale: 5.

31 = 22 + 9 .

Express this scale position as a proportion of

the total range possible.

Example:

31 2c 100 = 39*6. 52

Multiplying by 100 is merely for convenience in handling the numbers. The means and standard deviations of the variables and criterion are given in Table IV.

The standard devia24 tions were computed by the raw score formula:

a— Correlations.

=

i/nzx1 - (zxf

Table V presents the correlations

found between the three Honesty Indexes and the factor Co. Table VI presents the correlations between the Honesty Indexes and supervisory ratings.

All of the correlations

reported are product-moment coefficients computed by the 23 formula:

r

-

N Z X Y - (ZXj(ZY) - / [ N 2 X 2-(ZX)1] [ N Z Y z - ( Z Y Jz]

The correlations found for Honesty Indexes I and II are considered to be distorted.

2>i Ibid., p. 57. Ibid., p. 204.

Those individuals receiving

34

TABLE IV MEANS AND STANDARD DEVIATIONS OF HONESTY INDEXES, FACTOR CO, AND CRITERION FOR 287 FOREMEN

M

S.D.

HONESTY INDEX I Key A

64.^7

13-34

Key B

47.26

15.32

Key A

12.15

3.98

Key B

11.40

3.40

Key A

70.50

8.10

Key B

80.37

8.82

60.63

15.92

3.02

1.00

HONESTY INDEX II

HONESTY INDEX III

FACTOR CO SUPERVISORY RATINGS

35

TABLE V PRODUCT-MOMENT CORRELATION COEFFICIENTS BETWEEN HONESTY INDEXES AND FACTOR CO. SIGNIFICANCE OF CORRELATIONS (N-287)

' FACTOR CO

SIGNIFICANCE

Key A

.401

.01

Key B

.347

.01

Key A

.226

.01

Key B

.404

.01

Key A

.134

•05

Key B

.263

.01

HONESTY INDEX I

HONESTY INDEX II

HONESTY INDEX III

36

TABLE VI PRODUCT-MOMENT CORRELATION COEFFICIENTS BETWEEN THE HONESTY INDEXES AND SUPERVISORY RATINGS. SIGNIFICANCE OF CORRELATIONS (N=28T)

SUPERVISORY RATINGS

SIGNIFICANCE

Key A

.126

.05

Key B

.012

HONESTY INDEX I

HONESTY INDEX II Key A

+. 000



Key B

+. 108



Key A

+ .069



Key B

+. 084

- -

HONESTY INDEX III

37 either a high or a low raw score on factor Co appear to be more honest on Index II than the average foreman.

For

Honesty Index II, a high difference score which represents greater falsification cannot occur if but a few keyed responses are given.

Therefore, individuals answering only

a relatively few items that appear on the standard key receive the distinction of being honest.

In a similar

manner, any individual giving all or nearly all of the keyed responses necessarily receives a low difference score because of the nearly even distribution existing between hard-toinfluence and easy-to-influence responses.

This individual

receives an honesty index characteristic of frankness that may not be deserved. Honesty Index I is subject to the same error that is discussed above.

An individual answering a large number of

the keyed items would have a hard-to-influence score approximately equal to his easy-to-influence score.

The

ratio would be near unity and would represent a validity that might not be warranted.

Those individuals answering

only a few items are not affected by this error. The scores assigned by the use of Honesty Index III are not limited by the number of keyed responses answered. Each score can vary from zero to 100 regardless of the number of items answered. Honesty Indexes I and II were considered to be too

58 dependent upon the number of keyed responses given by an individual to be of any more use in evaluating the present data.

The errors inherent in such indexes are only to foe

avoided in future work in this field. Attempts to increase prediction of the criterion. Mackie

26

using the same data as in the present study found

the factor Co and supervisory ratings to have a linear cor­ relation of .27 .

A correlation of this magnitude with an

H of 287 is classified as very significant.

However.,

because of the small predictability of such a relationship, the practical usefulness of such a relationship is limited. A measure of the factor Co with the variability due to falsification removed should correlate higher with the criterion than uncorrected cooperativeness scores.

A

corrected Co score was found by increasing the scores of those individuals who were honest and decreasing the scores of those who were dishonest.

The medians of the distribu­

tions of lie scores for keys A and B were found. for key A was 70.6, and for key B it was 80 .5 .

The median Individuals

receiving scores of higher than the median earned larger falsification indexes as measured by the two keys.

If our

operational definition of honesty is correct, the more

Mackie,

ojd.

cit . ,

p .

58 .

59 honest group should have received a higher cooperativeness score than they did while the cheaters should have received lower scores.

For scores below the median on the lie scale,

the honest scores, the difference between their score and the median was added to their Co score.

For those more

dishonest, those above the median of Index II, the difference was subtracted from their cooperativeness score. Thus, two distributions of corrected cooperativeness scores were obtained.

Table VII summarizes the data for these

distributions. By inspecting Table VII it is obvious that prediction of the criterion is lessened considerably.

The granting of

premiums to honest scores and the penalizing of falsified scores resulted in a decrease rather than an increase in predictability. effect.

There are two possible explanations of this

The first is that the weights given to the cooper­

ativeness scores added unreliable variance to the independ­ ent variable, thereby cutting down the correlation with the criterion.

The second explanation is that the controlling

of cheating in the independent variable reduced its correlation with the criterion to the extent that "plus getting” was accounting for the common variance between Co and supervisory ratings. The reliabilities for keys A and B, as reported earlier, are reasonably high.

