2.3 .
3?},

 

aw umnmbukm
. Jim man?

5: . $1 .
25.... I.
I.» .

3.: 5!!
a: ..I::);.
I!

 

‘2;
1 a... 7.!
‘1: but...)

 

Mquugufnu-T-u 'w v

I

. "won-9'!”

    

  

épﬁgévfziﬁ . . , .

 

.

  

 

      

gﬁgsmsiﬁéﬁ

[II

E. E.

‘ . .3125

 

JN__xmmu

  

 

 

 

 

 

THESIS
7

4'-
’-" OC F
.g,’ . ‘-_.

 

 

IHIHJHIHUHJHIHUHIIHIWHIMllJlllllllllllmllll

193 02048 6530

 

LIBRARY

Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

TOWARD AN INTEGRATED MODEL OF
APPLICANT FAKING

presented by

Lynn A. McFarland

has been accepted towards fulﬁllment
of the requirements for

Ph . D . degree in W

 

 

Date l'7L/ 3‘ 5/00

 

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0. 12771

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

EAJEDDUE DATE DUE “DATE DUE

[In

 

‘wvv1

DEC 1 9 zbua

 

 

 

 

 

 

 

 

 

 

 

 

 

 

moo c/CIHCJDﬂDtIJ55-p.“

 

TOWARD AN INTEGRATED MODEL OF APPLICANT FAKING
By

Lynn A. McFarland

A DISSERTATION
Submitted to
Michigan State University

In partial fulﬁllment of the requirements
For degree of

DOCTOR OF PHILOSOPHY
Department of Psychology

2000

.1“:

 

ABSTRACT
TOWARD AN INTEGRATED MODEL OF APPLICANT FAKIN G
Bv

J

Lynn A. McFarland

The present study used the theory of planned behavior to predict faking on a personality
selection test. Participants’ attitudes toward faking, subjective norms toward faking,
perceived behavioral control over faking, and intentions to fake a selection test were all
assessed. In addition, participants were asked to take the personality measure under
instructions to be honest and also under instructions to respond like an applicant. Their
faking behavior was assessed through the use of difference scores and a social
desirability scale. The theory of planned behavior predicted both the intention to fake the
selection test and actual faking behavior. However, proposed moderators of these
relationships, such as valence toward performing well on the test, warning of a lie scale,
and knowledge of the constructs assessed by the personality test, were generally not
supported. Practical implications of these results and directions for future research are

discussed.

o'

i-‘

n'“
~..L

-.

C.

SI

ACKNOWLEDGMENTS

I want to everyone who made this dissertation possible, both directly by providing
feedback and advice on the dissertation, and indirectly for providing support and
encouragement to me personally throughout the process. First, thanks to Ann Marie
Ryan, my chair, for providing guidance and support throughout the dissertation process.

I also want to thank my committee members, Drs. Neal Schmitt, Rick DeShon, Dan
Ilgen, and Murray Barrick for their insightful comments and recommendations.

A thank you goes to Melissa Kabin for helping me collect my data and to
Christine Scheu for helping me to get those tasks done that were so difﬁcult to do from
Maryland. I also thank Christine for her general support and friendship along the way.

I want to thank my family, especially my parents, William and Rosanne
McFarland. They have always provided me with love and encouragement and have been
there for me in the toughest of times cheering me on. Without their constant support I
would not have been able to do any of this. I also want to thank my parents for taking the
time to learn what industrial/organizational psychology is. This is a true demonstration
of their love, concern, and support, and I am sincerely touched.

Finally, most of all, I want to thank my husband, Robert Ployhart, for his advice,
support, friendship, and love. He read this dissertation almost as many times as I did and

I greatly appreciate it!

iii

TABLE OF CONTENTS

LIST OF TABLES .............................................................................................................. v
LIST OF FIGURES ............................................................................................................ vi
INTRODUCTION ............................................................................................................... 1
Deﬁnitions and Why People Fake ............................................................................... 3
Evidence that Applicants Can and Do Fake ................................................................ 5
Inﬂuence of Faking on the Results of Personality Measures .................................... 10

A Model of Faking .................................................................................................... 30
Method .......................................................................................................................... 47
Sample ....................................................................................................................... 47
Design ........................................................................................................................ 47
Procedure ................................................................................................................... 56
Results ........................................................................................................................... 59
Descriptive Analyses ................................................................................................. S9
Manipulation Checks ................................................................................................. 64
Preliminary Analyses ................................................................................................ 75
Hypothesis Tests ....................................................................................................... 77
Exploratory Analyses .............................................................................................. 102
Summary ................................................................................................................. 103
Discussion ................................................................................................................... 104
Inﬂuences on Intention to Fake ............................................................................... 107
Predictors of Faking Behavior ................................................................................. 110
Effect of Faking on Criterion-Related Validity ....................................................... 116
Implications and Directions for Future Research .................................................... 117
Limitations .............................................................................................................. 121
Conclusion ............................................................................................................... 122
REFERENCES ................................................................................................................ 124
APPENDDC A: Pre-Questionnaire ................................................................................. 133
APPENDIX B: Post-Questionnaire ................................................................................ 136
APPENDD( C: Demographic Questionnaire ................................................................. 138
APPENDDi D: Consent Form ....................................................................................... 139
APPENDD( E: Protocol ................................................................................................. 140
APPENDIX F: Debrieﬁng Form .................................................................................... 146

iv

at";

15*-
Us.»

In.
B\,_

 

 

LIST OF TABLES

Table 1. Demographic Characteristics of Study Participants ........................................... 47
Table 2. Reliabilities of Personality Scales Across the Honest and Applicant Conditions,
and Reliabilities of Difference Scores ................................................................ 51
Table 3. Correlations Among Demographic Variables and Measures. ............................ 60
Table 4. Analysis of Variance Across the Eight Experimental Conditions ..................... 67
Table 5. Means and SDs Across the Eight Between-Subject Conditions ........................ 69

Table 6. Paired Samples t-tests Comparing Means Across the Honest and Applicant
Condition ............................................................................................................ 71

Table 7. Regression Analyses Regressing Pre- and Post-Intentions on Attitudes,
Subjective Norms, and Perceived Behavioral Control. ...................................... 72

Table 8. Post Intentions Regressed onto Attitude Toward Faking, Subjective Norms,
Perceived Behavioral Control, Warning of a lie scale, and Valence. ................ 73

Table 9. Regression Analyses Regressing Post-Intentions on Attitudes, Subjective
Norms, Perceived Behavioral Control, and Valence .......................................... 84

Table 10. Regression Analyses with Post Intentions Regressed on Attitudes, Subjective
Norms, Perceived Behavioral Control, and Warning of a Lie Scale .................. 85

Table 11. Regression Analyses Regressing Pre-Intentions and Knowledge on Faking

Behavior ............................................................................................................. 86
Table 12. Regression Analyses Regressing Post-Intentions and Knowledge on Faking
Behavior ............................................................................................................. 88
Table 13. Regression Analyses Regressing Faking Behavior on Pre-Intentions to Fake
and the Knowledge Scale ................................................................................... 90
Table 14. Regression Analyses Regressing Faking Behavior on Post-Intentions to Fake
and Knowledge ................................................................................................... 92
Table 15. ANOVA Predicting Faking Behavior. ............................................................. 94

LIST OF FIGURES

Figure 1. Model of Faking Behavior. ............................................................................... 31
Figure 2. Interaction Between Warning and Valence When Predicting Faking as
Measured by the Openness Difference Score ..................................................... 96
Figure 3. Interaction Between Warning and Knowledge on Faking as Operationalized as
the Openness Difference Score. ......................................................................... 97
Figure 4. Interaction Between Warning and Knowledge on Faking as Operationalized by
the Conscientiousness Difference Score. ........................................................... 98
Figure 5. Supported Relationships. ................................................................................ 105

vi

it?»

55.1:
me;
O .
mew

Cut?

V1521

 

 

 

 

INTRODUCTION

Personality tests are increasingly being used for personnel selection because these
tests generally yield moderate validities (Barrick and Mount, 1991; Hough, Eaton,
Dunnette, Kamp, & McCloy, 1990; Tett, Jackson, & Rothstein, 1991), yet produce less
adverse impact than more traditional selection measures (e. g., cognitive ability tests;
Bobko, Roth, & Potosky, 1999). Despite this, there is concern that intentional distortion
on the part of applicants may make the results of these tests difﬁcult to interpret because
it may result in changes in the test mean, reliability, and validity. However, there is still
debate about the effects of faking on the results of personality tests. For example, some
researchers have found evidence that faking does not negatively inﬂuence the validity of
these measures (Abrahams, Neumann, & Githens, 1971; Barrick & Mount, 1996;
Cunningham, Wong, Barbee, 1994; Hough et al., 1990; McCrae & Costa, 1983; Ones,
Viswesvaran, & Reiss, 1996). However, others disagree with these ﬁndings and provide
their own evidence that faking does attenuate the validity of non-cognitive measures
(Douglas, McDaniel, & Snell, 1996; Dunnette, McCartney, Carlson, & Kirchner, 1962;
Kluger, Reilly, & Russell, 1991; Pannone, 1984).

There are several reasons that results of research on faking personality and other
self-report measures may be inconsistent. An explanation of some of the observed
differences between these ﬁndings may be that faking is Operationalized in different ways
across studies. For example, some researchers have used scores on a lie scale as a
measure of faking (e.g., McCrae & Costa, 1983; Hough et al., 1990; Rosse, Stecher,

Miller, & Levine, 1998), whereas others have used deviations from the mean (e. g.,

Dunnette et al., 1962). It is not clear the extent to which each of these measures of faking
actually measure intentional distortion (Paulhus, 1986). Second, some of these
discrepancies may exist because situational variables are not taken into account. Some
situations may inhibit faking on these measures (e. g., knowledge that a lie scale is
included on the measure), while other situations may exacerbate the problem (e. g., high
competition for a job). Third, a number of the studies that have investigated the
relationship between faking and test validity have assumed that there is no variance in
faking (e.g., Becker & Colquitt, 1992; Zickar, Rosse, & Levin, 1996). That is, these
studies have assumed that everyone who is faking increases their scores the same amount
(i.e., by a constant). As Lautenschlager (1994) suggests, if all applicants increase their
scores the same extent (i.e., there is no variance in faking), faking will not change rank
orders and therefore should have no effect on criterion-related validity. However, if
faking does result in changes in the rank orders of applicants (i.e., some applicants
increase their scores through faking more than others), then faking has the potential to
distort criterion-related validity. Initial research suggests that there is variance in the
extent to which individuals fake (Ellingson, Smith, & Sackett, 1999; McDaniel, Douglas,
& Snell, 1997; McFarland & Ryan, in press; Rosse et al., 1998). Therefore, the failure to
consider this variability in some studies may explain why these studies typically do not
ﬁnd that faking has an effect on validity, while others do. Finally, the discrepancies
within the faking literature draw attention to a larger and more fundamental problem.
Currently, there is no theory or model of faking behavior. Understanding what
contributes to the extent of faking is needed before we can resolve debates about effects

due to distortion. Therefore, the present study will use theory to understand an important

 

>4!
h".

 

practical problem so that we may better determine when faking may adversely affect
personality test results.

The present study proposes and tests a model of faking behavior. This model is
presented in Figure l. The model incorporates the theory of planned behavior which has
been used for several years to predict behaviors as diverse as quitting smoking to stealing
(Norman & Conner, 1996; Reinecke, Schmidt, & Ajzen, 1996; Theodorakis, 1992).
Speciﬁcally, the way in which attitude toward faking, subjective norms about faking, and
perceived behavioral control inﬂuence the intention to fake will be examined. I will also
explore the extent to which the intention to fake predicts faking behavior. Finally, three
probable moderators of these relationships will be explored (i.e., valence toward doing
well on the test, warning of a lie scale, and knowledge of the constructs assessed by the
test). By considering the model of faking presented here, future studies may be able to
I consider the relevant issues and inﬂuences on faking. Only then can we determine under
what circumstances faking does matter.

Before describing this model in detail, a definition of faking will ﬁrst be given,
and reasons applicants may fake will be discussed. Second, evidence that applicants can
and do fake will be examined. Third, the way in which faking may inﬂuence the results
of tests will be reviewed. Finally, the model of faking will be described in detail.
Deﬁnitions and Why People Fake

A number of labels have been given to describe the tendency of some
individuals to select answers on tests that will result in others viewing them in the most

favorable way. This has been referred to as response distortion, social desirability,

,4“

Po;

 

L"

r»-

“at.

4
Alt.

.\u

elm.

(a

[(7‘-

ﬂ-

\
in

faking, impression management, self-enhancement, and intentional distortion (Becker &
Colquitt, 1992; Douglas et al., 1996; Mumford & Stokes, 1992, Zerbe & Paulhus, 1987).

Paulhus (1984; 1986) suggests that socially desirable responding contains two
components: self-deception and impression management. Self-deception refers to the
unconscious tendency to see one’s self in a favorable light. Impression management is a
conscious attempt to present false information to create a favorable impression on others.
Therefore, when applicants taking a selection test consciously do not select the most
accurate answer (the one that most closely describes them) but instead select the answer
that they believe will make them look most favorable, they are using impression
management. In the present study, this type of response distortion, whereby an individual
consciously distorts answers in order to be viewed favorably, will be referred to as
faking.

There are a number of reasons to suspect that people may fake responses on a
selection test. Leary and Kowalski (1990) suggest that people are motivated to manage
their impressions when the impressions they make are relevant to the fulﬁllment of their
goals and when these goals are greatly valued. Generally, when an applicant takes a
selection test, he or she desires to get the job. Individuals may fake in order to obtain a
higher score, thus increasing their chances of being hired. Pandey and Rastagi (1979)
showed that attempts to manage impressions increased during a job interview when
competition for the job became more intense. Therefore, it seems that as the stakes

increase, attempts to look the best one possibly can also increase.

Other less obvious reasons individuals may fake are suggested by Schlenker and
Weigold (1992). They suggest that people regulate the impressions they make to
maintain or enhance self-esteem. People generally act to maximize their self-esteem and
will inﬂuence the content of self-presentation to do so. Answering a test in such a way
that others will perceive one positively will help maintain self-esteem.

It is clear that there are reasons applicants may attempt to make themselves look
favorable. The next questions to be addressed are: Can individuals fake such tests? If
they can fake, do they? The following section will review the literature suggesting that
applicants can and do fake personality measures.

Evidence that Applicants C& and Do Fake

For over sixty years research has examined if personality tests can be faked (Bass,
1957; Borislow, 1958; Hunt, 1948; Kelly, Miles, & Terman, 1936; Longstaff, 1948).
Findings are very consistent, showing that people can increase scores on such tests when
they choose to do so. For example, Dunnette et al. (1962) had 62 salesmen take a forced
choice self-description checklist twice in counterbalanced order. In one administration
the subjects were asked to answer honestly, whereas in the other one they were told to
fake (select the best answers). Subjects in the fake condition had mean scores more than
half a standard deviation above those in the honest condition.

Since then several more studies have demonstrated similar ﬁndings (e.g., Cohen
& Lefkowitz, 1974; Hough etal., 1990; Hurtz & Bohon, 1997; Krug, 1978; McFarland &
Ryan, in press; Rosse et al., 1998). For example, Douglas et a1. (1996) administered a
personality test which contained an agreeableness and a conscientiousness scale to 600

college students. They had a between-subjects design in which half of the subjects were

,4

It!

 

mi

0"

l“l
-..
ﬁ

V <‘
.3

Ltd“

.\l..

 

l":

\a !

 

randomly placed into the honest condition and the other half were placed into the faking
condition (told to make themselves look as good as possible). Results revealed
signiﬁcant group differences of up to .70 standard deviations on the personality scale of
agreeableness, and .86 on conscientiousness, with all differences in favor of the fake
condition.

However, there are some limitations of this research that should be pointed out.
Much of the research that has examined the fakability of personality tests has used within
subjects designs which require participants to take the test honestly and also under
instructions to fake. A problem with such designs is that they generally result in order
effects such that individuals who take the test under instructions to respond honestly ﬁrst
increase their scores when asked to fake more than those who did not take the test before
being asked to fake (Klein & Owens, 1965). It is not clear in some of these studies if
conditions were counterbalanced. Therefore, the results of these studies should be
viewed with caution.

These studies suggest that responses on personality measures can be faked. When
asked to do so, subjects can increase their scores by as much as 1 standard deviation
above honest respondents. However, just because applicants can fake personality tests
does not mean they do fake. It is important to make the conceptual distinction between
fakability and actual faking. If applicants can fake, but do not do so, then responses on a
fakable measure may not be biased. Hence, it is crucial to examine the literature for any
evidence indicating that applicants actually do fake when responding to personality

selection measures.

 

Du
conditiom
aimed :
3:16. 1 one:

5167771131:
the‘l‘fis:

Show;

Dunnette et al. (1962) compared the scores of those in the fake and honest
conditions to the scores of 64 applicants applying for a sales job. The applicants
achieved a mean score somewhere between the mean scores of the subjects in the fake
and honest conditions, indicating that the applicants faked to some extent. However, to
determine whether all or only a few applicants tried to create a good impression on the
checklist, a fake key was developed. This was done by weighting those responses
showing signiﬁcant differences between the honest and fake conditions. By examining
the fake score distribution it was determined that a cut score of 4 successfully identified
90% of the faked answer sheets at the expense of misclassifying only 9% of the honest
ones. Through this investigation it was determined that only a few of the applicants
actually faked. Therefore, it was the responses of these few that increased the mean for
the applicant group.