Honesty Index III is the

40

TABLE VII MEANS, STANDARD DEVIATIONS, AND PEARSON r*s FOR CORRECTED COOPERATIVENESS SCORES

M

S.D.

r with rating

Signif­ icance

Co corrected by key A

61.2

16.79

.083

--

Co corrected by key B

66.4

16.24

.158

.05

41 consequence of a number of mathematical manipulations; therefore* the original reliabilities no longer apply to these derived scores.

Hew reliabilities were computed using 27 the split-half method. Index III values computed from pages one and three of keys A and B xvere compared with the values obtained from pages two and four. The reliability of Honesty Index III* when based on key A* equaled .14. .18.

The reliability based on key B equaled

Even allowing for the uneven distribution of the two

halves of the distribution* these reliabilities are extremely low.

The decrease in the prediction of the criterion

resulting when raw cooperativeness scores were corrected for cheating can be attributed to adding unreliable variance to the independent variable. One source of error present in Honesty Index III that lowers its reliability is the relatively small number of keyed responses answered by the average subject.

The

means and standard deviation of the hard-to-influence and easy-to-influence items are given in Table VIII.

A small

change in the manner of responding from one half of the test to the next would be reflected in an appreciable change in the respective Honesty Index III value.

Honesty Index

III is based on a proportion of the total possible difference

^

Supra* p. 28.

42

TABLE

V I I I

MEANS AND STANDARD DEVIATIONS OF EASY-TO-INFLUENCE AND HARD-TO-INFLUENCE CO RESPONSES

M

S.D.

Easy-to-influence

2^.88

5-55

Hard-to-influence

11.79

4.61

Easy-to-influence

16.92

3.89

Hard-to-influence

5.47

2.52

Key A

Key B

43 score obtainable for a given number of responses.

If the

number of responses is small, a slight change in the base will be reflected by a large deviation in the Index III value. From a practical viewpoint Honesty Index III is too unreliable a measure to be of any value.

Had Index III been

a reliable measure, its correlation with the criterion might have been increased to a point where it would be of value. To check this hypothesis, the correlations of Honesty Index III with the criterion were corrected for attenuation by the f o r m u l a : ^ r*CO CO----—----------------Y ------

vr** r Yr The corrected correlation coefficients were .18 for key A against the criterion and .20 for key B.

It is

recognized that correlation not significantly different from zero cannot legitimately be corrected in this manner.

These

increased correlations are of no more use in a practical situation than the uncorrected correlations.

From a

theoretical point of view, however, the corrected correla­ tions indicate that had the honesty index been reliable, there might have been the possibility of increasing predic­ tion by such a method. pO

Guilford, o p . cit., p. 287.

^4 The second hypothesis suggested as an explanation of the decreased predictability obtained when Co scores were corrected for falsification can be either verified or refuted.

This hypothesis was that the common variance

between Co and supervisory ratings was due to an error factor of ”plus getting” common to both variables.

The

unreliability of Honesty Index III obscured any relation­ ship that might exist. If there is no difference in the correlation obtained by comparing the number of hard-to-influence responses against the criterion when compared to the correlation of easy-to-influence items and the criterion, then the hypothesis has to be discarded.

If easy-to-influence items

correlate higher with the criterion, then the evidence would support the hypothesis of a ”plus getting” variance in the criterion.

Table IX summarizes the necessary

correlations. The significance of the differences between E responses and H responses on the same key were evaluated 29 by the method described by Peters and Van Voorhis. Hone of the differences were statistically significant.

The only

conclusion that can be drawn from the data is that there is

C. C. Peters and W. R. Van Voorhis, Statistical Procedures and Their Mathematical Bases (New York: McGraxtfHill Book Company, Inc. , 19*1-0), p. 155*

45

TABLE XX

PRODUCT-MOMENT CORRELATIONS FOR EASY-TO-INFLUENCE RESPONSES (E), HARD-TO-INFLUENCE RESPONSES (H) WITH CO AND THE CRITERION (N=287 ) CO

SIGNIFICANCE

CRITERION

SIGNIFICANCE

Key A . 1 9 8 ........... 01

E

.925

. .

.01

H

.888

. .

.01

E

.854

. .

.01

. 1 7 9 ........... 01

H

.757

• -

-01

. 1 5 3 ........... 05

.144

Key B

05

46 no difference in predictability when using easy-to-influence responses as compared to using hard-to-influence responses. The hypothesis of correlating common error has to be rejected.

II.

SUMMARY

In this chapter the data accumulated have been presented, analyzed, and discussed.

The analysis of the

data was limited to a complete investigation of the factor Co. Three lie detection indexes were employed in the investigation.

Two of these, the difference score and the

ratio score, had to be discarded.

The third index, the

range score, was analyzed as to its usefulness. The correlations of Honesty Index III with the criterion were not statistically significant.

Correlations

of Honesty Index III with raw scores on factor Co were significant at the .05 and .01 level for key A and key B respectively. When the degree of falsification was analyzed and frankness given a premium in the individual Co scores, the corrected Co correlations with the criterion dropped from .27 to .08 and .16 for key A and key B respectively. cause of this decrease was found to be due to the

The

4j unreliability of Honesty Index III.

The reliabilities were

.14 and .18 for keys A and B respectively. To check further on a hypothesis of "plus getting" accounting for the common variance between the criterion and the Co scores, the number of hard-to-influence and easy-to-influence responses given by each subject was correlated with the Co and supervisory ratings.

The

difference in predictability of the two scores was not statistically significant and the hypothesis was regarded as untenable.

CHAPTER V SUMMARY AND CONCLUSIONS In this chapter a summary of the study performed will be given, followed by those conclusions that can be drawn safely from the data. One hundred fifty-four college men were asked to participate in a psychological experiment.