Pannone (1984) administered a rational biographical questionnaire to over 200
applicants for an electrician position. He included an item which made reference to a
piece of electrical equipment that did not exist. Those applicants who said they had used
the piece of equipment were considered fakers. It was determined that over one-third of
the applicants were fakers and there was a signiﬁcant mean difference between the scores
of fakers and non-fakers such that the mean score of the fakers was over .5 standard
deviation units greater than those identified as non-fakers.

This study may be criticized in that it is possible that a number of the fakers were
not identiﬁed. That is, some people who did fake throughout the test did not say that they
had operated the ﬁctional piece of equipment. Similarly, those who did not fake

throughout the test may have been identiﬁed as fakers by their response to the fake item

 

“‘9
.5.

@971

Lyn!

R..

.-
is

but were actually just confused by the item or were just careless when responding to the
item. Therefore, this one item measure of faking may also be tapping carelessness of
responding rather than isolating conscious distortion.

Further evidence that applicants may fake selection tests come from Schmit and
Ryan (1993). They administered the NEO-FFI, a personality test used to measure the
traditional ﬁve factor model (extroversion, agreeableness, conscientiousness, emotional
stability, and intellectance or openness to experience), to applicants and students. Using
conﬁrmatory factor analyses, they found the ﬁve factor model ﬁt the data from the
student sample but the applicant data contained six factors. The authors suggested this
sixth factor (an “ideal-employee factor”) that surfaced in the applicant sample may have
resulted from applicant faking.

Becker and Colquitt (1992) administered a biodata inventory to 250 real
applicants. They also had another group of individuals take the same test under
instructions to fake and some under instructs to respond honestly. Results indicate that
the scores of the applicants fell between the scores of those in the honest and the fake
conditions (they scored lower than the fakers but higher than the honest respondents),
indicating that applicants do fake responses.

McDaniel et a1. (1997) had 192 subjects complete a conscientiousness and a
social inﬂuence measure that were mailed to them. Respondents ﬁrst completed the
conscientiousness and social inﬂuence measures as they would if they were applying for
a job that they really wanted. They were then instructed to respond to these measures a
second time honestly. For the conscientiousness measure, 23% of the sample improved

their score in the applicant condition and 26% improved their scores in the applicant

 

conditi
the san
the con
their cc

Y
Scalest

“Mining,

condition for the social inﬂuence measure. However, over 65% of the sample obtained
the same score across conditions for the social inﬂuence scale as compared to 72% for

the conscientiousness measure. When responding as applicants, respondents increased
their conscientiousness scale score by .26 standard deviations and their social inﬂuence
scale score by .16 standard deviations.

McDaniel et al.’s (1997) ﬁndings may be questioned on the grounds that those
who misrepresented themselves may have been less likely to respond to the mail survey
for fear that their responses would be traced. However, if this is true then their results
probably underestimate the true misrepresentation rate of the sample. This study may
also be criticized because the conditions were not balanced; thus, order effects may have
been present.

Hurtz and Bohon (1997) instructed subjects to answer a personality-based
integrity test as they would if they were applicants for a retail sales position. The mean
score of those in this applicant condition was over .5 standard deviations higher than
those who took the test under instructions to answer honestly.

Although there is considerable evidence that applicants do fake responses on
personality tests, there are some who do not believe this to be true. Hough et al. (1990)
administered a temperament inventory to military incumbents that included a social
desirability scale to detect intentional distortion in a favorable direction. All respondents
were told to answer the inventory honestly but some were later asked to answer so as to
look good (fake condition). The inventory was also administered to 125 individuals
going through the Military Entrance Processing Station (MEPS) who were told that the

test would be used to make decisions about their careers. Results showed that

incum

appllC

of the

ﬁres: v
not to.“

been. .

incumbents were able to increase scores by faking when told to do so; however, the real
applicants did not fake. The social desirability scores of the MEPS were similar to those
of the incumbents in the honest condition.

One must keep in mind that the MEPS are not traditional applicants. That is,
these were individuals who were already enlisted in the military. Therefore, they were
not job applicants and may not have been as motivated as job applicants would have
been. Additionally, the researchers used scores on a social desirability scale to identify
fakers. Research has shown that these lie scales often fail to isolate conscious distortion
because they also measure the self—deception component (i.e., the unconscious part) of
socially desirable responding (Paulhus, 1986). Not enough information was provided
about the social desirability scale used in this study to determine if it was ﬂawed in this
manner. Thus, results should be viewed with caution.

In summary, most of the evidence suggests that at least some applicants do fake
their responses on personality measures. There is also evidence that applicants increase
their scores by as much as a half of a standard deviation by doing so (e. g., Dunnette et al.,
1962; Stokes, Hogan, & Snell, 1993). Now I will examine why applicant faking may be
a concern to those using personality measures for selection purposes.

Inﬂuence of Faking on the Results of Persorglity Measures

There is concern that applicant faking may make it more difﬁcult to interpret
personality test results due to its effect on selection decisions, criterion-related validity,
and construct validity (i.e., factor structures, scale reliabilities, and correlations with other
measures). However, others disagree that faking negatively affects test results. In the

following section, the research that has been conducted with respect to the effects of

10

a--

-.‘.U tea:

and CU.C
the 350.6
the mean
“'35 3141:

s

“dense

faking on the results of personality tests will be examined. First I will present the
research that has examined how faking may inﬂuence selection decisions that are based
on personality measures. Second, the possible effects on criterion-related validity will be
explored. Finally, the way in which construct validity has been shown to be affected by
faking will be examined.

Selection Decisions. As discuSsed earlier, research has consistently found that
faking increases the mean scores of applicant groups (Becker & Colquitt, 1992; Dunnette
et al., 1962; McDaniel et al., 1997; Pannone, 1984). Again, ﬁndings indicate that when
asked to fake, groups can increase scores by as much as a standard deviation through
faking, and applicant groups tend to score a half of a standard deviation greater than
groups asked to respond honestly (Hough etal., 1990; Hurtz & Bohon, 1997).

Such changes in applicant scores have obvious implications if an organization is
using an absolute cut score to select applicants on this type of test. More applicants than
would have been selected if everyone responded honestly will be selected if applicants
fake. For example, Becker and Colquitt (1992) administered a biodata inventory to over
250 real applicants. To determine the impact of faking on selection decisions, Becker
and Colquitt assumed that every applicant distorted his or her score. They then adjusted
the score distribution by subtracting three points from everyone’s original score (this was
the mean difference between the honest and applicant scores). The hiring decision that
was made on the original score (hire or not hired) would stand for 83.1% of the
applicants when the new distribution of scores was considered. This study provides

evidence that, even if we assume that faking is a constant across individuals (everyone

11

mmgm

11.
£1 32:11:.
in change
sciecwn

A
commie.
the 16PF

organizer

fakes and increases their scores the same amount through faking), it has the potential to
change hiring decisions, such that those who normally would not be hired will be.

However, Becker and Colquitt (1992) based this analysis on the assumption that
all applicants faked the same amount. Thus, they failed to consider that faking may result
in changes in the rank order of applicants. If rank orders change this would likely change
selection decisions more drastically.

A more thorough analysis of how faking inﬂuences selection decisions was
conducted by Christiansen, Gofﬁn, Johnston, and Rothstein (1994). They administered
the 16PF and a response distortion scale to 495 incumbents in a large factory products
organization who were participating in an assessment center. Subjects were told that
responses to these measures would be used for future operational purposes (e.g.,
selection, promotion). They then examined the effect of a social desirability correction
on selection decisions for different selection ratios. It was determined that, depending on
the selection ratio, correction of scores would have resulted in different hiring decisions
than those that would have been made on the basis of uncorrected scores. For example,
with a 15% selection ratio 12 candidates who would have been selected on the basis of
uncorrected 16PF scores would not have been selected on the basis of faking corrected
scores. Results indicated that at lower selection ratios there was a greater percentage of
discrepant hires. In fact, they observed a change in rank order for over 85% of the
applicants with the application of the social desirability correction to the personality scale

SCOTCS .

12

Hough er
decrsroris
responses
was beset
increase 1

seiectzon
percents;
decreased
34? 0." ti.
Results u
the YEP-.1; c
O:
1:6 scale t.
39min} n
”ﬂuke .
fake Se} {-1
reg“ DU:

mss‘ RI):

Ellingson, Sackett, and Hough ( 1999) found similar results. Using the data of
Hough et a1. (1990), Ellingson et al. determined the percentage of correct top-down hiring
decisions that would be made using the fake responses and then using the corrected
responses (i.e., lie scale scores partialled from fake scores). The “true” hiring decision
was based on the applicant’s standing in the honest condition. Results indicated that an
increase in the selection ratio corresponded to an increase in the proportion of correct
selection decisions in the faking condition. Also, holding everything else constant, as the
percentage of individuals faking increased, the pr0portion of correct selection decisions
decreased. For example, given a 10% selection ratio, if 30% of applicants faked, only
34% of those who would have been hired under the honest condition would be hired.
Results were similar for the corrected scores. Therefore, even when faking is corrected,
the rank orders are not similar to those in the honest condition.

One problem with the majority of these studies is that most have used scores on a
lie scale to operationalize intentional distortion. It is not clear how well these scales
actually measure faking behavior. However, despite this, there is still considerable
evidence to support that there are individual differences in the extent to which individuals
fake self-report measures and that this may signiﬁcantly inﬂuence selection decisions
(e.g., Dunnette et al., 1962; Hough, 1998; McDaniel et al., 1997; McFarland & Ryan, in
press; Rosse et al., 1998). If faking does result in changes in rank orders, it has the
potential to alter the criterion-related validity of these measures. Research that has
examined how faking inﬂuences the criterion-related validity of personality measures

will now be reviewed.

13

 

 

 

and ,0

Criterion-Related Vaﬂm. In their study described earlier, Dunnette et al. (1962)
found that faking exerted an effect on the validity of their personality checklist.
Managerial ratings were provided for 45 of the salesmen who had answered the checklist
under the honest and the fake conditions. Correlations between ratings and checklist
scores under the honest condition were relatively large (reaching .38) and signiﬁcant.
However, when the ratings were correlated with scores of those in the fake condition, the
correlations decreased and became non-signiﬁcant. As described earlier, they determined
whether all or only a few applicants tried to create a good impression by developing a
fake key on the checklist and determined that only a few of the applicants actually faked.
Therefore, the authors concluded that even the presence of a few fakers may decrease the
validity of such measures.

Douglas et al. (1996) obtained job performance ratings from employers of
subjects who worked and gave permission to contact their supervisor. Evidence that
faking decayed the criterion-related validity of the scales (conscientiousness and
agreeableness) was found. The average validity of the scales was .26 for honest subjects
and .04 for those who took the tests under instructions to fake. Further analyses revealed
results similar to those found by Dunnette et al. (1962). Douglas et al. randomly selected
a few of the subjects from the faking condition and added their data to the data of the 97
honest subjects so that the combined sample contained the data of 10 to 25 percent of the
faking subjects. Through this procedure they found that as the percent of fakers in the
sample increased, the number of fakers scoring within the top ten increased as well. For
example, it was found that if 10% of the people faked, ﬁve of the top ten subjects were

fakers, whereas if 25% of the sample faked, nine of the top ten subjects were fakers. The

14

mean “'1:
Moreox'ei
validity 0
area. film
.\1
their met
tor: ‘ient:

CKICDI IO ‘

II P.“
his ”013;;
rm ..

mean validity decreased somewhat as the percentage of fakers in the sample increased.
Moreover, it was found that as the number of fakers in the top ten increased, the mean
validity of the measures dropped substantially for those who faked. They found that even
a few fakers may distort the results of personality measures.

McDaniel et al. (1997) had respondents rate how their supervisor would evaluate
their overall job performance (on a four point scale) in addition to the measure of
conscientiousness, social inﬂuence, and misrepresentation. Results indicated that the
extent to which respondents misrepresented themselves was negatively correlated with
self-reported job performance. They also found that the validities were higher in the
honest condition for both the conscientiousness measure (.14 vs. .06) and the social
inﬂuence measure (.23 vs. .19). A criticism of this study is that the two measures were
validated against a self-reported performance measure. However, one would expect that
if the self-reported performance measure was not accurate it would be slanted in the
positive direction. This would lead to an underestimation of the decline in validity.
Therefore, these results may be conservative.

Although there is some evidence that faking changes the criterion-related validity
of personality measures, there is also evidence to the contrary. Hough et al. (1990)
developed a temperament inventory called “Assessment of Background and Life
Experiences” or ABLE. This inventory consisted of ten content scales that measured six
temperament constructs and a social desirability scale that was included to detect
intentional distortion in a favorable direction. Hough et al. administered the ABLE to
military incumbents who took the test under both honest and fake conditions. All of

these respondents were told to answer questions honestly, but half also took the test under

15

instruct

i'aftea' ,

mentor

1......‘1”- oft

DC 93711:»

authors I

‘l‘
¥
.
3M 1~~ .
.f-1\ ‘1‘

instructions to fake good and the other half were told to fake bad (possibly so as to not be
drafted). The conditions were counterbalanced. Results indicated that the response
validity scales did detect the different types of distortion. Moreover, intentional
distortion, in a positive direction, did not attenuate the criterion-related validity of the
inventory.

The method Hough et al. (1990) used to obtain these results is questionable. The
data of the incumbents was skewed so that meaningful moderated regressions could not
be performed to determine the impact of faking on the inventory. To deal with this the
authors took the mean social desirability scale score of those in the fake good
experimental condition and used this to divide the concurrent validity sample into two
groups. Those scoring at or above the mean of the fake group were labeled “overly
desirable,” while those scoring below this mean were labeled “accurate.” They then
calculated the criterion-related validities of the ABLE for each of these groups separately.

Hough et al. (1990) found that the criterion-related validity did not change
substantially for those in the “overly desirable” group; however, the incidence of faking
in this group still may not have been very high. As Dunnette et al. (1962) showed,
applicants can fake but few actually do. Therefore, it is likely that even those within the
top half were not fakers. Instead of splitting the sample in half, Hough et al. should have
also examined the top one-third or top ten percent of scorers on the social desirability
scale within the incumbent sample. Perhaps if this was done the criterion-related validity
would have dropped substantially for the “overly desirable” group. It may also be that

the individuals in the “overly desirable” group had true high scores on the measure.

16

dam and
con-diner
as the pe
decision:
concigde
orders. l1
va‘idzr} o
c521} cor
Ii :
data as H:
31.52:} t
he sure if :
TS:
(”Jimmie
fab? Frc
P3511101] 0::
Selected 3?!
C“wilful-re:
Confuse; St

ta: {hate u

 

 

550185 be,» C

 

As discussed earlier, Ellingson et al. (1999) reanalyzed the Hough et al. (1990)
data and found that the corrected mean score was lower than the mean score in the fake
condition but greater than the mean score in the honest condition. It was also found that
as the percentage of individuals faking increased, the proportion of correct selection
decisions decreased. Therefore, using the same data as Hough et al., Ellingson et al.
concluded that applicants can fake a personality measure and faking will change rank
orders. If rank orders change, faking has the potential to inﬂuence the criterion—related
validity of these measures. They also demonstrated that the effects of faking are not
easily corrected.

It should be pointed out that, since the Ellingson et al. (1999) study used the same
data as Hough et al. (1990) it is subject to the same criticisms. That is, the Ellingson et
al. Study may be ﬂawed because it used a lie scale to assess faking. Therefore, we cannot
be sure if the results would hold if an alternative measure of faking was used.

The study discussed earlier by Christiansen et a1. (1994) also examined how the
criterion-related validity of a personality measure would be affected if corrected for
faking. From the 495 participants, 84 were selected for an upper level supervisory
position on the basis of the assessment center evaluation. Using the data of the 84
selected applicants for whom criterion data were available, the effect of correction on the
criterion-related validities of the traits relevant to the job was examined. Using faking-
corrected scores and through using hierarchical regression analyses, it was determined
that there was not a signiﬁcant increase in the criterion variance explained by corrected

scores beyond that explained by uncorrected scores. The authors concluded that this

17

 

 

 

demonstr;
personalir
B.
measure 0‘
Responin
samples 0'
ad self-cl
faking or. i
\‘oiuntm
T:

from a m:
desrrabm:

mltﬂl 2..“

 

 

Tire) eh-

d; X,
m PEFSOIIE

demonstrated that faking does not inﬂuence the criterion-related validity of the
personality measure.

Barrick and Mount ( 1996) administered the Personal Characteristics Inventory (a
measure of the FFM of personality) and Paulhus’ Balanced Inventory of Desirable
Responding (which measures both self-deception and impression management) to two
samples of long-haul semi-truck drivers. They found that both impression management
and self-deception related to scores on the personality scales. However, they found that
faking on the tests did not signiﬁcantly reduce their predictivness of supervisor ratings or
voluntary turnover.

The strongest evidence that faking does not alter criterion-related validity comes
from a meta-analysis conducted by Ones et al. (1996). They meta-analyzed the social
desirability literature to determine if social desirability operates as a predictor of some
criteria, as a mediator, or as a suppressor when personality tests are used as a predictor.
They also set out to determine if social desirability is related to real individual differences
in personality. They did this by comparing scores on social desirability scales to self and
other reports on a personality inventory. The authors found social desirability correlated
.37 with emotional stability, .20 with conscientiousness, .14 with agreeableness, and .06
with extroversion. There was no correlation between Openness to experience and social
desirability. To determine if these correlations represent the susceptibility of these scales
to socially desirable responding or if the correlations represent the relation of social
desirability to substantive personality constructs, other people’s (e. g., friends, relatives)
ratings of an individual on the personality measures were examined. There were similar

correlations between social desirability scale scores and others’ ratings of personality.