They filled out

the Guilford-Martin Personnel Inventory under two conditions. Condition I was to falsify their answers in a manner which would make them look like a good job prospect for a position of industrial supervisor.

Three weeks after taking the test

under condition I they were asked to fill in the form a second time.

Under condition II the subjects were asked to

make themselves look poor as an industrial supervisor. An item analysis was run to determine the propor­ tionate change in the way each item was answered from condition I to condition II.

These changes were expressed

as critical ratios in order to make equal differences equal. Using distributions of the differences two lie keys, key A and key B, were constructed.

Key A used all of the

items on the standard Guilford-Martin test, assigning a “plus” to those whose change from condition I to condition II was greater than the median change, and assigning a "minus” to those items that changed less than the median

49 change.

These items were classified as easy-to-influence

and hard-to-influence respectively.

Key B differed from

key A only in that the middle third of the C R ’s was omitted* thus obtaining a key that weighted the extreme third only. It was felt that this t^ould eliminate those items that were not clearly defined as easy-to-influence or hard-toinfluence* thereby making the key more reliable. A group of 207 Industrial supervisors who had taken the Guilford-Martin test were used to check the usefulness of the lie keys.

The supervisors had been rated on a five

point scale as to proficiency. criterion.

This scale was used as the

The inventory booklets for these supervisors

were rescored using key A and key B.

The data from this

rescoring constitutes the major portion of this study. Reliabilities were found for hard-to-influence and easy-toinfluence items using the Spearman-Brown prophesy formula. These reliabilities were high enough to justify putting confidence in the keys. A series of honesty indexes were investigated and one was chosen that appeared to be unaffected by certain peculiarities of the data.

This index* Honesty Index III*

was used to evaluate the lie keys. Scores on this honesty index were shown to be signi­ ficantly correlated with the factor cooperativeness of the Guilford-Martin Personnel Inventory.

There was a small

50 There was a small but not significant correlation between the honesty index and the supervisory ratings.

The raw

score Co correlated .27 with the criterion. A correction factor was applied to the Co scores to allow for the degree of honesty with which the test was taken.

The corrected cooperativeness scores were then

correlated with the criterion.

These correlations were

markedly lower than that obtained before the honesty factor was partialed out. The decreased predictability was found to have been caused by the adding of an unreliable correction to factor Co.

The Co scores were then divided into a hard-to-influence

and easy-to-influence category for each supervisor.

The

hard-to-influence component was correlated with Co and the ratings. responses.

The same was done for the easy-to-influence The result of this analysis was that there is

no difference in predictability between the hard-to-influence responses and the easy-to-influence responses. The conclusion that can be drawn from this study can be summarized as follows: 1.

College students demonstrated their ability to

change their Guilford-Martin Personnel Inventory scores markedly when given instructions to do so. 2.

No independent and reliable measure of falsifi­

cation was found to evaluate the degree of cheating.

51 3.

From a practical viewpoint, no increase in the

predictability of the criterion was demonstrated. 4.

In spite of the Inconclusive results, the

methods used are suggestive of future research in the field of employee selection and placement. This study has pointed out many of the errors and problems that are encountered in doing research on the problem of correcting falsification on personality tests. These errors offer the opportunity for further research. Such research could: 1.

Use a sample of on-the-job employees to construct

a falsification key.

One assumption that had to be made

about the original population, in the present study, was that they could simulate the behavior of a good and poor industrial supervisor. 2.

Use job applicants as a population on which to

evaluate lie keys.

The foremen tested in the present study

were secure men of long standing at a manufacturing plant. Job applicants under competitive conditions would be likely to give more meaningful lie scores than were found in the present study.

B I B L I O G R A P H Y

52 BIBLIOGRAPHY Bernreuter, R. G., "Validity of the Personality Inventory,11 The Personnel Journal, 1 1 s585-86 ,- April, 1933. Goldstein, H., f,A Malingering Key for Mental Tests," Psychological Bulletin, 42:104-18, 19^5* Guilford, J. P., Fundamental Statistics in Psychology and Education. New York: McGraw-Hill Book Company, Inc., J53 PP. Guilford, J. P., and H. G. Martin, The Guilford-Martin Personnel Inventory, Manual of Directions and M o m s . Beverly Mills, California: SKeridan S u p p l y ‘Company, 19*3Hansen, Alice Viola, "An Experimental Study of the Ability to Simulate Humm-Wadsworth Temperament Performance." Unpublished Master*s thesis, The University of Southern California, Los Angeles, 1942. 44 pp. Humm, D. G., and G. ¥. Wadsworth, "The Humm-Wadsworth Temperament Scale," American Journal of Psychiatry, 92:163-200, 1935. Kelly, E. L., C. C. Miles, and C. M. Terman, "Ability to Influence Onefs Score on a Pencil-and-Paper Test of Personality," Character and Personality, 4:558-64, 1934. Kimber, J. A. M . , "The Insight of College Students into the Items of a Personality Test." Unpublished Doctor*s dissertation, The University of Southern California, 1945. 186 pp. Mackie, R. R., "Norms and Validity of Sixteen Test Variables for Predicting Success of Foremen." Unpublished Master*s thesis, The University of Southern California, Los Angeles, 1948. 58 pp. McKinley, J. C., S. R. Hathaway, and P. E. Meehl, "The MMPI: VI the K Scale," Journal of Consulting Psychology, 12:20-^1, 1948. Meehl, P. E., and S. R. Hathaway, "The K Factor as a Suppressor Variable in the MMPI," Journal of Applied Psychology, 50:525-64, 1946.

5? Peters., C. C., and W. R. Van Voorhis, Statistical Procedures and Their Mathematical Bases, New Y o r k : McGraw-Hill Book Company, Inc., 194UI 5l6 pp. Rosenzweig, S., ftA Basis for the Improvement of Personality Tests with Special Reference to the M-F Battery, Journal of Abnormal and Social Psychology, 33:476-88,

T95^ _______ , 11A Suggestion for Making Verbal Personality Tests More Valid,” Psychological Review, 4l:400, 1934. Ruch, F. L., "A Technique for Detecting Attempts to Fake Performance on the Self-Inventory Type of Personality Test,” Studies in Personality. New York: McGraw-Hill Book Company, Inc., 1942. Fp. 229-^4.

A P P E N D I X

5^

EASY-TO-INFLUENCE ITEMS KEY A AND KEY B* Item Number 8.

9. 10. *11.

*14. 16.

*17.

18.

*19. *20.

Yes

?

No

Do you feel that there are too many useless laws which hamper an individual’s personal freedom? . .

Co

Do you lack patience with the "shrinking violet" type of man?

Ag

.

Are you inclined to think about yourself much of the time? . . . .

0

Do you think that most people who help others secretly dislike going to the trouble to do s o ? ........

Co

When you are criticized does it disturb you b a d l y ? ........ .

0

Do you find that very few workmen nowadays do a job as it should be d o n e ? ...........................

Co

Does it bother you a lot to see someone else bungling a job that you know perfectly well how to m a n a g e ? ........................

Ag

Have you found that, in general, people higher up tend to dodge the dirty work, leaving it for others to do? . ..............

Co

Are you annoyed when people tell you how you should do a thing? . .

Ag

Do you get upset rather easily?

0

.

* All of the items appear on Key A; those with an asterisk also appear on Key B.

55 EASY-TO-INFLUENCE ITEMS (cont inued) Item

Number *21.

24.

Yes

?

No

Are you ever bothered by the Idea that someone is reading your thoughts?..................

0

Are you inclined to worry too long over humiliating experiences? .................. .

0

2 5 . Do you think that the kind of person who would "turn the other cheek" deserves to get slapped? .

Ag

*2 7 . Do you think that large business corporations should be prohibited?

Co

*28.

*37.

Do other people often try to take the credit for things you yourself have accomplished? . . .

0

Does your conversation tend to center around your own interests and hobbies rather than those of other people? ................

0

*3 9 . Are you often getting into scrapes which you did not seek to stir up? 40.

*41* *43*

*46.

0

Have certain people talked about you and yet you were unable to prove i t ? .........

Ag

Do you sometimes think that most people are stupid?..............

Ag

Co you know of any people at present who are intentionally trying to avoid you? . . . . . . . Do you believe in the parole system for p r i s o n e r s ? ..........

0

Co

56

EASY-TO-INFLUENCE ITEMS (continued) Item

Number *48.

*50. 51.

*53.

*5 6 . *5 8 .

60.

*62.

65.

66.

Yes

?

No

Have you ever felt that someone was hypnotizing you and forcing you to do things you did not .................. want to do?

0

Do other people deliberately say or do things to annoy you? . . . .

0

When you enter a new group (business or social) do you like to be tipped off as to who are the important people to line up with? .

Co

Do you believe that only people with money can be sure of getting a square deal in courts of law? .

Co

Do other people often watch you on the s l y ?

Co

.

Do you think that an unusually bright person is likely to be physically weak? ................

Co

Were you ever ignored or given a "raw deal” through spite? . . .

Co

Do you suspect that most people who do you a good turn are really expecting something in return for i t ? ........................

Co

Do you believe that most people shirk their duties whenever they can without appearing to do so? .

Co

Do you hate to lose an argument even when the issue is of little i m p o r t a n c e ? ....................

Ag

57 EASY-TO-INFLUENCE ITEMS (continued) Item Number

Yes

?

No

*6 7 . Does It seem to you that other people generally "have all the l u c k " ? ...............

Co

70. Are most people thoughtless of the rights of o t h e r s ? ...........

Co

*73. Generally speaking, do you believe that the boy who has not learned to defend himself deserves to "take a b e a t i n g " ? ...............

Co

7 6 . Do you consider yourself a rather nervous person? .................

*77. Do you think your generation has as many opportunities for success as your parents 1 generation had?

0

Co

7 8 . Do other people pay more attention to your comings and goings than they s h o u l d ? ..................... 8 l.

Do you usually receive criticism of .yourself without resenting it?

Co 0

*8 3 . Is there anyone you know personally whom you would like to see behind prison bars? . . . .

Ag

8 5 . Do you think that most people are

overpaid for what they really contribute to society? ...........

Co

*8 7 . Have you very much resented having friends or members of your family give you o r d e r s ? .................

Ag

8 8 . Are you as quick as other people

to pay compliments when they are d e s e r v e d ? ..................

0

58

EASY-TO-INFLUENCE ITEMS (continued) Item Number

Yes

*8 9 . Do people frequently talk about you behind your b a c k ? ..........

?

No Co

9 6 . Should the government take over

more and more the management of private business? .............. 97. Are your feelings rather easily h u r t ? ..........................

Co 0

*99 . Do you usually get more than your share of the blame when things go wrong? . . . . ..................

Co

*100. Are there many kinds of work that you would not consider doing because they are beneath you? . .

0

*103. Have you often felt that certain persons are secretly trying to get the better of you? ........

Co

*104. Would you prefer a hard job that is very interesting to an easy one that is uninteresting? . . . .

0

*105. Have you frequently wished for enough money or power to impress people who regard you as an i n f e r i o r ? ......................

Ag

108.

If you could have your way about it would you change a lot of things about human nature? . . . .

Co

*109.