18

T335. 11'

#350712

pert or:

of five

tr;

\

P‘
P

III;

Thus, there is evidence that social desirability is related to real individual differences in
personality.

Ones et al. (1996) also found that social desirability did not predict job
performance. They calculated the operational validity of social desirability’s prediction
of ﬁve frequently used performance criteria (school success, task performance,
counterproductive behaviors, training success, and supervisory ratings of job
performance). Only training success was predicted by social desirability scale scores.

To be a mediator social desirability must correlate with both job performance and
personality variables. Because it does not correlate with most performance criteria, social
desirability does not act as a mediator. In addition, by partialling social desirability from
personality measures, Ones et al. (1996) were able to investigate if such responding
impacts the criterion-related validities of the Big Five variables. When this was done, the
validities did not change; thus, it was concluded that social desirability did not attenuate
the criterion-related validity of the personality measures. It was also determined that
social desirability did not act as a suppressor.

The authors concluded that the concern with the susceptibility of personality tests
to socially desirable responding is needless. Their results indicate that social desirability
measures true variance in personality and does not attenuate criterion-related validity.
Therefore, controlling for such responding is not necessary. However, Ones et al. (1996)
may have overstated their ﬁndings because the study contained a major ﬂaw.

The scales used by Ones et al. (1996) to measure socially desirable responding not
only measure respondents’ conscious attempts to present a positive image (impression

management) but also measure the self-deception component of social desirability. For

19

‘3‘”

AI‘LSL

'"D"
hkua.

‘1

example, it has been shown that the Marlow-Crowne scale and the MMPI K Scale, which
were used by Ones et al., load on both the self-deception and impression management
factors (Paulhus, 1984; 1986). Therefore, these scales are not only measuring conscious
attempts to distort, but also the unconscious component.

It is not surprising that measures of self-deception would correlate with
personality constructs and not attenuate the criterion-related validity of the personality
measures. It has been suggested that if self-deception is controlled for, the usefulness of
self-reports may actually decrease because self-deception explains true personality
variance (Zerbe & Paulhus, 1987). Therefore, it may very well be that the self-deception
component of social desirability was the cause of the correlation between desirability
scale scores and personality dimensions. It would be interesting to see if this relationship
still existed if self-deception was partialled out of the social desirability scores.

Ones et al. (1996) can conclude that the aggregate social desirability scale scores
(measuring both self-deception and impression management) used in their analyses
correlated with personality dimensions and did not attenuate the validity of personality
tests. However, they cannot conclude anything about the relationship between
respondents’ conscious attempts to distort their answers to present themselves positively
and personality dimensions or validity. Therefore, Ones et al.’s suggestion that faking
should be of no concern to those using self-reports of personality is premature. Perhaps
self-deception should be of no concern; however, there is still no conclusive evidence
which indicates that applicant attempts to consciously distort responses do not bias

personality test results.

20

lie

or: person;

 

Theon ['81

different. i.

.. 4 -,
SILDIQCM i

shew: 1;

Zickar et al. (1996) used a different approach to analyze the effect that faking has
on personality measures. They conducted a Monte Carlo simulation using Item Response
Theory (IRT) to model faking on a personality test. Results indicated that across
different faking magnitudes (variance of faking between respondents) and faking
prevalence rates (percentage of fakers in a validation sample), validities did not decrease
a signiﬁcant amount. Also, the mean observed scale difference between validation
samples in which fakers were included and samples with no fakers was less than .25
standard deviations for all but the most extreme faking conditions. Thus, faking was not
shown to moderate validity. However, their ﬁndings indicated that as the percentage of
fakers in the sample increased, the percentage of fakers in the top end of the distribution
increased as well. Such ﬁndings have implications for top-down selection procedures as
this indicates that the fakers may be the ones selected.

Zickar (1997) sought to replicate the ﬁndings of Zickar et al. (1996) but based his
analysis on a more complex and realistic model of faking. He examined a computer
simulation of faking on the Work Orientation scale from the Army’s Assessment of
Background and Life Events (ABLE). In this simulation, Zickar considered three factors
not considered by Zickar et al. First, he varied the variance of faking magnitude or the
variance between subjects in the extent of faking. Second, the correlation between faking
magnitude and the latent trait was systematically manipulated. Finally, the percentage of
items deemed fakable was varied. Results indicated that as the variance of faking
increased, the validity decreased; however, this correlation was small and not signiﬁcant.
The percentage of items faked had a curvilinear effect on the validity correlation; that is,

the most distortion occurred when 75% of the items were fakable and the least amount of

21

distortion occurred when 25% and 100% of the items were fakable. The correlation
between the squared percentage of items faked with observed validity was signiﬁcant (r =
-.61), while the correlation between faking magnitude and the latent trait and observed
validity was not signiﬁcant. Overall, validity was inﬂuenced the most when the variance
in faking was large, there were negative correlations between faking and the latent trait,
and 75% of the items were faked. Thus, it seems that in speciﬁc situations, faking may
decrease the criterion-related validity of non-cognitive measures.

Hough ( 1998), in an effort to determine why there are such discrepancies in the
literature regarding the relationship between faking and criterion-related validity, sought
to determine the cause of these differences. Hough suggested that the different ﬁndings
were a result of the different samples that were used. By reviewing much of the literature
she determined that when incumbents are examined, little faking (as measured by scores
on a lie scale) occurs and criterion-related validity is unchanged as a result of this faking.
When applicants serve as subjects, more faking (as measured by a lie scale) occurs,
therefore the criterion-related validity is slightly compromised, but not substantially.
Finally, when research subjects are the focus of the study and are directed to fake (i.e.,
there is a large amount of faking) criterion-related validity will be substantially reduced.
Therefore, Hough concludes that faking in real selection contexts should not be a concern
because applicants do not distort responses enough in such situations to reduce the
criterion-related validity substantially.

There is one major problem with Hough’s (1998) reasoning. She failed to
consider that the measure of faking (i.e., the operationalization of faking) may explain

these discrepancies between the type of sample used and how faking inﬂuences the

22

critenon’ri

 

mcumbent
Houever. '
esurratior,
nmczec
conscious
used in sq
a ho are 1’!

‘. 13110113

 

1a

 

criterion-related validity. For example, in all the studies cited by Hough that used
incumbent and applicant samples, a lie scale was used as the measure of faking.
However, when subjects are directed to fake there is no need to use the lie scale as an
estimation of faking because it is already known which group of people faked (i.e., those
instructed to do so). Therefore, again, it may be the failure of these lie scales to isolate
conscious distortion that accounts for the discrepancies of these results. The lie scales
used in studies involving incumbents and applicants may not really be identifying those
who are faking. Thus, we cannot know what the true relationship is between faking and
criterion-related validity in these samples.

Hough (1998) went on to conduct a study that investigated how applicant faking
related to criterion-related validity. However, once again, a lie scale was used as the
measure of faking. Therefore, it was found that faking did not alter the criterion—related
validity of the personality measure in question. Again, we cannot be sure to what extent
this scale isolates conscious distortion; therefore, these results are not conclusive.

Con_struct Valﬁty. Studies have also found that faking may inﬂuence the
construct validity of personality measures through distorting factor structures, the
correlations between personalin scales, and the reliability of scales. As discussed earlier,
Schmit and Ryan (1993) administered the NEO—FFI to applicants and non-applicants
(students). They found that factor intercorrelations were higher for the applicant sample
when compared to intercorrelations in the student sample. Moreover, the ﬁve factor
model ﬁt the data from the student sample, but the applicant data contained six factors.
The ﬁrst factor for the job applicant sample was “a large, work-related personality-

characteristic dimension” which they labeled the “ideal-employee factor.” This factor

23

contained most of the items from the Conscientiousness scale, but also included items
from the other four scales. One possible explanation for this extra factor is that
applicants, unlike students, wish to present themselves as a good employee. It may be
that applicant faking created this ﬁrst factor.

A similar study was conducted by Cellar, Miller, Doverspike, and Klawsky
(1996). They examined the factor structure of the NEO-PI based on a sample of 423
ﬂight attendant trainees. They also found that the six factor model ﬁt better than the ﬁve
factor model. However, the sixth factor in their study did not appear to be an ideal
employee factor. Instead, this factor appeared to be a method factor. The differences
between the ﬁndings of Schmit and Ryan (1993) and Cellar et al. may be attributed to the
samples used. Schmit and Ryan used actual applicants while Cellar et al.’s sample
consisted of trainees who had already passed the selection procedure to become a ﬂight
attendant. Therefore, it seems that the sample used by Cellar et al. may have been less
motivated to do well or to fake. Cellar et al. also examined applicants for a different type
of job. These two inﬂuences could have resulted in different factor loadings on the sixth
factor.

When Ellingson et al. (1999) reanalyzed the Hough et al. (1990) data they also
determined the effect of correcting for social desirability on the construct validity of the
measure. Ellingson et al. factor analyzed the faked scores, the corrected scores, and the
honest scores for the ten personality scales. Results indicated a one factor solution for
both the fake and corrected scores, but a two factor (more complex) solution for the

honest scores. The authors concluded that faking dissolved the complex

24

multidimensionality of their measure. Thus faking results in scores which no longer
reﬂect personality traits but instead reﬂect faking behavior.

However, Ellingson et al. (1999) Operationalized social desirability by artiﬁcially
inducing its presence. That is, they speciﬁcally asked participants to fake good. Perhaps
giving direct instructions to distort responses produced an extremely distorted socially
desirable response set consistent with an overpowering social desirability factor. An
applied setting may elicit a form of faking that is more reﬁned and may have a different
effect (or no effect) on the factor structure of a personality measure. To deal with this
limitation of Ellingson et al. (1999), Ellingson, Smith, and Sackett (1999) examined how
socially desirable responding inﬂuenced the factor structure of personality measures
across four different samples. One sample included incumbents, the second examined
applicants, the third included applicants and incumbents, and the fourth included
applicants, incumbents and students. Additionally, each data set included a different
personality measure, and different data collection contexts (i.e., applicants took it within
selection context, students within a research context, and incumbents within a
developmental context). Within each sample, the ﬁt of the factor structure of the
personality test in question was examined separately for those who were labeled “low in
social desirability” and those considered “high in social desirability.” Note that
individuals were categorized as low and high socially desirable responders according to
their scores on a lie scale (those who scored in the top and bottom third of the scale were

examined).

25

 

deception

bersonelir

a result .3
11'] a deer:

Comm;-

Results indicated that correlation matrices computed among the personality scales
for the high and low social desirability groups were highly similar. Additionally, across
all samples the factor structure models demonstrated good ﬁt when applied to the high
socially desirable groups. However, socially desirable responding did seem to affect
measurement error as error variances were not equal across the two groups (i.e., those
high and low in socially desirable responding). However, the authors concluded that
faking did not result in signiﬁcant changes in factor structures.

In a study discussed previously, Barrick and Mount (1996) found neither self-
deception nor impression management altered the criterion-related validity of a
personality test. However, they did ﬁnd that the correlations observed between the
personality scales were higher within their applicant sample as compared to non-
applicant samples. They suggest that this increase in the relationship among the scales is
a result of applicant faking. Therefore, although they ﬁnd no support that faking results
in a decrease in criterion-related validity, they did provide some evidence that the
construct validity of personality measures may be compromised by faking.

McCrae and Costa ,( 1983) found evidence that construct validity is not
compromised by applicant faking. They administered the NBC inventory and two social
desirability scales (the Marlowe-Crowne and the Lie scale from Form A of the Eysenck
Personality Inventory) to 215 volunteers. Six months later, the subjects’ spouses were
asked to rate their spouse’s personality using the same personality inventory completed
by the subject. All correlations between the self and spouse ratings were signiﬁcant and
ranged from .25 to .61. Partialling social desirability out led to signiﬁcant correlation

increases between self and spouse ratings for only two traits (activity and positive

26

emotions .
when soci.

support t'r.

 

 

 

emotions). For most traits, the correlation between the self and spouse ratings dropped
when social desirability was partialled out. The authors concluded that these ﬁndings
support the argument that socially desirable responding does not decrease the validity of
personality measures. However, McCrae and Costa only examined socially desirable
responding, not intentional faking. The subjects in their study were volunteers so they
were most likely not motivated to fake their responses. Also, the two social desirability
scales used were not designed to detect respondents’ attempts to consciously distort
answers. These measures tap both self-deception and impression management.
Therefore, this study may say little about how faking distorts the validity of personality
tests.

Collins and Glese (1998) found that the Five Factor Model held in an applicant
sample. Assuming that the applicants in their sample faked, this is evidence that faking
did not inﬂuence the construct validity of the personality measure.

Ones and Viswesvaran (1998) conducted a meta-analysis to determine the
inﬂuence of social desirability on the convergent and divergent validity of a personality
scale. The number of correlations analyzed exceeded ﬁve thousand and the total sample
size was over 4 million. Results indicated that both convergent and discriminant
validities of the Big Five dimensions remained relatively the same when social
desirability was partialled out. The authors conclude that their ﬁndings indicate that
faking does not adversely inﬂuence the construct validity of personality measures.
However, again, the meta-analysis included studies that Operationalized faking as scores
on a lie scale. Therefore, we cannot be sure if results would be the same had intentional

distortion been isolated and partialled out of the correlations.

27

13
personal:
faster an
expicratt
The data
and tn 0
respond:

shat Sc':

10 distor

faking d

IOU? Sca
xICFaﬂC

Persona

Douglas et al. (1996) found faking distorted the construct validity of their
personality and biodata scales as measured by multi-trait and multi-method analyses and
factor analyses. When the data of those in the honest condition were examined using an
exploratory factor analysis, four factors were generated. All four were construct factors.
The data from the faking applicants generated four factors but two were method factors
and two were construct factors. Thus, a measure based on the responses of honest
respondents may have a different structure when respondents fake. This is similar to
what Schmit and Ryan (1993) found with the ﬁve factor model of personality. Although
this study tells us that, when asked to fake, the responses of subjects may distort factor
structures, it is not known if the extent to which applicants really fake would be enough
to distort the factors in this way. Nonetheless, these results provide evidence that if
faking does occur it may change the psychometric properties of the measure.

Douglas et al. (1996) also found that intemal consistency reliabilities for their
four scales were higher for the faking condition than for the honest condition.
McFarland, Ryan, and Ellis (2000) found that when individuals were instructed to fake a
personality measure, the reliability of the scales increased. These ﬁndings suggest that
honest responding may hurt the homogeneity of the items. Subjects who fake make more
of an effort to report consistently positive responses. In other words, faking creates an
artiﬁcial inﬂation of consistency. Honest applicants report inconsistent behavior because
their behavior truly is inconsistent. If it is true, and alphas do change as a result of
faking, then these measures are tapping both conscious distortion and the construct which

they were designed to measure. This is a violation of the unidimensionality assumption

28

of alpha. This implies that alpha should perhaps not be used as a measure of reliability in
settings where some respondents may be faking their answers.
Mam

There is still debate about the way in which faking inﬂuences test decisions.
Although there is consistent evidence that applicant faking may alter selection decisions,
it is less clear how the criterion-related validity and construct validity of these measures
is affected. There are several reasons this area may be plagued by such discrepancies.

First, as stated earlier, the operational deﬁnition of faking changes across these
studies. Some studies use a score on a lie scale as a measure of faking, while others use
one item to measure faking, and still others examine increases in scores from an honest
condition. Second, it may be that different designs produce different results. Some
studies have used within-subject designs, while others have use between subject designs.
Third, the samples used may explain these discrepancies. Some studies have used
college students and asked them to pretend to be applicants, while others have used
applicants or incumbents to determine the effect of faking on validity. Finally, perhaps
ﬁndings are so discrepant because we still do not understand how various factors may
inﬂuence the results of each study (e. g., situational cues, knowledge of what the test is
measuring). For example, some studies asked participants to fake good but also warned
them that a lie scale was included on the measure (Doll, 1971; Hough, 1990), while
others have included no such additional instructions (McFarland & Ryan, in press;
Pannone, 1984). Some have explained to participants what types of constructs the test in
question measures (Cunningham et al., 1994; Dwight & Alli ger, 1997), while many have

not done this (Hough, 1998). Therefore, it is not surprising that studies have found such

29

‘ ﬁ
.W W~.I
Irilldellgl

011 a per
fiﬁure

TClCiIIOn:

 

conﬂicting results. Before we can really understand how faking may inﬂuence test
results, we have to understand how these different factors may inﬂuence faking behavior.
That is, we need a model of faking behavior to understand how and in what way various
factors may inﬂuence applicant faking.
A Model ofjgkﬁg

This study is being conducted to test the model of faking behavior presented in
Figure 1. This model is being proposed as a way to predict, explain, and ultimately to
understand faking behavior. This model does not address all of the possible explanations
for the discrepancies in the faking literature (e.g., study design). However, it does
address issues of the operationalization of faking and understanding of the various
inﬂuences on faking behavior.

The current study will examine the usefulness of this model for predicting faking
on a personality measure, it may be used to understand faking behavior on any self-report
measure (e.g., biodata form, integrity test). Below, each variable in this model and its

relationship to faking behavior will be described.

30

 

 

£2;

 

 

coca—32
wEom 833.80
Co omnogocx-

3E 9 3:5...

332.5 mac—am we Eco—2 A Semi

 

 

 

 

 

 

 

8320m—
was;

 

28m 34 he wEEaB-
3:58:33.
L8 coco—w?

moose—DEE 3:23.35

 

 

 

 

 

 

 

33m
3 83:25

 

 

 

wise
Beach. 062:2

. L.
F 882 0330.33
/ 35:00
EoFEem

©0385;

 

 

 

 

 

 

 

 

 

31

of plnne.
behavior
intenzion ~

Norma :

 

 

 

 

'he theori
Regent's-
smoking

10°11. .

x]

330' PCIC:

intenttor

Attitudes, Beliefs, and Perceptions of Control, and Intention to Fake. The theory

of planned behavior, originally proposed by Ajzen (1985), will be used to predict faking
behavior. This theory has been used extensively within social psychology to predict
intentions, and (through intentions) behavior (Ajzen, 1991; Ajzen & Madden, 1986;
Norman & Conner, 1996; Reinecke et al., 1996; Schifter & Ajzen, 1985). For example,
the theory of planned behavior has been used to predict condom use (Boldero, Moore, &
Rosenthal, 1992; Reinecke et al., 1996), recycling (Boldero, 1995), success quitting
smoking (Godin, Valois, Lepage, & Desharnais, 1993), and stealing (Beck & Ajzen,
1991). According to this theory, one’s attitude toward the behavior, subjective norms,
and perceived behavioral control predict the intention to perform the behavior. In turn,
intentions predict behavior.

Attitude toward the behavior refers to the degree to which a person has a
favorable or unfavorable evaluation of the behavior in question. More speciﬁcally, it is
an evaluative appraisal of the consequences of acting or not acting (Ajzen, 1991). This
appraisal may be either favorable or unfavorable, and is generally expressed with
semantic differential items (e.g., good-bad, pleasant-unpleasant).

Subjective norm is a social factor predicting intentions. This refers to the
perceived social pressure to perform or to not perform the behavior. For example, items
used to measure subjective norms assess the likelihood that important individuals (e. g.,
parents, friends) would disapprove or approve of the behavior. Therefore, an individual
who thinks friends and family members who are important to them would disapprove of

faking on a personality test will have a negative subjective norm for faking behavior.

32

51'

C)

n

(D

‘.’ '7

I".

Perceived behavioral control is the third factor proposed to predict intentions.
This refers to the individual’s belief regarding the ease or difﬁculty with which the
behavior can be performed. Individuals who believe they can perform the behavior are
more likely to intend to behave in that manner.

The theory of planned behavior is an extension of the theory of reasoned action
(Fishbein & Ajzen, 1975). Unlike the theory of reasoned action, the theory of planned
behavior does not assume that the behavior in question is completely under one’s
volitional control. Therefore, the theory of planned behavior goes a step further than the
theory of reasoned action by considering an individual’s perception of control over the
behavior in question. It is believed that the inclusion of this additional variable is why
the theory of planned behavior has been shown to predict behavior better than the theory
of reasoned action (Ajzen, 1991).

The theory of planned behavior will be used to predict the intention to fake in the
following way. Speciﬁcally, it is hypothesized that attitudes toward faking personality
tests in selection contexts, subjective norms of faking (i.e., perceptions of whether or not
important others would approve or disapprove of faking behavior), and perceptions of
whether or not one could fake the test, will have a direct effect on the intention to fake.

It has been suggested that attitudes regarding dishonesty (whether it be theft,
cheating, lying, etc.) are related to intentions to behave in a dishonest manner. For
example, it has consistently been found that positive attitudes toward a behavior lead to
greater intentions to perform that behavior (e.g., Ajzen, 1991; Boldero, 1995; Boldero et
al., 1992). Beck and Ajzen (1991) found that attitudes toward dishonest actions predicted

intentions to cheat on a test, shoplift, and lie. Since positive attitudes toward a behavior

33,

have con:
following
Hmothe

fake.

or acre;
instance
that mo.
based 0
other Ct

society

0"]

A;

have consistently been linked to stronger intentions to perform that behavior, the
following prediction will be made:

Hypothesis 1: Attitude toward faking will be positively correlated with the intention to
fake.

It has been suggested that the belief that certain dishonest behaviors are common
or acceptable is a determinant of many forms of dishonesty (Murphy, 1993). For
instance, it has been found that an individual is more likely to steal if he or she believes
that most people steal (O’Bannon, Goldinger, & Appleby, 1989). Most honesty tests are
based on evidence that suggests that individuals who are more likely to commit thefts and
other counterproductive acts are more likely to believe that such behavior is accepted by
society (Hollinger, 1989).

Research on the theory of planned behavior has consistently found that when
subjective norms toward a behavior are positive, the intention to perform that behavior
will be greater (Ajzen, 1991; Beale & Manstead, 1991; Schifter & Ajzen, 1985). This
means that when individuals perceive that important others would approve or encourage
the behavior in question, they are more likely to intend to engage in that behavior. For
example, Beck and Ajzen (1991) found that favorable subjective norms of cheating,
shoplifting, and lying were positively related to intentions to behave in these ways. That
is, individuals who thought their friends and family would approve of such behavior (or
not disapprove) would be more likely to intend to engage in these behaviors. It is
expected that these results will generalize to faking on a selection test.

Hypothesis 2: Subjective norms toward faking will be positively related to the intention

to fake.

34

 

 

 

behavlt‘r

behavior

 

suggest:
Hos ex e
predict 1'

Se l {-6 iii

Perceived behavioral control is the third variable in the theory of planned
behavior that is hypothesized to inﬂuence intentions. Ajzen (1991) suggests perceived
behavioral control refers to an individual’s perception of the ease or difﬁculty of
performing the behavior of interest. As Ajzen points out, this deﬁnition is very similar to
Bandura’s perceived self-efﬁcacy (Bandura, 1986; 1991). He suggests that the theory of
planned behavior places this construct (self-efﬁcacy) within a more general framework of
the relations among beliefs, attitudes, intentions, and behavior. Therefore, Ajzen
suggests self-efﬁcacy is very similar to, if not the same as, perceived behavioral control.
However, since the model in this paper is integrating the theory of planned behavior to
predict faking behavior, the term perceived behavioral control will be used instead of
self-efﬁcacy because this is the term used within the theory’s framework.

Research has consistently found that perceived behavioral control is positively
related to intentions such that when individuals perceive that they have control over the
behavior in question, the intention to perform the behavior is greater (Ajzen, 1991; Beale
& Manstead, 1991; Schifter & Ajzen, 1985). For example, Beck and Ajzen (1991) found
that perceptions of behavioral control were highly and positively correlated with
intentions such that those who believed they had control over cheating, lying, and
shoplifting were more likely to intend to behave in these ways. Similar results are
expected in this study, such that individuals who perceive control over faking behavior
(i.e., those who think they could fake a personality selection test) will have greater
intentions to fake.

Hypothesis 3: Perceived behavioral control of faking will be positively correlated with

the intention to fake.

35

intent-0
is. the g
the bet.
Mansre.
Ajzen. i
from 10.
Thereft:
tntentror

31036 “j

 

Intention to Fake and Faking Behavior. It has consistently been found that the
intention to engage in a behavior is related to performing the behavior in question. That
is, the greater the intention to perform a certain behavior, the greater the likelihood that
the behavior will be performed (Ajzen, 1991; Ajzen & Fishbein, 1980; Beale &
Manstead, 1991; Beck & Ajzen, 1991; Boldero, 1995; Boldero etal., 1992; Fishbein &
Ajzen, 1975; Schifter & Ajzen, 1985). Intentions have been shown to predict everything
from losing weight (Schifter & Ajzen, 1985) to stealing (Beck & Ajzen, 1991).
Therefore, in the present study it is expected that a similar relationship between the
intention to fake and faking behavior will be found.

Hypothesis 4: The intention to fake will be positively related to faking behavior such that
those with a greater intention to fake will fake to a greater extent.

Although the relationship between intentions and behavior is strong, it is far from
perfect. This indicates that there are other variables at work that are inﬂuencing this
relationship. The same is true of the relationship between intention to fake and faking
behavior. The faking literature suggests some moderators of these relationships.
Therefore, the model of faking presented in this paper integrates the theory of planned
behavior with some variables that have empirically been shown to affect faking behavior.
Speciﬁcally, it is expected that situational factors will moderate the relationships between
attitudes toward faking, subjective norms toward faking, and perceived behavioral control
and the intention to fake (see Figure 1). Additionally, ability to fake (i.e., knowledge of
the constructs being measured) will moderate the relationship between the intention to
fake and faking behavior. A detailed discussion of these moderators will now be

presented.

36

orrna)'

 

Moderators. Although the theory of planned behavior has been successful at
predicting intentions and behavior, some have suggested that the theory would explain
more variance if moderators were considered in the framework (Bagozzi, 1992).
Therefore, other predictors of the intention to fake and faking behavior will be examined
within the context of the theory of planned behavior.

Research in the faking literature has found that there are multiple situational
factors that may inﬂuence faking behavior in applied settings. These situational factors
may be situations the applicant is placed in with respect to his or her life circumstances,
or may be more directly related to the selection context (e.g., valence for a job, concern
with getting caught faking). For example, it has been suggested that when valued
outcomes are at stake, applicants will fake more (Mabe & West, 1982). Therefore, an
individual in a ﬁnancial crisis may really need the job in question such that she would
fake to ensure she gets the job. It has also been shown that the more desirable a job, the
greater the likelihood that the individual will fake responses on a selection test (Fletcher,
1990). Therefore, there is evidence that valence for a job may predict faking behavior.
In general, it has been suggested that the more desirable the outcome is (doing well on a
test), the more likely an individual is to distort his or her responses (Schlenker &
Weigold, 1992). In the present study, although valence for a speciﬁc job will not be
measured (as this is a lab study), the valence one has for the outcome of doing well on the
test will be examined. Speciﬁcally, a monetary incentive will be offered to some
participants, such that those who perform well on the test will receive money (this will be

explained in more detail below).

37

annude

ofnotd

0065 DC

thecon-

Skater
Sublet:
have a

H\

    

 

 

As Figure 1 shows, situational factors will moderate the relationship between
attitudes toward faking, subjective norms toward faking, and perceptions of behavioral
control and the intention to fake. It is expected that even if an individual has a negative
attitude toward faking (or negative subjective norm toward faking), the individual may
still intend to fake if the outcomes of doing so are very desirable, or if the consequences
of not doing so are negative (e.g., not obtaining a job). Likewise, even if an individual
does not think he or she can fake a personality test, he or she may still intend to fake if
the consequences of doing so are very positive. In the present study it is predicted that
those who are offered an incentive will have more positive valence for doing well on the
test. Based on the above discussion, the following predictions will be made:

Hypothesis 5a: Valence for doing well on the test will moderate the relationship between
attitudes toward faking and intention to fake such that those with high valence will have a
greater intention to fake, regardless of attitudes toward faking.

Hypothesis 5b: Valence for doing well on the test will moderate the relationship between
subjective norms for faking and intention to fake such that those with high valence will
have a greater intention to fake, regardless of their subjective norm of faking.

Hyp_othesis 5c: Valence for doing well on the test will moderate the relationship between
perceptions of behavioral control toward faking and the intention to fake such that those
who have high valence will have a greater intention to fake, regardless of their
perceptions of behavioral control.

Characteristics of the speciﬁc selection situation may make applicants more or
less likely to fake. For instance, it has been suggested that concern with getting caught

faking may decrease faking behavior on selection tests. Research has consistently found

38

 

vent-la?

that Mi:

, .
apph.ar

 

that in self-report measures, veriﬁable and objective items are faked less than items that
are unveriﬁable and subjective (Becker & Colquitt, 1992; Shaffer, Saunders, & Owens,
1986; Stokes et al., 1993). For example, Stokes et al. found that applicants faked
veriﬁable items less than items measuring personal preferences. Research has also found
that when applicants are told their test answers will be veriﬁed, they fake less than
applicants who are not given such information (Hough et al., 1990; Mael, 1991). It has
been suggested that this decrease in applicant faking (as measured by both lie scale scores
and mean differences from honest groups on the test) on veriﬁable and objective items
(and when told their responses will be veriﬁed) is because applicants fear that their
attempts to fake will be detected. If applicants are identiﬁed as fakers, they fear this
would result in decreasing their chances for the job (Lautenschlager, 1994). As faking is
generally done to increase one’s chances of getting a job, it seems that concern with
being caught, and thus being eliminated from the selection process, would decrease
faking behavior.

When using personality tests, it is less feasible to verify answers or to create
objective items (as compared to many biodata forms or integrity tests) because whether
or not an individual has desirable personality characteristics is being measured. A
measure of this type is inherently less veriﬁable and more subjective (Wheeler, Hamill, &
Tippins, 1996). Therefore, warning applicants that their responses may be verified (in
order to reduce faking) may not work as it will be obvious to the applicants that such
veriﬁcation would be difficult, if not impossible, to obtain. Therefore, other means have
been used to deter faking on personality tests and other less veriﬁable self-report

measures.

39

 

chore.

lllfflai.

1.,
Altlflt‘x',

1111“,}:

 

junior
hill“ 01“
l

mini:

“ﬁfe “

Research has shown that telling individuals that a lie scale is included on the test
that will identify those who are faking responses may decrease faking (Doll, 1971;
Kluger & Colella, 1993; N ias, 1972; Schrader & Osbum, 1977; Wheeler et al., 1996).
To determine if warning that faking attempts may be detected decreased faking on a test
of mental incompetence, Hiscock, Layman, and Hiscock ( 1994) administered two forced
choice tests — a knowledge test and moral reasoning test - to 105 inmates. A third of the
inmates were instructed to take the tests honestly, a second third were told to fake bad,
and the last third were told to fake bad but also that there were ways that the inmates may
be identiﬁed as fakers (e.g., by appearing excessively impaired or answering difﬁcult
items correctly and easier ones incorrectly). Results indicated that those asked to fake
without being warned scored signiﬁcantly lower on the knowledge test than those who
were told to fake but given a warning. Both groups scored lower than those responding
honestly. For the moral reasoning test, although not statistically signiﬁcant, those in the
unwarned fake condition scored lower than those in the warned fake condition.

Similarly, Nias (1972) had 262 twelve and thirteen year old children take the
junior Eysenck Personality Inventory. All children were told to take the test honestly, but
half of them were also warned that a lie scale was included in the measure to detect
individuals who may not be responding honestly. Results indicated that the children who
were warned of the presence of a lie scale scored signiﬁcantly lower on the lie scale and
had higher mean scores on psychoticism and neuroticism. Nias also examined if the
warning reduced faking more than if one just corrected scores for such responding. To
test this, those who had high lie scale scores within the group that was not warned of a lie

scale were taken out of the sample and the means for the various scales were recalculated

4O

with those peoPle excluded. Results indicated that mean scores on psychoticism and
neuroticism were still much lower compared to the means for the group that was warned
a lie scale was included on the measure. Therefore, it seems that warning may be more
effective at dealing with faking than correcting scores for faking.

Doll (1971) had 300 college graduates receiving armed service training take a
biodata instrument under instructions to respond honestly. Some of the subjects were
then told to fake to look good but to be prepared to defend their answers, another group
was told to fake to look good but to be aware that a lie scale may be included to detect
faking, and the last group was told to fake to look as good as possible. Results indicated
that subjects could increase their scores by faking and that the largest score increase was
in the “fake to look as good as possible” condition while the least amount of increase was
seen by those in the condition which instructed subjects to be aware of a lie scale.

Schrader and Osbum (1977) examined how warning that a lie scale was included
on a biodata form inﬂuenced faking. They asked participants to fake a biodata form.
However, before taking the biodata form, half of the participants were told a lie scale was
included in the measure to identify those who were giving false responses. It was found
that the group warned that a lie scale may have been included in the measure scored a
half of a standard deviation lower than those who were not warned.

Recently, Wheeler et al. (1996) replicated these results, ﬁnding applicants who
were warned a lie scale may be included on a personality measure scored lower on the lie
scale and had lower mean scores then applicants not given the warning. It is believed
that this occurs because applicants do not want to be identiﬁed as liars and thus decrease

their chances of being selected for the job (Kluger & Colella, 1993).

41

Therefore, it seems that warning that a lie scale is included in the measure may
decrease the intention to fake, even if subjective norms, attitude toward faking, and
perceptions of behavioral control are high. Therefore, the following hypotheses will be
made:

Hypothesis 6a: Warning that a lie scale is included in the measure will moderate the
relationship between attitude toward faking and the intention to fake such that those
warned that a lie scale is included in the measure will intend to fake less, regardless of
their attitude toward faking.

Hypothesis 6b: Warning that a lie scale is included in the measure will moderate the
relationship between subjective norm toward faking and the intention to fake such that
those warned that a lie scale is included in the measure will intend to fake less, regardless
of their subjective norm toward faking.

Hypothesis 6c: Warning that a lie scale is included in the measure will moderate the
relationship between perceptions of behavioral control over faking and the intention to
fake such that those warned that a lie scale is included in the measure will intend to fake
less, regardless of their perceived behavioral control.

Another moderator considered in the model of faking behavior is the ability to
fake. The ability to fake refers to how successful an individual is at increasing his or her
score through faking, when he or she chooses to fake. An individual may intend to
distort responses and attempt to do so, but not have the resources available to
successfully increase his score through distortion. That is, he may not have the ability to
distort. There are several factors that may increase the ability of an individual to fake

self-report measures. One that has been well researched, and will be examined in this

42

study, is knowledge of the construct being measured (Cunningham et al., 1994; Dwight &
Alliger, 1997; Holden & Jackson, 1981; Napier, 1979).