Are many of your supposed friends really insincere? ..............

0 Co

110.

Do many people think you are "hard-boiled"? ..................

Ag

59 EASY-TO-INFLUENCE ITEMS (continued)

Item Number 112.

*113. *114. 115.

116. *117.

*118,

*120.

122. 124.

Yes

?

No

Do some people become so bossy or domineering that you want to do the opposite of everything they tell you to do? ..........

Ag

Do other people often deliberately make things hard for you? . . . .

Co

Are there times when it seems that everyone is against you? . . . . . .

0

0

When criminals make a daring escape from prison do you sometimes secretly hope that they will avoid capture?..................

0

Do you sometimes feel contempt for the opinions of o t h e r s ? ........

Ag

Does it bother you to have other people tell you what you should d o ? ............ ......... .

Ag

Do you often feel that you are left out of things, perhaps unintentionally, in group activities? ....................

0

Does the United States government owe every one of its citizens a decent living? ..................

Co

Have some people criticized you unjustly to others? ............

0

Are you deserving of things far better than is your present lot? .

Co

*125 . Do you think no one would keep on the "straight and narrow path" were it not for the fear of being c a u g h t ? ..................

Co

60 EASY TO INFLUENCE ITEMS (continued) Item

Number When you lose something do you often begin to suspect someone of either having taken it or having misplaced i t ? .

126.

Yes

*129.

?

0

Were you ever seriously doublecrossed? ........................

0

135. Were you ever so deeply in debt that you did not know where the money was coming from to pay what you o w e d ? .................. *137 . Do other people often blame you for things unfairly?. 138 .

In most cases is it important to get what you want even if you have to come into conflict with other people in order to get it? .

* 139 . Have other people been too ready to accept credit which rightfully belongs to you? . . . . . . . . . *141. *142.

*143.

*144.

*147.

No

Co 0 Co

Ag

Co

Have you had more than your share of hard l u c k ? ..................

Co

Do other people often try to put things over on you when you are not w a t c h i n g ? .................

Co

Is there any person whom you would particularly like to "put in his (or her) place”? ................

Ag

Do you believe that most people require someone to tell them what to d o ? ........................ Do you often find it necessary to return merchandise to a store because it turns out to be not as represented?

Ag

Co

EASY-TO-INFLUENCE ITEMS (continued)

Item Number *148.

Yes Do people near you sometimes whisper or look knowingly at one another when they think you are not noticing them? . . .

62

HARD-TO-INFLUENCE ITEMS KEY A AND KEY B* Item Number

Yes

?

*7 . Does It seem to you that human beings hardly ever learn to avoid making the same mistakes t w i c e ? .......................... *8 . Do you feel that there are too many useless laws which hamper an individuals personal freedom? . . *11.

Do you think that most people who help others secretly dislike going to the trouble to do s o ? ........

Co

Co

Co

*12. Do you feel disgusted when someone escapes paying the full penalty of the law through some technicality? *13. Do you frequently seek the advice of other people? .............. *15.

Do you despise a "yes" man?

No

Ag 0

...

0

Ag

16. Do you find that very few workmen nowadays do a job as it should be d o n e ? ..........................

Co

*1 7 . Does it bother you a lot to see someone else bungling a job that you know perfectly well how to manage ? ........................

Ag

*22.

Do you have days in which it seems that everything goes wrong? . . .

* All of the items appear on Key A; those with asterisks also appear on Key B.

0

65 HARD-TO-INFLUENCE ITEMS (continued) Item

Number *2 3 .

*26.

Yes

31.

Co

Have you ever had fears about other people that you later found to be without foundation? ............

0

When a person has gone out of his way to be nice to you, do you try to see what his real reasons are? Do you usually feel that in group undertakings your own plans are b e s t ? .....................

*3 2 . Do you think that in most places the traffic regulations are seriously in need of improvement? 33•

*3^.

35. 36.

*33.

No

Generally speaking, do you think the head of a firm should have risen through the ranks, that is, having worked his way up in the b u s i n e s s ? ......................

*2 9 . At a movie or a play do you often feel that one of the main characters is a bit like you? . . 30.

?

0

0

Ag

Co

Is money necessary for complete happiness?......................

Ag

Do you think that all secret societies should be done away w i t h ? ..........................

Co

Do many men deserve higher pay than their b o s s e s ?..............

Co

Have you ever been severely punished for something you didnrt do? . . .

Co

Do you sometimes feel sorry for a person who Is convicted of a crime even though you realize he is guilty?

0

64 HARD-TO-INFLUENCE ITEMS (continued) Item Number

Yes

?

No

Co

Co

*4 5 . Do you think that most people live far beyond their income?

. .

*4 7 . Do people sometimes offend you without knowing it because you hide your feelings from t h e m ? .................. *4 9 . Are you inclined to let other people have their own way even when it disturbs your peace and comfort?........................

0

Ag

5 2 . Do you enjoy taking part in a good fight? ...............

Ag

*54 . Do you believe that most people will tell a lie now and then in order to get a h e a d ? ....

Co

5 5 . Are there some things about

yourself concerning which you are rather touchy? .........................

0

*57. On the whole, are your own ideas of how things should be done superior to the plans suggested by others?

Ag

*59 * Do you often feel very badly about about other people's troubles? . .

0

*61. Do most groups of people behave like a bunch of sheep, that is, blindly follow a leader? ........

Co

6 3 . Do you often feel that a lecturer is talking about you

personally? *64.

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

If a person is not playing fair, do you like to see someone beat him at his own g a m e ? ............... .

0

Ag

Ag

65 HARD-TO-INFLUENCE ITEMS (continued) Item

Number 68.