Goldman and Olczak ( 1976) examined the effect of knowledge about self-
actualization on faking the Personal Orientation Inventory (a measure of self-
actualization). All participants ﬁrst took the inventory under instructions to respond
honestly. Then, half of the participants were given a class that described and discussed
self-actualization. After this, one-third of the participants were told to take the inventory
again but were told to fake bad, one third were told to fake good, and the last third were
told to take the test honestly again (in each condition half of the participants had been
given information about self-actualization while the other half had not). Results
indicated that both groups (those not given information about self-actualization and those
given the information) asked to fake good were able to signiﬁcantly increase their scores
and both groups were also able to decrease their scores by a signiﬁcant amount when
asked to fake bad. However, those given knowledge of self-actualization were able to
increase their scores more than those not given this information when asked to fake good
and decreased their scores signiﬁcantly more than those not given this information when
asked to fake bad. Thus, when provided with information about the construct the test was
measuring, participants were able to fake to a greater extent.

Cunningham et al. (1994) explored the effects on test scores of providing speciﬁc
content information about the major constructs involved in the Reid Report, a popular
overt integrity test. The authors instructed all participants to respond to the measure as
they would if they were taking the test as a job applicant. However, some participants

were given information about the punitive scale (what it was trying to measure and how it

43

din: r. -~

receix : '

 

desist.

on the

 

 

measured it), some about the projective scale, and some about both dimensions of the
test, while another group was given no information about any test dimensions. In an
attempt to motivate subjects, they were told that those scoring within the top ﬁve percent
would receive ten dollars. Results indicated that those who scored lowest on the test
were subjects who received no information about what the test measured. Those given
information on a dimension of the test had signiﬁcantly higher mean scores on that
dimension as compared to those who did not receive this information. The groups
receiving information on both dimensions had the highest mean total scores (.5 standard
deviations greater than the control group). However, there were some generalized effects
on the non—targeted dimensions such that information on one dimension increased the
mean score on the other dimension slightly. These results suggest that applicants can
improve scores through faking if they are provided with information about the test.
These ﬁndings may be exaggerated given that students served as subjects. Real
applicants may give more thought (due to greater motivation) to what the test is trying to
measure and not need to get information on the nature of the test because they would be
motivated to ﬁgure it out on their own. However, such information may have helped
students who have less experience with selection tests and who are less motivated to
increase scores through faking. Nonetheless, this study indicates that the ability to ﬁgure
out what the test is measuring may be a good predictor of faking.

Dwight and Alliger (1997) replicated the ﬁndings of Cunningham et al. (1994).
They had 120 undergraduates take the Employee Integrity Index (EH) which is an overt
integrity measure developed by Ryan and Sackett (1987). Subjects were placed in one of

three conditions: fake good, honest, or a coaching condition. Those in the coaching

condr
respe.’
of the

than :1

 

the is.
those
donor

111C763

 

condition were told what integrity tests are designed to measure and taught how to
respond to such items so as to obtain a high score. Results indicated that the mean score
of the coaching condition was signiﬁcantly higher (by almost .5 standard deviation units)
than the mean score of those in the fake good and honest conditions. The mean score of
the fake good group was also signiﬁcantly greater (by over one standard deviation) than
those in the honest condition. Cunningham et al. (1994) and Dwight and Alliger (1997)
demonstrated that coaching individuals about the nature of the construct being measured
increased individuals’ ability to increase their scores.

It should be noted that no research could be found that has examined knowledge
of constructs and faking on personality measures. There is no reason why the results
pertaining to integrity tests should not also apply to personality testing. Both are self-
report measures. Also, although not empirically tested, it has been suggested that
applicants will be more successful at increasing their scores on personality measures
through faking if they can tell what the test is measuring (Anastasi & Urbina, 1997).

Therefore, it is predicted that knowledge of the constructs that the test measures
will moderate the relationship between the intention to fake and faking behavior. Again,
an individual may intend to distort responses, but not have the necessary ability to do so.
Hymthesis 7: Knowledge of constructs measured will moderate the relationship between
the intention to fake and faking behavior such that those who are given information about
the constructs measured on the personality test and have a high intention to fake the test
will increase scores more than those with similar intentions but not given this

information.

45

 

 

COUt

ling.

 

It should be noted that the present study will operationalize faking in two ways:
scores on a lie scale and difference scores (i.e., change in scores from the honest to
applicant condition). As discussed previously, one of the reasons findings may be so
discrepant in this literature is a failure to have one operationalization of faking across
studies. Some studies use a difference score to measure faking, while others use lie scale
scores. To determine if the results are consistent across different measures of faking, the
present study will explore the effects on faking and the effects of faking, using both a lie
scale and difference scores.

Faking and Validity. As discussed, one of the main reasons faking is even a
concern to those using selection tests is the potential impact faking may have on the
validity of personality measures. Therefore, in addition to testing the hypotheses above,
the extent to which faking inﬂuences the validity of the personality measure will also be
examined using GPA as the criterion measure.

Research has demonstrated that conscientiousness is a valid predictor of job
performance. Additionally, conscientiousness has been shown to predict educational
criteria such as GPA. Therefore, to determine if the validity of the test changes as a
result of faking, GPA will be correlated with conscientiousness test scores from both the
honest and the applicant conditions. If the correlations change substantially across
conditions, we can be fairly certain that faking is changing the validity. However, given
that the empirical results are mixed with regards to how faking may inﬂuence the results

of such tests, no hypotheses will be made, but the issue will be explored.

46

Method
m
Participants were 547 undergraduates from a large university in the Midwest.
This sample size results in considerable power to detect even small effects (Cohen, 1988;
1992). Table 1 contains speciﬁc information about the demographic makeup of the
participants. As the table shows, the majority of the participants were White females

under the age of 20.

Table 1. Demographic Characteristics of Study Participants

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Total Sample Size 570
Sex
Male 171 (30%)
Female 376 (66%)
Race
(’Whitc 426 (74.7%)
1 African American 45 (7.9%)
Hispanic 11 (1.9%)
Asian 40 (7.0%)
Other 21 (3.7%)
Age 19.80 (2.96)
Past Experience With
Personality Tests
Yes 405 (71.2%)
No 43 (7.5%)
Design

A 2 (incentive vs. no incentive) x 2 (warning of lie scale vs. no warning) x 2
(knowledge of constructs vs. no knowledge) between-subjects design was used. For the
ﬁrst factor, participants were asked to respond to a personality measure as if they were an

applicant for a job they would like to have. However, approximately half of the

47

participants (11 = 253) were also given an incentive such that they were told that those
who scored in the top 15% would receive $20 while the other participants (11 = 294) were
not offered this incentive. Other studies have shown that $20 is seen as very desirable to
undergraduates (McFarland & Ryan, in press; McFarland et al., 2000). Therefore, it is
reasonable to expect that those offered this incentive would have signiﬁcantly higher
valence toward doing well on the test than those not offered the incentive. The second
factor was warning of a lie scale. Approximately half of the participants (p = 285) were
told that a lie scale was included in the measure so that fakers could be identiﬁed, while
the other half (p = 262) were not told a lie scale was included in the measure. The third
factor was knowledge of constructs being measured. Half of the participants (n = 276)
were told what constructs the personality test measured and these constructs were deﬁned
for them prior to taking the personality test. The other half of participants was not given
this information (13 = 271).

It should be pointed out that in addition to taking the personality test under
instructions to respond like an applicant, participants also took the personality test under
instructions to respond honestly. These honest responses were only collected so that
difference scores could be created (these difference scores serve as one measure of faking
behavior). However, the honest and applicant conditions were counter-balanced to
determine if the order of the conditions effected results.

Pre-guestionnaire. Approximately one week before attending the experiment,
participants took the pre-experiment questionnaire (as a pre-questionnaire on the
Internet). This measure included items that measured participants’ attitudes toward

faking on selection tests, subjective norms for faking, perceptions of behavioral control

48

over faking, and intentions to fake in the future on selection tests (pro-intentions to fake).
In the study conducted by Beck and Ajzen ( 1991), participants were asked to respond to
similar items (e.g., items that asked about attitudes, subjective norms, and perceived
behavioral control toward stealing and cheating). The results indicated that individuals
were very willing to indicate what would be perceived as undesirable responses (e. g.,
positive attitudes toward theft). However, to increase the likelihood of honest responses
to this measure, participants were told that their responses would be kept completely
anonymous and that the answers they provided would not affect them in any way.
Additionally, they were encouraged to respond honestly. All items were adapted from
Beck and Ajzen who found adequate reliabilities for the above measures (all greater than
.65). See Appendix A for a reproduction of these measures.

Attitude Toward Faking. Participants indicated their attitudes toward faking a
selection test on ﬁve, 5-point semantic differential-type response scales: good-bad,
pleasant-unpleasant, foolish-wise, useful-useless, unattractive-attractive. The internal
consistency of this scale was .86.

Subjective Norms. Four items were used to measure subjective norms toward
faking. These items were answered on a 5-point scale from strongly disagree to strongly
agree. An example of an item on this scale is “Most people who are important to me
would look down on me if I lied on a selection test.” The internal consistency of this

scale was .73.

49

Perceived Behavioral Control. Three items were used to measure perceived
behavioral control over faking. These items were on a 5-point scale from strongly
disagree to strongly agree. An example of an item on this scale is “It would be easy for
me to lie on a selection test.” The internal consistency of this scale was .79.

Pre-Intentions to Fake. Generally, intention measures are given prior to the
performance of the behavior that is to be examined (Ajzen, 1991). Therefore, ten items
were used to measure intentions to fake in the future on selection tests. Items were on a
5-point scale from strongly disagree to strongly agree. An example of an item on this
scale is “I would NEVER lie on a selection test.” The internal consistency of this scale
was .89.

Personality Test. The NEO-FFI (Costa & McCrae, 1989), a measure of the ﬁve
factor model of personality, was used to assess personality. This measure is a short
version of Costa and McCrae’s (1985) NEO-PI. The five factors measured are
neuroticism, extroversion, openness to experience, agreeableness, and conscientiousness.
Each of the ﬁve factors are measured by 12 items with a Likert response format ranging
from 1 (strongly disagree) to 5 (strongly agree). The internal consistencies for each of

the scales across both the honest and applicant conditions are reported in Table 2.

50

Table 2. Reliabilities of Personality Scales Across the Honest and Applicant Conditions.

 

and Reliabilities of Difference Scores

 

 

Scale Honest Applicant Difference
Score
Neuroticism .88 .88 .61
Extroversion .84 .84 .42
Openness .76 .72 .21
Agreeableness .77 .78 .24
Conscientiousness .86 .88 .75

 

51

Post-Questionnaire A post questionnaire was handed out after the participants
took the personality test under the applicant condition. Most of the items were used as
manipulation checks (i.e., to assess if the manipulations were salient to participants).
Intentions to fake the personality measure were also assessed. See Appendix B for a
reproduction of this questionnaire.

Valence Toward Doing Well on the Test. This measure assessed how much the
participants valued the outcome of doing well on the test. It was anticipated that those
who were offered the incentive for doing well on the test would have higher valence
toward doing well on the test than those not offered the incentive. Thus, this measure
served as a manipulation check to determine if those who were offered the incentive felt
more positively toward doing well. Five items were included on this measure.
Responses to all items were made on a Likert type response scale ranging from 1
(strongly disagree) to 5 (strongly agree). An example of an item on this measure is
“doing well on the test is very desirable to me.” The internal consistency of this scale
was .90.

Concern With Being Caught Faking. This measure served as a manipulation
check. It was used to estimate how concerned individuals were that they would be
identiﬁed as a faker and whether or not they believed such faking attempts would be
detected by the scale. It was expected that those who were warned about the lie scale
being included on the test would have greater concern that their false responses would be
detected by the lie scale. This measure contained nine items. These items had a Likert

type response format ranging from 1 (strongly disagree) to 5 (strongly agree). An

52

example of an item included on this scale is, “I was concerned that the lie scale would
identify me as a liar.” The internal consistency of this scale was .77.

Knowledge of Constructs Being Measured. Eight items were used to measure
knowledge of constructs being measured. This measure served as a manipulation check
to determine if those who were given information about the constructs the personality test
was measuring were better able to determine which items related to the constructs. This
measure contained items that asked participants to indicate if they could successfully
identify the constructs that were being assessed on the test. Items were on a 5-point scale
from strongly disagree to strongly agree. An example of an item on this scale is, “Some
items measured how reliable and responsible a person is.” The internal consistency of
this scale was .80.

Post-Intention To Fake. A questionnaire assessing the extent to which
participants intended to fake the personality test under the applicant condition was
administered. As stated earlier, intention measures are generally given prior to the
performance of the behavior that is to be examined (Ajzen, 1991). This is why the
present study administered an “intentions to fake in the future” measure on the pre-
questionnaire. However, this pre-intention to fake measure was administered about a
week before the participants actually took part in the experiment. It has been suggested
that, for it to predict behavior, intentions should be measured as close to the time at which
the individual has the opportunity to perform the behavior in question (i.e., faking; Ajzen,
1991). In this study, if participants were asked their intentions to fake immediately prior
to the administration of the test, it might have inﬂuenced their responses on the test (i.e.,

some people may have never even thought about faking, but since the experimenter asked

53

about faking right before the test, some participants may have been likely to fake;
Feldman & Lynch, 1988). Also, the design is such that post-intentions must be measured
to test two of the moderation effects (valence and lie scale). That is, an intention to fake
measure must be administered after participants are offered the incentive and after they
have been told that a lie scale is included on the test. Otherwise, it cannot be determined
if these manipulations inﬂuenced intentions to fake. For these reasons, intention to fake
was also measured immediately following the test administration.

However, just because the post-intention measure was given after the test does not
mean that this is simply a measure of faking behavior on the test. Post-intention items
asked if individuals intended to fake the test. Some individuals may have intended to do
so, but may not have been successful. Therefore, other measures of faking behavior (i.e.,
difference scores and impression management scale scores) serve as the actual measures
of the extent of faking.

This post-intention measure included seven items that used a Likert type response
format ranging from 1 (strongly disagree) to 5 (strongly agree). An example of an item
on this scale is, “I intended to lie on this test.” The internal consistency of this scale was
.91.

Demographic Questionnaire. A questionnaire assessing various participant
demographics was also included and assessed factors such as race, sex, years in college,
and GPA. This measure also had participants indicate the amount of experience they
have had with personality tests. This measure also asked participants whether or not they
were thinking of any speciﬁc job for which they were applying when they took the

personality test under instructions to behave like an applicant. If so, they were asked to

54

indicate what kind of job they imagined they were applying for. See Appendix C for a
reproduction of this questionnaire.

Faking Behavior. Across the faking literature, two operationalizations of faking
are most often used: lie scale scores and difference scores. Therefore, in this study,
faking behavior was measured in these two ways. First, within the NEO-FFI, items from
Paulhus’ (1984; 1991) BIDR scale were included. This is a lie scale that is used to
measure both self-deception (unconscious distortion) and impression management
(conscious distortion). The self-deception scale contained 20 items. An example of an
item on this scale is “I have not always been honest with myself.” The internal
consistency of the self-deception scale was .74 in the honest condition and .86 in the
applicant condition.

The impression management scale also contained 20 items. Scores on the
Impression Management scale were used to measure intentional distortion (i.e., faking;
Paulhus, 1986; 1991). An example of an item on this scale is “When I hear people
talking privately, I avoid listening.” The internal consistency of the impression
management scale was .76 in the honest condition and .90 in the applicant condition.
Although both self-deception and impression management are measured, we are most
concerned with impression management scores because this is the measure used to assess
intentional distortion.

Faking behavior was also assessed by examining differences in scale scores
across the honest and applicant conditions. Several studies have used changes in test
scores from honest to applicant or fake conditions to operationalize faking (i.e., the

difference between the two administrations is considered to be due to faking; e. g., Doll,

55

1971; Ellingson et al., 1999; McDaniel et al., 1997; McFarland & Ryan, in press).
Therefore, a measure of faking behavior (in addition to the impression management
scores) was derived for each of the personality scales by subtracting the score received in
the honest condition from the score received in the applicant condition. This resulted in 5
difference scores for each individual. These difference scores were used as one measure
of faking behavior. By using the difference score we can be more conﬁdent that
intentional distortion (i.e., faking) is being isolated. Self-deception may occur in both the
honest and applicant conditions, but as described earlier, self-deception remains constant
across situations (Paulhus, 1986) and so should not inﬂuence the difference score.

Table 2 contains the reliability of the difference scores. The following formula
was used to calculate the reliability of the difference scores: rdd = (O'd2 - Ocdz) \ 0,12; where
God: = 01.2 (l - rm.) + of (1 - ram); with h representing the measure in the honest condition,
a for the applicant condition, and of representing the variance of the difference score
(Rogosa, Brandt, & Zimowski, 1982; Tisak & Smith, 1994). Only two of the difference
scores achieved adequate reliabilities (neuroticism and conscientiousness). Thus, further
analyses that include the other three difference scores (agreeableness, openness, and
extroversion), should be viewed with caution.

Procedure

Participants signed-up on-line via the Internet to participate in the experiment.
They were directed to an Internet site and could select to participate in a number of
experiments for credit. Before participants could sign-up to participate in the present
study, they were required to ﬁll out a 22 item questionnaire. This was the pre-

questionnaire containing the attitude toward faking, subjective norms toward faking,

56

perceptions of behavioral control, and pre-intention to fake measures. Therefore, on
average, this survey was taken one week before the participants came to the experiment.

The experiment session began by having participants read and sign an informed
consent form (see Appendix D for a reproduction of the consent form). Half of the
participants were randomly assigned to take these tests under the honest condition ﬁrst
and the other half were asked to complete the tests under the applicant condition first. In
the honest condition participants were given the following instructions:

Please answer the following questions on the personality test as honestly as
possible. Your answers will remain completely conﬁdential. Your answers will be used
for research purposes only, and will not be used to evaluate you in any way. The nature
of the project requires that you answer the following questions as honestly as possible so
please provide as accurate answers as you can.

Half of the participants were also told the five constructs the personality test was
created to measure. A definition of each construct was also provided (see the protocol in
Appendix E for speciﬁc instructions).

1 Before taking the personality test within the applicant condition, all participants
were given the following instructions:

When answering the questions on the personality test imagine that you are a job
applicant. Please answer as you would if you were really taking the test to get a job.

Keep in mind that your answers will be kept completely anonymous.

57

After these instructions were given (for the applicant condition), half of the
participants were given the following instructions:

To make this situation more like an applicant situation, we are offering an

incentive. Those of you who score in the top 15% on this test will receive $20.

Also prior to taking the test in either condition (depending on which
administration was given ﬁrst, and after the general instructions (honest or applicant)
were given), half of the participants were told that a lie scale was included in the
measure. These individuals were told the following:

Sometimes people try to make themselves out to be better than they really are.
Therefore, this test includes a scale to check on this. Such scales are used to identify
individuals who give inaccurate or false responses. Individuals who have a high score on
this scale (i.e., those individuals that the lie scale identiﬁes as faking responses) will be
eliminated from the selection process (and from consideration for the incentive if they are
in that condition). Please keep this fact in mind as you take this selection test.

Therefore, it was made clear to those that were in the incentive and warning
condition that those who were caught faking would not be eligible for the money.

After the necessary instructions were given, the personality test was administered.
After taking the test under the applicant condition, participants were given the Post-
Questionnaire that contained the items for the final four measures (knowledge of
construct being measured, intention to fake, valence toward doing well on the test, and
concern with being identiﬁed as a faker). This was followed by the administration of the
demographic form. After the experiment was complete, participants were given a

debrieﬁng form (see Appendix F for a reproduction of this form) and were debriefed.

58

Results
Descriptive Analyses

The equivalence of the eight between-subject conditions was examined to ensure
that differences between these groups were a result of the manipulations. AN OVAs and
chi-squares were conducted to ensure that the eight groups were equivalent in terms of
age, race, sex, GPA, past experience with personality tests, and all pre—questionnaire
scales (e.g., pre-intentions to fake). When controlling for the number of analyses that
were conducted (e. g., controlling for Type I error with a Bonferroni correction), no
signiﬁcant differences were found, indicating that the eight groups did not differ on these
variables.

Table 3 presents the means and standard deviations for each of the measures
across the two conditions and the correlations among all the measures. There were
moderate intercorrelations between the different personality scales. However, the
intercorrelations of the personality scales in the honest condition are similar to normative

information provided by the test publishers (Costa & McCrea, 1989).

59

 

on. em. 5. g. 3. 3. an... 3.. na: mu... mm: Nu. mo. 8... EN Om; EeEowmcmﬁ
5685:: .3

E. on. we. mN. an. on: mu: 5.- we; 3.. co. 5... No... bad :-._ scavenge-howd—
ow. cm. 00. cm. 5... ha: 5.. mm: mu: we. mg. 8. mos mnAm mmocmsocc2om=ou.m_
R. ma. an. mm: .. mm: a: ma: me. No. co. :6 SNm Ease—amoch/x .2
on. mw. cm: co. we. we. vo. 3. 8. mo. mmd vmdm 80:25.9
vw. an: 3.. mo.. 3.. 8.- 5. oo. ac: Sac wmdm commao>obxmd
mm. :- Nc. mo. 2. mo.- 8... war mm.w Eda Em_o:oSoZ.w
ZOHEDZOU
HmmZOE
am. am. hm. Nb. 5. we. Nor mu. Sum exam—

8 22635.05 .N.
8. 8. mm. 8. 8.- 8.- 8. 3% 388.8255

“52089—6

8. mm. 8. 8.- 8. 8. com 2:02 03.8358
8. 8. 8. 8.- 8. mg 83.8

Emacs 83:2. .4

mas/42299320

-5:

S.- 2. 8. as 2.8. 5:285
.33 cocotomxm .m
E.- S. .3 Ema
88 8.2 a... ._
muiaéeozmo

 

3 2 a Z 2 o w h o m v m N _ .Q.m :32

 

 

 

 

8:532 Ea 853:; 2:92.onch w=o8< Scum—BSD .m DBMH

60

 

 

 

5.- mo. 3. 5. 3. mo. 5.. 55. No. 55. co. 5.- 5.- 55. mod on; “560?:va
22829:— .mm
5. 5.- 5. no 3. no. 5.- 5.- 5. m5. m5. 5.- co. 5.- wwd 5N4 535805-5635
3.. 3.. pm: 3.- 3. 3.- NN. cm. 3. 3. an. 3.- 5.- 5.- :6 mod mmocmsoucoﬁmconvbm
3.- 5. 3.. me: no. 3.. 3. 3. 3. 3. 3. 5.. co. 5.- 5% 5d mmocoEmon<wN
5.- 5.- 5. Nov 3.. No. 5. 5.- 5.- 5.- 5.- No. 5.- 5.- 26 mm. 83550.3
5.- 5.- 3.- 3.. 5. ma: 3. 3. mo. 3. 5. 5.- 55. 5. 56 NN.N :o_ﬁo>obxm.mm
mo. 5. mm. 3. 5.- 3. mm: 5.. 5.- 5.- 5.- we. no. 5. ems 5w- EﬂosquZNN
mum—00m MUmemmn—HD

3. R. 3. ca. 3. 3. 5a.. 3.- 3.- 5.- 3.- 5.- 5.- 5.- 5:» mod Eva-Homage:
cognac: . _ N
R. 5. ea. 5. 3. 3. 3.. 3.- 5. m5.- 5.- 55.. 5.- 5.- 556 SN cosmooovéumdm
3. 3. 3. 3. 3. ca. 3.- 5.. 55. co. 5.- 5.- 3. co. mud Exam 805525562006—
3. 3. an. 55. m3. 5N. ma: 3.- 5.- 5.. 3.- 5. 5. 55. Now 5.3 amuse—naoohw<.w_
3. cm. 55. 5. 3.. 3. 3.- 5. 5. mo. 5. 3. 5. 5. 53.0 3.3 86530.2
3. 3. 3N. 3. 5N. Nb. ma: 5.- no. 5.- 5. oo. 5. 5.- who when 58335372
.. mm: ma: mm: but mm: 55. 3. 5.- No. 5. mo. 5. 5.- 8.5 5.2 Em_o:oSoZ.m_
ZOEHQZOU
HZ<U~AAE<

3 2 Q I o_ a w h c w v m N _ dd :32

 

 

5.253 3:532 mus-m 3333/ oﬁmﬂoﬁoﬂ wcoE< macaw—280 .m 033.

61

 

.:.I:_:.: .J......u....—\.¢ .J...

 

 

 

2. S. 3.. m... 3. 2.. a..- mm. mm. mm. an. 2. mm. a... Esowgz 838.95 .mm
a. 3.. an. S. 3. a... 8. at am. on. a. 2.. 9... 538028 .R
3. em. 3. E. a... 8. a. S. 3. S. 3. mm: 3058:5328 .8
a. E. 3.. 5.. mm. cm. 3. mm. 3. 3. 3.. 3053323 .2
a. a. 8.- 8. 8. S. 8. an. 3. 8. 3258.3
9.. a... an. mm. an. a. a. an. mu: SEBOEm .mm
6. an. a: s... «N: 2.. 3..- R. 528.32 .mm
mmmOUm muzmmmmnna
8. E. mm. a... S. S. E: “cueowgz 8535.5 .8
cm. E. on. an. 3.. mm; =o_a8%..=om .8
mm. am. 3. mm. R: 3838:2880 .2
MK. 3. 2. me: 305382? .2
9. an. 8.. 3258 .:
vw. wv... commuo>obxm .0—
3. 523232 .2
zoEmzou 555.54
mm mm ow mm «m mm mm _N om o_ w. h. o. m.

 

 

.3558 8:5on 98 moﬁmta> oEmeoEon—wdoﬁ< macaw—ebony .m 035.

62

 

.3. .v- m 32 Eon E muons—280 .802

 

2.. mo: 3. we. 3.- 3.. 3.- 8.- 8.- 3.- 3.- MN: 3.- 5.. 3.. 3.- 3.. 3. 3.- 23m 05 NE @88qu
wEom 5;? 58:00 .mm
ow. 3N. we. 5. 3. 3. mo. 8.. co. 3.- 3. mm. on. ma. 3. ea. an: au. commomm< 3023.50
.8 032305— 4m
om. an. an. 3. 3. 3. 5. 3. 3.. an. 3. mm. mu. .8. a. mu: No. “mo-H 5:03 mEoQ
6338. 85:3 .3
5. en. mm. m. mu. co. an. em: 3N. MN. em. 3. 3. mu. S...- 3: exam
2 32225.38 .3

ma<zzoﬁwmao
.58

 

Nm 3 am am mm mm cm mm vm mm mm _N ON 9 M: E 0— 3 E

 

 

3.- 3.. 3.- 3.- 3.- mu. 8. we. 5.- S. 8.. co. 8.. oo. mod 23m 05 NAm @8859
3:3 5:5 58.80 .mm

an. an. 3. 3. ca. 3.. 3.- co.- .3.- 3: 3. mo. 8.. 9.. mod 388$ 825200
no omco_Bo:M ._m

.3. N3. 3. mo. 3. 2..- 5. co. 5. mo. 8.. No. mo- 5. om.m “mo-r 5:03
wEoO c338. oo=o_a> dm
3.. 3.- NN: ac. 8.- mm. 3m. NN. 3. 3. 3.. 8. co.- 3. mmd exam 822225-5033

ma<zzoﬁmmao $8

63

 

 

2 Q I 3 m w - c n v m N _ .D.m 53—2

 

 

 

3.258 mega-no.2 Es mozmtglﬂﬁﬂwaqmcoE/x acorn—280 .m oSJaH

Manipulation Checks

The post-questionnaire administered after the participants took the personality test
under the applicant condition contained items to measure valence toward doing well on
the test, concern with being detected by a lie scale, and knowledge of what the test was

measuring. These scales were included to serve as manipulation checks. To determine if
the manipulations were effective, the differences across the two groups (e. g., those who
were warned of the lie scale versus those who were not) on each of these measures was
assessed. Three independent samples t-tests were conducted. In the ﬁrst t-test, whether
or not the participant was offered the $20 incentive served as the independent variable
(those not offered the incentive; “0;” versus those offered the incentive, “1”) and valence
toward doing well on the test served as the dependent variable. Signiﬁcant differences
between these two groups was found, with those who were offered the incentive having
signiﬁcantly greater valence toward doing well (_t (542) = 4.99, p < .001). Therefore, the
valence manipulation was successful.

Second, the lie scale manipulation was examined. Whether or not the participant
was told a lie scale was included on the test served as the independent variable and
concern with being identiﬁed as faking by a lie scale was the dependent variable. Again,
results indicate that those who were warned about the lie scale being included on the test
were signiﬁcantly more concerned that they would be identiﬁed as faking (t (542) = 9.15,
p_ < .001). Thus, this manipulation was also successful.

Finally, the knowledge manipulation was examined by assessing if those who
were provided information about what constructs the test measured felt they had greater

knowledge of what the items on the test were measuring. Results indicate that this

manipulation was not successful. There was no signiﬁcant difference in knowledge
between those who were given the information versus those who were not (1 (542) = .52,
N.S.).

As will be discussed shortly, there were signiﬁcant mean differences across the
eight manipulations. In fact, those who were given knowledge of what constructs the test
assessed tended to have the highest scores on the personality scales in the applicant
condition. Therefore, exploratory analyses were conducted to determine why the
knowledge manipulation may have failed. First, participants’ experience with personality
tests was examined and controlled for in a factorial AN OVA. Although past experience
with personality tests did predict scores on the knowledge scale, the manipulation did not
become signiﬁcant when experience was controlled for. Second, the extent to which
knowledge varied across the other manipulations was examined. It is possible that
knowledge did not work because it interacted in some way with the other manipulations.
An AN OVA was conducted so that knowledge scale scores across the eight between
subject conditions could be examined. No differences in knowledge of the constructs
were found across these eight conditions.

As can be seen in Appendix B, eight knowledge items composed the knowledge
scale that was used to assess the knowledge manipulation. Five of the items ask
respondents whether the test they just took measured speciﬁc constructs, while the other
three asked whether it was clear what the test was measuring in general (e.g., “it was
usually clear to me what personality trait an item was measuring). Therefore, it was
thought that these two different types of items may have yielded different results. For

instance, we should certainly expect differences in knowledge of the specific constructs

65

 

ortq

."
dd

[8

:1

 

being measured between those given knowledge of the constructs and those not given this
information. However, we may not ﬁnd this difference in the perception of the general
transparency of the measure. Therefore, two different knowledge scales were created,
with the ﬁve speciﬁc items making up the first scale and the three general knowledge
items composing the other scale. The mean differences on these two scales for those who
were given knowledge of what the test measures versus those who were not, was tested
with an independent samples t-test. Neither test was signiﬁcant. Therefore, even when
one simply considers speciﬁc knowledge of the constructs measured there are no
differences between those given knowledge and those who were not. Overall, it seems
that the knowledge manipulation was not successful in manipulating knowledge.
However, the knowledge manipulation did alter something as those given knowledge of
the constructs measured did score significantly different from the other groups. Perhaps
the manipulation check is not capturing the differences between these two groups because
something different was altered. That is, instead of manipulating knowledge, something
else was manipulated. Another explanation is that the manipulation did work, but the
manipulation check was ﬂawed (i.e., did not adequately measure knowledge of what the
test was measuring). Unfortunately, there is no way to determine if the manipulation was

ﬂawed or if the manipulation check was inadequate.

66

Table 4. Analysis of Variance Across the Eight Experimental Conditions

 

 

Scale E 113
HONEST CONDITION

Neuroticism 1 .68 NS.
Extroversion l .33 NS.
Openness 305* .04
Agreeableness 1 .05 NS.
Conscientiousness 3 .3 l * .04
Self-Deception l . 15 NS.
Impression Management 1.07 NS.
APPLICANT CONDITION

Neuroticism 7.61 * .09
Extroversion 482* .06
Openness 3.41 * .04
Agreeableness 405* .05
Conscientiousness l 3 .54* . 15
Self-Deception 652* .08
Impression Management 455* .06

 

Table 4. Analysis of Variance Across the Eight Experimental Conditions (cont’d)

 

 

 

Scale E If
DIFFERENCE SCORES

Neuroticism 9.88* .1 1
Extroversion 3.72* .05
Openness 233* .03
Agreeableness 344* .04
Conscientiousness 9.12* . 1 1
Self-Deception 8. 12* , . 10
Impression Management 6.21 * .08
POST-

QUESTIONNAIRE

Post Intentions 7.34* .09
Valence 4.14* .05
Knowledge 233* .03
Concern with Lie Scale 1295* .14
*p _<_ .05

Note. The df = 7 for all analyses.