Yes

?

Does it distress you considerably

to see another person in pain? . . 6 9.

*71. 72.

*74.

*7 5 .

*77.

*79.

*8 0 .

*82.

*84.

No 0

Do you feel that many young people .get ahead today because they have "pull"?....................

Co

Are you continually comparing yourself with other people? . . .

0

Do a lot of people you have known tend to form "cliques" or closed g r o u p s ? ........................

Co

Do you find that generally if you want a thing done right you must do it y o u r s e l f ? ................. Can a person get ahead by his own efforts if he does not look out for himself at every turn? . . . . Do you think your generation has as many opportunities for success as your parents r generation had? .

Co

Co

Co

Co

When things become dull do you feel the urge to stir up some e x c i t e m e n t ? ....................

Ag

Do you think the educational system in this country is seriously wrong in many r e s p e c t s ? ..............

Co

Have you ever found out that a

person who was supposed to be an expert did not know as much as you did about something?............

Co

Are there important changes you would make immediately in your manner of living if you could have a sub­ stantial increase in income? . . .

Co

66

HABD-TO-INFLUENCE ITEMS (continued)

Item Number * 86.

*90.

*91.

92. *93. 94.

95.

98.

Yes

No

In most situations is it all right to conceal the truth when by doing so you give aid to a friend? ........................

0

Is it almost unbearable for you to see a close friend or relative suffering intense pain? . . . . .

0

Is It true that people will generally have contempt for a person who does not assert himself once in a while?

0

Have you known many fftwo-faced" individuals personally?

Co

Can most people be trusted completely? Co Do you derive considerable satisfaction from making other people do as you want them to? . .

Ag

Do you believe that all public office-holders sooner or later look out for their own Interests first? ..........................

Co

Generally speaking, do you believe that people use a veneer of politeness to cover up what is usually "cut-throat" competition?

Co

*101 . Do you think that a lot of people exaggerate their hard luck in order to gain sympahty from others? . .

Co

102 . Do people ever accuse you of being *1 0 6 .

selfish, and with some reason? . .

Ag

Have you frequently become involved in conflicts in defense of your friends or members of your family? . . . .

Ag

67 HARD-TO-INFLUENCE ITEMS (continued) Item Humber

107.

*111.

119.

*121.

Yes

Is it difficult for you not to bear a grudge against someone who has injured y o u ? ...

?

Ho

Ag

Does everything that happens seem to have a relationship to your own life or experience? . . .

0

Is it difficult for anyone to * impose upon you for the reason that you are usually wise to their i n t e n t i o n s ? .....

Ag

Is the person who carelessly

leaves valuable properly lying around as much to blame as the person who appropriates it for his ownuse? .

Co

123 . Will most people in business bear

close watching when you are dealing with t h e m ? ..... *127.

128,

*130.

131.

Co

Do you think that most people who allow themselves to come under the rule of a dictator are to blame for their p l i g h t ? ........

Co

Is there any subject on which you would like to hold a public indignation meeting for the purpose of organizing a mass protest?........................

Co

Do you believe the parole system works more to the advantage of the prisoner than to the advantage of s o c i e t y ? ...............

Co

Do you tend to let people run over you more than you should for your own g o o d ? ......................

Ag

0

68

HARD-TO-IDJFLUENCE ITEMS (continued) Item

Humber *131.

Yes Do you tend to let people run over you more than you should for your own good? . ..........

*134.

*136 .

*140. *144.

*145.

*146. *149.

Do you usually seek to become an officer in any organization to which you b e l o n g ? ....

No

Ag

Ag

Ag

*132 . Have you frequently felt like telling "nosey” people to mind their own b u s i n e s s ? ............ *133, Have you often found it necessary to stand up for what you believe to be r i g h t ? ..........

?

Ag

Ag

In group undertakings do you usually manage to have your own plans put into effect?. . . .

Ag

Do you sometimes feel sorry for all the people in the world? . . .

0

Do you believe that most people require someone to tell them what to d o ? ..........

Ag

Is there any piece of music that usually makes you weep when you hear i t ? ..............

0

Are people in general out to get more than they g i v e ? ..

Co

If someone rudely crowds ahead of you in line, do you tell him

where he b e l o n g s ? ....

Ag

69 TABLE X PROPORTION OF COLLEGE STUDENTS ANSWERING YES, NO, AND ? TO EACH ITEM OF THE GUILFORD-MART IN PERSONNEL INVENTORY UNDER EXPERIMENTAL CONDITIONS I AND II. CRITICAL RATIOS OF THE DIFFERENCES (N=154)

Item Num­ ber 1 2 3 4 5

Condition I Look good ?

Yes

No

94 94 4 34

6

0 1

24 9 73 51 30

73 9

62 56 33

16

9 10

19 27

11 12 13 14 15

10 67

56 11 66

5 93

5

44 76 92

55 23

8

1 1 1

5 4

89

11 9

2.07 2.62 8.21 7.52 19.7 18.8

0 0

16.7 14.5

1 5 1 3 3

90 48 28 82 42

9 51 71 17

1 1 1 1 2

22.6 22.6 3.36 4.17 5.07 5.07 17.25 16.30 4.24 4.47

0.00 1.90 0.00 1.14 O .52

5 3 2 0 0

81 62 91 93 82

18

1 0 1 0 2

9.74 8.80 4.60 5.38 11.6 11.5 26.3 26.3

1.90 2.00 0.67 0.00 1.67

1 2 5 1 12

82

18

86 51

14 46 24 12

82 76 69

2

89 28 43

37 9 5

61 91

21 22 23t, 24 25

6 49 68 11 24

93 49 26 88

a

65

4 4

19 20

95

65

•>

48

3 3

40 63

63 12

85

Ho

17.4 16.0 27.3 28.1 17.0 16.35 3.01 2.62 5.80 6.88

10

86 31

?