68

 

mV.m OV.N 00m 00m 3V mmV and 55._ VN.0 2V 00d cod mVV 00.0 V0.V S...” .:oEom:::§

 

5:68:95
0:“ 05.5 mmd mm; w5.V V0.V 0m.~ z: _5.V VmV 2d 3; EV 00.0 $0 wV.m 808800—00
mV.0 00.0m 3.0 8.00 50.0 wmdm 55.0 303 00.0 00.00 5.0 0o.Vm 00.0 05.00 ~00 00.00 88:
8.6088800
w0.0 N000 00.0 M500 0m.0 w0.0m 0 _ .0 5.3 00V wV.0m 5.0 NV.mm 0V0 2 .00 N _ .0 00Vm 38:20:80.4
00.0 MVdN 0 _ .0 0000 V0.0 00.50 5.0 00.3 55.V mm. _ m mN.0 50.0w ~00 am. _ 0 ~00 _0. _ m :8580
$0 mVVm 05.0 mV.~m $0 05.00 00.0 55.mm :0 00.0m 00.0 3.50 «0.5 V0.00 5.0 wV.0m 5:880:05
_V.w VV.5_ 00m om.5_ 2.x 8.2 V0 00.2 55.5 00.2 V0.0 5.2 00.0 00.2 V0.0 V0 .2 8:602:02
HZ<OSAE<
05d 00d _o.m mV._ 0~.~ _0._ 00; o: 505 _0._ o. .N 0V5 mfm wV._ 8N 0V._ .:oEow:::2
:28895
00.5. cod _5._ wV._ 0w.m mad mm; Vm._ mV.~ _5._ mo; V0._ 3.5 cm; 0M: 00.. 808800.00
V0.0 0. .Nm 3.0 056m 30 50.Vm ~00 00.3 8.5 ac. _ m 3.0 55.00 N50 05.00 0m.w OVdm 3»:
-msozcﬂowcou
00.5 N0. _ 0 ~00 _ _Nm 2 .0 3.00 w0.0 ~0.~m 55.V w5.~m 05.0 00. _ m E .0 mmdm 00.0 mm. _ m mmo:o_0:8:w<
55 50d~ 50.0 3.3 G00 0m.~m 5.5 005m 3.0 3.3 Vw.0 NVdN 0—0 8.3“ 3.0 0~._m 88:80
05.0 V000 5V.5 m0._m Vw.0 0w.mm 00.0 w0.~m w5.0 00.00 3.0 3.00 00.5 _0.Nm 50.0 Vw.~m :o_m8>o:xm
V0.5 2 .om 0V0 dem 3.0 oz: 0N5 55.x. 3.x 00. _ N V0.5 0. . _ m _m.w 0:: Vm.w V0. _ N 8:808:02
Hmmzom
m5. 05.5 05. V5.N N0. 05d 50. 00d o5. 05.m 05. SN 05. m5.~ 05. 50.~ 83:25-95
Va. 305 55 _w.~ 00. 05.5 mm. 5.5 mm. 0o..~ 0a. 005 M3. co.~ 5w. m5.m 0mm
0m. and V9. oV.~ m5. 00.5 N5. 2 .N mm. Sum w5. cmd 05. 5~.N mm. mad 2:82 9,0835
V5. 3:5 00. 0~.~ 00. 55.5 _0. mod 50. E .m 00. mmd N5. 0 _ .N o0. N _ .N 9.03.0
0830.5 0050?
800-9:

Om :82 am :82 Om :82 00 :82 Om :82 Om :82 am :82 00 :82

80:85
0:: w:_:._:>> 80:85 80:85 0:: mEEES 0:: 3:0

ow0o_30:¥ 0:: w:_:::>> $02255— ow0o_Bo:M 3:0 30:85 3:0 305:3 802395— 880 30:00 0.80

 

 

8:000:50 80.05-:838m E _m 20 m88< 820380 080530 0:: 882 .0 20:5

69

 

f—vtv—.—-— Avv U» lr—r‘—~——._-p~. ‘ h....—r.-—.J o.a..j-1‘. - .-—.~A.q.|._ : .hJ. .

 

 

mm. ~m.~ 2.. mm.~ 2: Sam on. Swm co.— 9: mp. w~.~ cw. mod 8.. vwd 22225-35
an. vmd wm. 2 .m we. 3d 3. E .m we. Ed _m. m3” 3.. Ed on. mud EoocoU
xv. mod 9.. 86 on. 26 wv. 34m ow. E .v 3. mod mm. 3 .v 3. mod ego—39$
on. 99m K. mm.m mm. mud on. ova mm. ﬂan :u mm.m S. m~.m 3w. om.m 85:3
820-52
$6 3 .m 36 36 2k. :6 one Sam was now an wm.m $6 ooé Ema 3.x awe:
-maoscoﬁmcou
omw mm; Qum no; wad and 9: we; :1» 3; _m.v wm.m owé ma.m wed moﬁ mmocoﬁaoew<
vmé 3 .- _m.m mm: own 2.. EN vm. Raw 5: 36 mm. 31‘ _c. cad mm. wmoccomo
mad mo; «Ga 3. mod 3N ave mm; o _ .m KM cod mm: mm.m aqm cmh wed 56.9055
«ﬁn SN- wwé w _ .m- mw.w Eye. 86 ohm- 5.x 3.x- 2 .m em.m mod adm- 5.x 3.x- 832.832
may—00m
muzmmmmma
2:28:—

28 maﬁa? 03:82: «52:85 98 wEEmB 23 3:0

032305— 98 mica? 035.322 once—Bocx 2:0 95:85 3.5 9:555 among—326— 9.80 3950 28m

 

 

 

 

 

. 32550 252280 “cubsméooiom E _m 05 383‘ mcocﬂgn 28.85 was 9802 .m 03$.

70

dozﬂgn 2.85%

8.03 2: .3 @636 was 22:38 anzmma 2: E :38 2: Bot c2228 :25: 05 E :88 05 wczoﬂﬁzm 3 622228
22> moat. Seam .2an ESQ cm a so @2885 Son 203 EoEowmcaE 5685:: can cosmoooqhom an: 35 co 8.58 kiwi
$586,: $2? :wE 5:5 038 tog: ESQ 3c a 5.3 383mm =w 203 35308 3:28th 05. .me n m mug—mam =m 8m .202

 

 

 

 

EoEommENE
wv. So. wed- nod 3N n _ .m 91 568:3:
3‘. So. :6- Ed mad mm.m :4 Samoan—£00m
3. So. mic ﬂ - No.0 Rdm _ _s an. _ m 3052222850
ow. So. Ed _ - Rum Ream 2 .o modm amuse—smoew/x
<2 .3 a. _- 3m 3.8 mg 8.8 32:25
Vm. _oo. 51:- $6 aﬁvm who omdm co§o>obxm
me. :5. 3.2 $5 3.2 2.x _ _dm EﬂouoSoZ
u m 02?; mm. :82 dd :32
29.50200 HZ<USAE< 205.5200 ram—ZOE 8382

 

 

 

< new “mono: 2: 383‘ £802 ES :50 283 m2 Sam 3me .c 2an

 

71

mava*

 

 

 

 

Q. *~_. mac—am Eaaoh _obcoU _EEZEQm 338th

no. No. mac—an— EaBo... mESZ 2583mm

5 _. *Nm. mac—mm Enid. 203:2
*oog _ no. om. wcoccooczmom

\. _. *m _. magnu— Ewaoo. _obcoU _SEZEQm 338:5

E. *2. was—mn— Eaaoo. 2502 358.3%

mm. *mm. wcimm @338. 09:32
*ww.m _N hm. or. £2325..on

m M m m m .252

 

 

 

.75—SD _Eogmcom Simeon cam

 

.mEEZ 0363.96 £20333. so 902505-59“ was -05 5me cm mom _w:< :o_mw2mom .b DEMH

2
7

Table 8. Post Intentions Regessed onto Attitude Tmrd Faking, Subjective Norms,

 

Perceived Behavioral Control, Warning of a lie scale. and Valence.

 

 

 

Model 2 at ﬁat _If .E A 32 A BZ_F
Entr Entry
1/
STEP 1 .140 1527*
Attitude Toward Faking 23* .17
Subjective Norm .02 .02
Percieved Behavioral .12* .12
Control
Valence .09 .28
Warning of A Lie Scale -.47* -.26
STEP2 .145 7.11* .005 .414
Attitude*Valence .02 .03
Subjective Norm*Valence .06 -.08
PBC*Va1ence .03 .05
Attiude*Warning .01 .01
Subjective .07 .09
Norm*Warning
PBC*Warning -.15 -.25

 

73

AN OVAs were also conducted to determine if there were mean differences across
the eight conditions for each of the personality measures, the difference scores, and the
post-questionnaire measures. Results are presented in Table 4 . As this Table shows, the
manipulations had little effect on honest responses. However, there were signiﬁcant
differences across the manipulations for all the personality scales in the applicant
condition, for the difference scores, and for all post-questionnaire measures. The effect
sizes associated with these differences are also presented in Table 4. The effects were
small to moderate. These results provide support that when asked to respond honestly,
the respondents did follow these instructions. That is, if participants were really
responding honestly, then it should not matter what condition they were in (there should
be no mean differences across the eight groups when responding honesty). We found
that the manipulations only affected two of the seven scales (openness and
contentiousness) within the honest condition, and these effects were small (effect sizes
were .04) and smaller than effects found in the applicant condition. However, within the
applicant condition, the manipulations had effects on all scores and these effects were
typically stronger. The means and standard deviations across the eight conditions for
both the honest and applicant condition are presented in Table 5.

Past research has found that applicants score about a half a standard deviation
greater than participants who respond to personality measures honestly (Hough et al.,
1990). Therefore, to determine if the applicant condition was an effective manipulation
(i.e., really made participants behave like applicants), paired samples t-tests were
conducted for each of the personality scales comparing the honest responses to the

applicant responses. The results of this analysis are presented in Table 6. There were

74

signiﬁcant differences between the honest and applicant personality test scores for all but
the Openness scale. For the remaining six scales (the four other personality scales and
the two social desirability scales), participants scored signiﬁcantly higher in the applicant
condition. Additionally, the effect sizes in the last column on the right of Table 6 show
that participants scored, on average, a half of a standard deviation greater than in the
honest condition. Therefore, the mean differences between the honest and applicant
responses that were observed in this study mimic those found using real applicants. This
lends support that the applicant manipulation was effective.

It should be pointed out that the self—deception mean increased signiﬁcantly in the
applicant condition. This is surprising because Paulhus (1986) suggests that self-
deception is a measure of how a person sees oneself and therefore should not change
across experimental manipulations. However, as Table 6 shows, self-deception increased
in the applicant condition as much as impression management did (which is a measure of
conscious distortion). Therefore, these results cast doubt on the construct validity of this
social desirability measure.

Preliminary Analyses

In the applicant condition, participants were told to imagine that they were taking
the personality test as an applicant applying for a job they really wanted. Therefore, they
were not given any information about a speciﬁc job for which they were applying.
However, it is possible that some participants had a speciﬁc job in mind while they took
the personality test in the applicant condition and this may have influenced results. To
test this, on the demographic form, participants were asked to indicate if they were

thinking of any particular job as they took the personality test in the applicant condition.

75

If they indicated they were thinking of a job, they were asked to indicate which job they
were imagining applying for. Results indicated that 135 of the participants were thinking
of a specific job when they took the personality test as an applicant. However, the type
of job they thought of varied widely. Salesperson was the job most frequently mentioned
(11 = 25). The next most frequent was a “business” job (3 = 18; no mention was made
about what kind of business). This was followed by people indicating they imagined they
were a manager (11 = 13). The rest of the jobs imagined were indicated by less than 5
participants. Clearly, it would be very difficult to conduct analyses within these job
categories considering the small sample sizes. Therefore, several independent samples t-
tests were conducted to determine if those who thought of a speciﬁc job scored
differently on the personality measures (or on any of the post-questionnaire measure)
than those who did not. After correcting for Type I error using a Bonferroni correction,
no signiﬁcant differences were found between those who were thinking of a speciﬁc job
and those who did not on any of the measures.

Some studies have found that order effects exist in faking research such that those
who are asked to take the test honestly before the applicant or fake administration will be
able to increase their scores more through faking (Klein & Owens, 1965). To test for
such order effects, 18 independent samples t-tests were performed (one for each of the
personality and social desirability scales across both the honest and applicant conditions,
and for each of the four post-questionnaire measures) with a Bonferroni correction. The
scale score for the test in question served as the dependent variable and the order of
administration served as the independent variable. Results indicate that only two scales,

conscientiousness from the applicant condition and valence from the post-questionnaire,

76

had order effects. The effects were such that those who were given the tests under the
honest condition ﬁrst had higher conscientiousness scores and greater valence for doing
well on the test. This is consistent with past research. Further analyses demonstrate that
the effect size for order was small (less than a quarter of a standard deviation). However,
to be conservative, tests that include these two scales will be conducted both with and
without order controlled. The effect of controlling for order will only be reported when it
signiﬁcantly alters results.

Hypothesis Tests

Hypothesis I predicted that attitude toward faking would be positively correlated
with the intention to fake such that those who had more positive attitudes toward faking
would have a greater intention to fake. As Table 3 shows, the correlation between
attitude toward faking and the pre-intention to fake (the measure administered via
Internet one week prior to the experiment) was .72 (p < .001), while the correlation with
the post-intention measure (the one administered after the test) was .24 (p < .001).
Therefore, support was found for Hypothesis 1.

Hypothesis 2 was also supported, as it was found that those who had more
positive subjective norms toward faking were more likely to intend to fake. As Table 3
shows, the correlation between subjective norms and the pre-intention to fake is .57 (p <
.001), and .17 (p < .001) with post-intentions. Thus, again we see that the correlation is

stronger with pre-intentions.

77

Support was also found for Hypothesis 3, which predicted that those who have
greater perceived behavioral control over faking would also be more likely to intend to
fake. Again, as Table 3 shows, the correlation between pre-intentions to fake and post-
intentions to fake was .54 (p < .001) and .22 (p < .001) respectively.

It is not surprising that attitude toward faking, subjective norms, and perceived
behavioral control correlated more highly with the pre-intention measure. It has been
suggested that the closer the attitude is measured to the intention measure, the stronger
the relationship will be (Fishbein & Jaccard, 1973; Ajzen & Madden, 1986). It may not
simply be the passage of time itself that attenuates the relationship between intentions
and these variables, but exposure to new information or other events that may alter
attitudes, subjective norms, perceived behavioral control, or intentions (Ajzen, 1991;
Marcey & Brown, 1983). In fact, two manipulations were used to alter the relationship
between intentions to fake and these three variables (valence and warning of lie scale). It
should also be noted that the high correlation between these three variables and pre-
intentions to fake is most likely not a result of method bias, as these three variables do not
correlate as strongly with each other as they do with pre-intentions.

However, it should also be pointed out that the pre- and post-intention measures
also differ in referent. That is, the pre-intention measure refers to tests in general,
whereas the post-intention measure asks about faking on a speciﬁc test. While the pre-
intention measure asked participants if they intended to fake on a selection test some time
in the future, the post-intention measure asked if participants intended to fake the
personality test they just took. This might explain some of the differences between

results for the two measures. The correlations with the post-intention measures may be

78

weaker, not only because it was measured at a different time from the independent
variables (i.e., attitudes, subjective norms, and perceived behavioral control), but also
because it is of a more speciﬁc nature.

To gain a deeper understanding of the way in which attitudes toward faking,
subjective norms, and perceived behavioral control predict the intention to fake, these
variables were examined using regression analyses. This type of analysis can allow for a
deeper understanding of the relationships between these variables and intentions to fake.
One regression analysis was run with pre-intentions as the dependent variable and
another with post-intentions to fake as the dependent variable. Attitudes toward faking,
subjective norms toward faking, and perceived behavioral control over faking served as
the independent variables in both analyses. Results of these analyses are presented in
Table 7. As the table shows, all variables explained a signiﬁcant amount of the variance
in pre-intentions beyond that explained by the other two variables. Taken together,
attitudes, subjective norms, and perceive behavioral control explained 57% of the
variance in pre-intentions to fake. When post-intentions serve as the dependent variable
the explained variance is only 7%. Additionally, subjective norms does not signiﬁcantly
predict post-intentions to fake beyond that explained by attitudes.

The results of the regression analyses obtained [when pre-intentions was the
dependent variable are very similar to estimates that have been found in other studies
testing the theory of planned behavior (see Ajzen, 1991 for a review). For example, on
average, research has found that attitudes, subjective norms, and perceived behavioral
control explain 55% of the variance in intentions to fake. However, when post-intentions

were used as the dependent variable, the relationships decrease substantially and are

79

much lower than has been found with other tests of the theory. Again, this is likely
because post-intentions were measured well after attitudes, subjective norms, and
perceived behavioral control were assessed. In other tests of the theory of planned
behavior, these variables are assessed at the same time as intentions. Therefore, it is
expected that relationships observed with the pre-intention measure would be more
consistent with past research.

Hypothesis 4 predicted that the intention to fake would be positively correlated
with faking behavior. This hypothesis was tested in two ways. First, both pre and post-
intentions to fake were correlated with the difference scores for the ﬁve personality
scales. Keep in mind that the difference scores serve as an operationalization of faking
because any change in scores from the honest to the applicant condition may be
attributable to faking. It was found that both pre- and post—intentions were signiﬁcantly
correlated with the difference score for four of the ﬁve scales (see Table 3). Neither of
the intention measures was significantly related to the openness difference score (which
is not unexpected given the extremely low alpha for this difference score). Therefore, we
ﬁnd that the intention to fake does explain a signiﬁcant amount of variance in the
difference scores. However, the pre—intention measure explains only one to four percent
of the variance in the difference scores, while post-intentions explain fourteen to twenty-
six percent. Again, this may be explained by the closer proximity of the post-intention

measure with the administration of the tests.

80

Hypothesis 4 was also tested by correlating pre- and post-intentions with social
desirability scores in the applicant condition. The results, which are presented in Table 3,
indicate that both social desirability scale scores (self-deception and impression
management scores from the applicant condition) signiﬁcantly related to pre and post-
intentions to fake the test. However, unexpectedly, pre-intentions to fake correlate
negatively with both social desirability scales, while post-intentions correlate positively.
This means that those who had high pre-intentions to fake a selection test got lower
scores on the social desirability scale in the applicant condition, but those who indicated
that they did try to fake the test received high social desirability scores. Another
interesting ﬁnding is that both self-deception and impression management correlated
similarly with both intention measures. Paulhus (1986) suggests that self-deception
assesses unconscious distortion, while impression management measures intentional
faking. Therefore, one would expect to ﬁnd no relationship between self-deception and
intention to fake, because self-deception is unconscious and unintentional. Again, these
results do not support the construct validity of the two components of Paulhus’ (1986)
social desirability scale.