0 2 1

63 28

32

17 18

Yes

No

6

16

Critical Ratios

Yes

7

8

Condition II Look poor

91

87

56 38 8 7 16

3

0 0 3 1 1

20.7 20.2

14.9 13.4

22.3

19.3 7.38 6.98 3.04 3.66 14.4 14.2 14.0 11.2

2.00 1.11

1.56

0.52 1.14

1.60 1.56 0.67

2.63 2.35

1.11

1.67 O .85 * 0.00 3.86

70

TABLE X (continued) PROPORTION OF COLLEGE STUDENTS ANSWERING YES, NO, AND ? TO EACH ITEM OF THE GUILFORD-MARTIN PERSONAL INVENTORY UNDER EXPERIMENTAL CONDITIONS I AND II. CRITICAL RATIOS OF THE DIFFERENCES (N=l54) Condition II

Condition I Look good Item Num­ ber 26

27 28 29

30

3

5 4

53 70 94 71

27

65

0

85

15

.56 43 59 77

8

82

5 5 5 7

71 92

18 26 8

49 79

21

0

10.2

3 3 3

12 8 66 10

2

3

8.89 17.8

1 10

77 90 31 90 94

3

95

5 39

54 3

43 95 75 53

36 52 36

31 32 33 34 35

18 28

36

31

66

19

79 31 90 71

31 38

39 40

18 66 8

66

41 42 43 44 43

16 18 6

55 37

93 43 57

46 47 1n 48 49 50

91 58 4 41 4

37 95 53

81 80

6

92

2

No

45

28

5

48

5

?

Yes

Yes

. No

43 35

Look poor ?

Yes

21

Critical Ratios

3

2 1 2 0 0

3 1

3

3 1 1

1 2 1 0

2 1 2 6

86

84

11

5

3 5

18 58 75

80

2 1 1 2 1

1 6

4

26 92

12

5

40 25 72 8

.172 16.4 18.5 5.00 10.2

8.92 3.44 12.2

5.97

6.42 24.1 19.4 22.3 7.88

22.7 8.85

9.40 18.9 0

19.8 2.49 30.8

No

.345 16.1

17.3 4.71 10.2

7.30 3.11

10.8

5.38 8.71

9.20 17.3

6.42

22.7

15.9 22.6 7.90 23.7

8.41 9.48 19.2 .531 17.2 3.44 25.9

? 0.00 0.00 1.90 0.96 0.00

3.48 .83 1.90 6.83 3.34 0.00 0.52 0.00 0.00 3.38

1.14 0.00 0.00 1.67

.37 0.52 1.90 0.00 1.72 1.58

71 TABLE X (continued)

PROPORTION OP COLLEGE STUDENTS ANSWERING YES, NO, AND ? TO EACH ITEM OF THE GUILFORD-MARTIN PERSONAL INVENTORY UNDER EXPERIMENTAL CONDITIONS I AND II. CRITICAL RATIOS OF THE DIFFERENCES (N=154) Condition I Look good Item Num­ ber

Yes

No

?

Yes

No

?

Yes

No

?

28

66 68

6 8 2

89 79 91 94 87

11 18

0

11.7

9 13

0 0 0

13.2 11.3

28.1 4.22 10.0

3.00 1.82 1.67 2.00 1.67

89

10

1

13

l l l

26.1

86 82

5 7

17 93

2

93 95 71

24 5 77 38

60

2

56 57

5 45

82

13 13

6

59

93 19

43 93 42

3

§0

26

66

61 62

f? 64

59 19 21 86

78 76

65

32

66 67 68 69

32 13 68

26

6

70

26

43 71

71

36 49 13 38 45

69 S°

72

73 74 75

Critical Ratios

Look poor

51 52 53 54 55

58

Condition II

12 66

87 84

84 60

48

3 3 3 3 1

68

94

6

16 82

7 6

4 27 28 6

4 3

95 93 19 95 90

79 5 9

1 1

63 93

3

88 87

34 7

3

1 8

17

5

6

12 11 81

3

8.17

1 0

20.2 7.00 16.1

1 1

20.0

7.53

0

9.96 3.77 14.4

0 1

22.9

3 4

3

0 1

3 0 0 1 1

14.8 9.87 9.07 14.6 4.82 9.55 19.3 10.0

5.44

10.1 2 6 .4

3.37 9.63 17.7 6.05 2 l.o 7.70 13.3

4.08 4.08 0.00 1.90 3.33

7.39 19.5 9.63 3.50 !3.7

2.00

0.67

1.14 0.00

0.44

14.3

1.11

21.6 10.7

1.20

8.41 14.1

2.35 1.14

5.43 9.33 17.8

10.3 6.34

1.14

1.14 1.11 2.00 0.00

2.83

72 TABLE X (continued) PROPORTION OF COLLEGE STUDENTS ANSWERING YES, NO, AND ? TO EACH ITEM OF THE GUILFORD-MAR TIN PERSONAL INVENTORY UNDER EXPERIMENTAL CONDITIONS I AND II. CRITICAL RATIOS OF THE DIFFERENCES (N-154) Condition I Look good

Condition II

Critical Ratios

Look poor

Item Num-

Yes

No

0

Yes

No

64 14

35 84 16 20

79

19

12.2 16.3 17.3 4.50 7.44

12.2

83 78

1 1 1 1 2

3 4

10

90

16.9 5.18

1

89

10 11

82

16

23.4 8.43

22.6

3 5

0 0 0 0 2

19.0



9

72 94

26 6

ber

Yes

No

76 77 78 79

8

83 12

91 15

82

1 2 6

54 40

39

7 4

80 81 82

§3 84 85 86 87 88 89

81 66 8

51

16

34 12

90 3

56 16

30 91 46 80 SI

86 8

2 2

8

28

5

70 92

5

84 93

90

63

28

91 92 93 94 95

69 42 37

26 56

38

97

2 8

98

43

96

91

S 7

65 56

99

6

100

96 92 50 91

19

77

2 6 6 1 0 6

3 3

9

8 66

2 0 1 2 6

7.04 24.4 13.7

28 8 88 16 6

2 0 1 0 1

.186 10.6

30

1 0 1 0 0

70 89

10

69

72 94 92 93

15.0

28 6 8 6

28.7

6.42

5.72 10.9 12.2 16.2

14.7 11.2 28.2 19.8

16.3 14.8 3.66

7.04 4.39 7.79 14.4 5.69

22.2 13.9 17.1 7.05

.388 10.2 6.32

9.97 11.0

15.9 14.7 9.66 25.0 17.6

0.00 0.67 2.29 2.63 0.96 2.00

2.35 1.11 2.00

1.33 2.61 1.67 0.67

4.84 0.66

1.33 1.67 2.29 2.63 2.29 0.00 0.00 2.29 2.00 2.00

73 TABLE X (continued) PROPORTION OP COLLEGE STUDENTS ANSWERING YES, NO, AND ? TO EACH ITEM OF THE GUILFORD-MAR TIN PERSONAL INVENTORY UNDER EXPERIMENTAL CONDITIONS I AND II. CRITICAL RATIOS OF THE DIFFERENCES (N=l54) Condition I

Condition II

Look good Item Num-

Look poor

ber

Yes

No

?

Yes

101

71 17 17 96 24

27 80 82

2

3

98 60

3

93 73 95

106

34

107

28

108

25

65 68 68 82 82

1 5 8 8 6

5 5

102

103 104 105

3 73

1 1

48 82

No

?

Yes

2

0 2 0 1 0

6.95 8.50 20.3 25.3

1 0 1 0

2.55

38 7 86

5 51

18

7

110

10 12

111 112

40

56

26

113 114

2 1

115

4 9

69 94 90

16

81

3

64

5 3

95 96 92 90

10

109

116

30

117

10

118

84 79

16 18

69

29

95 97 95 75

4 3 5 23

4 4 7

3 2 1 0 0 2

119

9 39

120

10

31

2 10

86

4

81

17

1 0 1 1 2

121 122

58

15 51

38 67 41

4 17

24 12

60 88

74 94 95

17

25 5 4 3 13

1 1 1 1 1

123 124 125

§7 89

Critical Ratios

8 1

86

17.8

No 6.43 8.11 19.6

25.5 16.4 2.55

11.0

10.0

15.7 18.9 15.2

14.0 19.5 14.2

5.25

4.88

17.0

15.6

41.0 27.4

36.4

12.6 15.3 28.5 25.0 10.8

17.1 2.97 21.9

9.75

27.8 12.1

9 1.67 0.52 1.11 0.00 2.00 0.00 2.78 2.83

3.48 1.20

1.33 1.90 1.67 1.11 0.52

13.8

1.90

25.3 24.5 8.55

2.00 0.67

16.2

2.42

14.4 8.23

3.39 0.96 1.56

4.95 2.83

17.8

13.4

4.49

18.7

19.2

0.00

74 TABLE X (continued) PROPORTION OF COLLEGE STUDENTS ANSWERING YES, NO, AND ? TO EACH ITEM OF THE GUILFORD-MARTIN PERSONAL INVENTORY UNDER EXPERIMENTAL CONDITIONS I AND II. CRITICAL RATIOS OF THE DIFFERENCES (N=154) Condition I

Condition II

Look good

Look poor

Item Num-

No

?

Yes

5

2 2 1 0 1

17.2

8.81

24.6 7.50

1 0 1 0 1

8.54 7.50 1.68 4.53 19.8

7.94 7.50 1.92 5.14 19.2

•• -52

2.12 25.8 13.8 21.6 0

2.29 2.00

24.1

24.1 20.5

2.29

16.5 5.12

16.0 4.00

3

4.91

4.80

3.00 0.00

0 1 1 1

8.50

7.29 16.5

3.00 0.00

24.1 3.32

1.90

4.30 2.20

1.86

1.14

ber

Yes

No

?

Yes

126 127 128 129 130

23 51 25

7 4 4

82 81

35

70 45 71 93 57

9

80

131 132 133 134 135

10 62 83 72

88 38 16 25

2 1 1

4

94

2

51 95 75 47 79

20

136 137 138 139

47 4

35

18

50 92

47

22

74

13 23

82

4 5 6

74

87 95

3

26

5

90 90 94 84 37

140

6

92

141 142 143 144 145

5 8 27

58 13

71 38 84

146 147 148 149

59 14 5

83 90

60

150

93

B 86

35 37 5

Critical Ratios

1

3

6 2 6

3 6 1

5 4 1

93

89

96 83 92

82 85

16 18 11 18

49 5 25 53

8 12

4 74 8

9 6 16 60

4 1§

8 18 11

2 1 1 1 0 2 1 0 0

3

5.96 11.5

30.8

14.8 24.3 .60 27.1

16.3

29.2

Wo 15.3 5.70 10.8

•>

2.07 0.96 1.56 1.11

3.12 0.67 1.11 0.00 2.00

0.67 4.68 1.56 1.99

1.33

1.67

1.56