As shown in Table 9, support was not found for Hypotheses 5a, 5b, and So. It
was expected that valance toward performing well on the test would moderate the
relationship between the three main effects on the intention to fake, such that those who
had high valence toward performing well (those offered the $20 incentive) would be
more likely to intend to fake, regardless of their attitudes, subjective norms, or perceived
behavioral control over faking. First, a moderated regression was conducted with post-

intentions serving as the dependent variable and attitude toward faking, the dichotomous

81

measure of valence (0 versus 1), and their interaction as the independent variables. Keep
in mind that post-intentions had to serve as the dependent variable (as opposed to pre-
intentions) because the pre-intention measure was administered before the manipulations.
Results indicate that only the main effect of attitude was signiﬁcant. Thus, valence
toward performing well on the test did not signiﬁcantly predict the intention to fake
beyond that predicted by attitude toward faking. Second, the interaction between
subjective norms and valence toward doing well on the test was examined in a similar
manner. Once again, only the main effect for subjective norms was signiﬁcant. Finally,
the interaction between perceived behavioral control and valence was examined and
again only the main effect of perceived behavioral control was signiﬁcant.

Next, Hypotheses 6a, 6b, and 6c were examined. These hypotheses predicted an
interaction between the three main effects (attitudes, subjective norms, and perceived
behavioral control) and concern with being detected as a liar by the lie scale on the
intention to fake. Three different moderated regressions were conducted with post-
intentions to fake serving as the dependent variable and the main effect of interest,
whether or not the participant was told about the lie scale being included on the measure,
and their interaction as the independent variables. As Table 10 shows, none of the three
interactions were signiﬁcant. However, for all three analyses, the two main effects were
signiﬁcant. Results were such that those with more positive attitudes toward faking,
positive subjective norms toward faking, and high perceived behavioral control, were
more likely to intend to fake. However, those who were warned about the lie scale being

included on the test were less likely to fake.

82

The same analyses described above with valence toward doing well on the test
and concern with being identiﬁed as lying by a lie scale, were run again with the
manipulation checks (the scales which measured valence and concern on the post-
questionnaire) in place of the dichotomous variable for valence and concern (0 or 1).
Although the results were slightly larger (more variance was explained) when these
continuous scales were used as independent variables, these analyses did not result in
significant interactions.

A regression was also run with post—intentions regressed onto attitude toward
faking, subjective norms, perceived behavioral control, warning of a lie scale, and
valence. Table ?? shows the results of this analysis. As can be seen, warning was the
strongest predictor of post-intentions to fake. Those who were warned that a lie scale
was included on the test were much less likely to have tried to fake the test. Attitude
toward faking, and perceived behavioral control were also predictors of post-intentions.
Additionally, subjective norm did not predict post-intentions. None of the interactions
were significant.

It should be noted that a paired samples t-test was conducting comparing pre-
intentions to post-intentions. It was found that pre-intentions to fake were signiﬁcantly
higher than post-intentions. This is not surprising given that those who were warned that
there was a lie scale included on the test were much less likely to intend to fake. Thus,

the warning contributed to lowering post-intentions to fake.

83

Table 9. Regression Analyses Regressing Post-Intentions on Attitudes, Subjective

Norms Perceived Behavioral Control,:and Valence.

 

 

 

 

 

 

 

Model Q g E E A 3} A R2f_
Attitude and Valence
STEP 1 .063 1573*
Attitude Toward Faking .33* .25
Valence .07 .04
STEP 2 .063 1050* .000 11
Attitude * Valence .04 .05
Subjective Norms and
Valence
STEP 1 .032 7.75*
Subjective Norms Toward
Faking .20* .17
Valence .05 .03
STEP 2 .032 5.18* .000 .08
Subjective Norm *
Valence .03 -.O4
Perceived Be:h_2_1viora_l
Control and Valence
STEP 1 .051 1264*
Perceived Behavioral
Control .22* .22
Valence .08 .04
STEP 2 .051 8.46* .000 .128
Perceived Behavioral
Control * Valence .03 .05
* p_ < .05

Note. The “Valence” variable was coded 0/1 (“0” being no incentive was offered; “1”

being a $20 incentive was offered to those who scored in the top 15% on the test).

84

Table 10. Reggssion Analyses with Post Intentions Regressed on Attitudes, Subjective

Norms Perceived Behavioral Control, and Warning of a Lie Scale.

 

 

2

Model b L

__ F A32 ARZF

Attitude and Warning

STEP 1 .126 3402*
Attitude Toward Faking .33* .24
Warning -.47* -.26

STEP 2 .126 2268* .000 .12
Attitude * Warning -.04 -.05

Subjective Norms and
Valence

STEP 1 .099 2593*
Subjective Norms Toward
Faking .20* .18
Warning -.48* -.26

STEP 2 .099 1726* .000 .04
Subjective Norm *
Warning -.02 -.03

Perceived Behavioral
Control and Valence

STEPl .116 3085*
Perceived Behavioral
Control .22* .22
Warning -.48* -.26

STEP 2 .118 21 .09* .003 1.50

Perceived Behavioral
Control * Warning -.11 -.18

a:
p < .05
Note. The “Warning” variable was coded 0/1 (“0” indicates the participant was not

6619’

warned that a lie scale was included on the test; indicates the participant was told a

lie scale was included on the measure).

85

Table 11. Regmssion Analyses Regressing Pre-Intentions and Knowledge on Faking

 

 

 

 

 

Billet—vim
Model E _R_2 E A 32 A R2 F
Impression Management
STEP 1 .025 6.14*
Pre-Intentions -.89* -. 16
Knowledge -.30 -.04
STEP 2 .029 458* .003 1.47
Pre-Intentions*Knowledge .63 .22
Self-Deception
STEP 1 .010 2.31
Pre-Intentions -.50* -.09
Knowledge .04 -.00
STEP 2 .010 1.54 .000 .003
Pre-Intentions*Knowledge .02 .05
Neuroticism Difference Score
STEP 1 .029 694*
Pre-Intentions -.83 —.08
Knowledge 208* . 14
STEP 2 .029 463* .000 .038
Pre-Intentions*Knowledge . 18 .03
Extraversion Difference Score
STEP 1 .013 3.1 1*
Pre-Intentions .74* .1 1
Knowledge -.30 -.03
STEP 2 .014 2.24 .001 .495
Pre-Intentions*Knowledge .44 . 13
Omnness Difference Score
STEP 1 .005 1.17
Pre-Intentions -. 1 7 -.03
Knowledge -.55 -.O7
STEP 2 .015 235* .010 469*
Pre-Intentions*Knowledge l . 16* .40
Agreeableness Difference Score
STEP 1 .023 564*
Pre-Intentions .94* . 1 5
Knowledge -.38 -.04
Step 2 .023 3.7 5* .000 .006
Pre-intentions*Knowledge -.04 -.01

 

86

Table 11. RegLession Analyses Regressing Pre-Intentions and Knowledge on Faking

 

Behavior (cont’d).

 

 

 

Model .12 L _3 E A32 ARZF
Conscientiousness Difference
Score
STEP] .060 1504*
Pre-Intentions 1 .96* .20
Knowledge -1.92* -.14
STEP2 .060 10.01* .000 .010
Pre-Intentions*Knowledge -.08 -.02
p< .05

Note. The “Knowledge” variable was coded 0/1 (“0” indicates the participant was not
given information about the constructs assessed by the test; “1” indicates the participant
was given information about the constructs the test assesses). Impression management

and self-deception scores are from the applicant condition.

87

Table 12. Reggession Analyses Reggessing Post—Intentions and Knowledge on Faking

Behavior

 

Model

IQ“

Impression Management Score

from Applicant Condition
STEP 1

Post-Intentions

Knowledge
STEP 2

Post-Intentions* Knowledge
Self-Deception
STEP 1

Post-Intentions .89*

Knowledge .10
STEP 2

Post-Intentions*Knowledge
Neuroticism Difference Score
STEP 1

.97*
-.02

-.15

-.14

Post-Intentions
Knowledge
STEP 2
Post-Intentions*Knowledge
Extpzwersion Difference Score
STEP 1
Post-Intentions 203*
Knowledge .12
STEP 2 .
Post-Intentions*Knowledge .70
anness Difference Score
STEP 1
Post-Intentions .02
Knowledge -.33
STEP 2
Post-Intentions*Knowledge
Agreeableness Difference Score
STEP 1
Post-Intentions
Knowledge
STEP 2
Post—intentions*Knowledge

1.41*

150*

-.s4*

214*
-.O9

-.37

—4.18*

A

.22
-.OO

-.05

.22

.01

-.05

—.53
.10

.29

.38

.01

.19

-.01

-.04

—.28

.424

—.01

-.11

88

R2

.049

.049

.051

.051

.303

.312

.143

.147

.002

.010

.181

.182

E

13.79*

924*

14.17*

956*

1 16.49

8092*

4463*

3072*

.439

1.87

59.17

3973*

A132

.009

.009

.001

ARzF

.17

.18

7.10*

2.63

4.72*

.881

Table 12. Regression Analyses Regressing Post—Intentions and Knowledge on Faking

Behavior (cont’d).

 

 

 

Model a J .33 .13 A 132 ARE
Conscientiousness Difference
Score
STEP] .270 98.97
Post-Intentions 390* .50
Knowledge -1 .48* -. 10
STEP2 .277 6831* .007 536*
Post-Intentions*Knowledge -1 .33* -.26

 

p<.05

Note. The “Knowledge” variable was coded 0/1 (“0” indicates the participant was not

given information about the constructs assessed by the test; “1” indicates the participant

was given information about the constructs the test assesses). Impression management

and self-deception scores are from the applicant condition.

89

 

Table 13. Regression Analyses Reggessing Faking Behavior on Pre-Intentions to Page

and the Knowledge Scale

 

 

 

 

 

 

 

Model 2 g _R_2 E A32 A33
13
Impression Management
STEP 1 .134 36.61
Pre-Intentions -.71 * -. 12
Knowledge 302* .33
STEP 2 .135 24.48* .001 .322
Pre-Intentions* Knowledge -.33 -.24
Self-Deception
STEP 1 .10 26.21*
Pre-Intentions .37 .22
Knowledge 243* .30
STEP 2 .102 17.92* .002 1.28
Pre-Intentions*Know1edge -.60 -.50
Neuroticism Difference Score
STEP 1 .022 534*
Pre-Intentions - l .04* -. 10
Knowledge -1.89* -. 12
STEP 2 .026 4.11* .003 1.65
Pre-Intentions*Knowledge -1 .39 —.59
Extgtversion Difference Score
STEP 1 .016 394*
Pre-Intentions .79* . 12
Knowledge .72 .07
STEP 2 .017 269* .000 .225
Pre-Intentions*Knowledge .35 .22
Openness Difference Score
STEP 1 .002 .45
Pre-Intentions -. 19 -.03
Knowledge -.30 —.03
STEP 2 .004 .56 .002 .80
Pre-Intentions*Knowledge -.57 —.42
Agreeableness Difference Score
STEP 1 .025 599*
Pre-Intentions .97* . 1 5
Knowledge .70 .07
STEP 2 .026 4.12* .001 .405
Pre-intentions*Know1edge .45 .30

 

9O

Table 13. Regression Analyses Regressing Faking Behavior on Pre-Intentions to Fake
and the Knowledge Scale (cont’d).

 

 

 

 

Model 2 g 33 _13 A _32 A R2 1:
Conscientiousness Difference
Sm
STEP 1 .046 11.44*
Pre—Intentions 2.08* .2 1
Knowledge 1 .09 .07
STEP 2 .049 8.02* .002 1.17
Pre-Intentions*Knowledge l . 14 .49
* p < .05

Note. The “Knowledge” variable was the continuous scale score from the post-
questionnaire. Impression management and self-deception scores are from the applicant

condition.

91

Table 14. Regmssion Analyses Regressing Faking Behavior on Post-Intentions to Fake

 

 

 

 

 

 

 

 

ﬂd Knowlegg
Model 12 g E E A g A RZF
Impression Management
STEP 1 .15 4919*
Post-Intentions .88* .20
Knowledge 295* .33
STEP 2 .176 3868* .023 15.09*
Post-Intentions* Knowledge 125* 1.24
Self-Deception
STEP 1 .14 4551*
Post-Intentions .85 * .2 1
Knowledge 245* .3 1
STEP 2 .155 3319* .012 7.46*
Post-Intentions*Knowledge .80* .88
Neuroticism Difference Score
STEP 1 .301 115.73*
Post-Intentions —4.2 1 * -.54
Knowledge -1.33* -.085
STEP 2 .301 77.15* .000 .294
Post-Intentions*Knowledge -.28 -. 16
Extraversion Difference Score
STEP 1 .145 4580*
Post-Intentions 2.02* .38
Knowledge .43 .04
STEP 2 .149 3136* .004 2.267
Post-Intentions*Know1edge .59 .49
Openness Difference Score
STEP 1 .001 .296
Post-Intentions -.01 -.00
Knowledge -.29 -.03
STEP 2 .004 .70 .003 1.51
Post-Intentions*Knowledge -.44 -.43
Agreeableness Difference Score
STEP 1 .182 5998*
Post-Intentions 2. 14* .42
Knowledge .38 .04
STEP 2 .185 4068* .003 1.89
Post-intentions*Knowledge .50 .44

 

92

Table 14. Regression Analyses Regressing Faking Behavior on Post-htentionsro Fake
and Knowledge (cont’d).

 

 

 

 

Model Q g .33 E A g A R2 F
Conscientiousness Difference
Score
STEP 1 .261 94.78*
Post-Intentions 396* .5 1
Knowledge .3 1 .02
STEP 2 .263 6384* .002 1.715
Post-Intentions*Know1edge .70 .40
* p < .05

Note. The “Knowledge” variable was the continuous scale score from the post—
questionnaire. Impression management and self—deception scores are from the applicant

condition.

93

Table 15. AN OVA Predicting Faking Behavior.

 

 

Source SS df MS F p q
Impression Management 499.41 7 71.34 4.55 .001 .056
Warning 465.87 1 465.87 29.69 .001 .053
Valence 33.58 1 33.58 2.14 n.s. .004
Knowledge .19 1 .19 .01 n.s. .000
Warning x Valence 1.49 1 1.49 .10 n.s. .000
Warning x Knowledge .28 ‘ 1 .28 .02 n.s. .000
Valence x Knowledge 7.03 1 7.03 .45 n.s. .001
Warning x Valence x .00 1 .00 .00 n.s. .000
Knowledge
Error 8346.93 532 15.69
Total 13593.00 540
Neuroticism Difference 3224.16 7 460.60 9.88 .001 .1 15
Score
Warning 2512.24 1 2512.2 53.88 .001 .092
4
Valence 17.69 1 17.69 .38 n.s. .001
Knowledge 415.85 1 415.85 8.92 .01 .016
Warning x Valence 23.31 1 23.31 0.50 n.s. .001
Warning x Knowledge 167.30 1 167.30 3.59 n.s. .007
Valence x Knowledge 6.70 1 6.70 .14 n.s. .000
Warning x Valence x 1.28 1 1.28 .03 n.s. .000
Knowledge
Error 24852.42 533 46.63
Total 41752.00 541
EXtroversion Difference 613.93 7 87.70 3.72 .001 .047
Score
Warning 520.33 1 520.33 22.06 .001 .040
Valence 7.86 1 7.86 .33 n.s. .001
Knowledge .07 1 .07 .00 n.s. .000
Warning x Valence 23.16 1 23.16 .98 n.s. .002
Warning x Knowledge 1.71 1 1.71 .07 n.s. .000
Valence x Knowledge 9.54 1 9.54 .40 n.s. .001
Warning x Valence x 63.01 1 63.01 2.67 n.s. .005
Knowledge
Error 12569.26 533 23.58
Total 15885.00 541
Openness Difference 271.91 7 38.84 3.23 .05 .030
Score
Warning 35.39 1 35.39 2.12 n.s. .004
Valence 1.97 l 1.97 .12 n.s. .000

94

Table 15. ANOVA Predicting Faking Behavior (Continued).

Source
Knowledge
Warning x Valence
Warning x Knowledge
Valence x Knowledge
Warning x Valence x
Knowledge
Error
Total

Agreeableness Difference
Score

Warning

Valence

Knowledge

Warning x Valence

Warning x Knowledge

Valence x Knowledge

Warning x Valence x
Knowledge

Error

Total

Conscientious. Difference
Score

Warning

Valence

Knowledge

Warning x Valence

Warning x Knowledge

Valence x Knowledge

Warning x Valence x
Knowledge

Error

Total

SS
15.46
79.27
99.47
19.31
46.75

8896.07
9192.00

503.96

452.88
13.27
7.77
1.59
2.01
.30
19.37

11149.34
15653.00
2988.66

2062.59

3.35
485.49
17.67
283.26
12.88
39.44

24793.52
41700.00

df

533
541

533
541
7

l
1
1
1
1
1
1

533
541

95

MS
15.46
79.27
99.47
19.31
46.75

16.69

71.99

452.88
13.27
7.77
1.59
2.01
.30
19.37

20.92

426.95

2062.5
9

3.35
485.49
17.67
283.26
12.88
39.44

46.52

.93
4.75
5.96
1.16
2.80

3.44

21.65
.63
.37
.08
.10
.01
.93

9.18
44.34

.07
10.44
.38
6.09
.28
.85

P
n.s.

.05
.05
n.s.
n.s.

.001

.001

[1.8.
11.8.
n.s.
n.s.
11.3.
[1.8.

.001
.001

ns

.001

n.s.

.01
n.s.
n.s.

T]

.002
.009
.01 1
.002
.005

.043

.039
.001
.001

.000

.002

.108
.077

.019
.001
.011
.001
.002

 

 

No Warning

    

W arning

 

 

—O— No Valence

+ Valence

 

 

Figure 2. Interaction Between Warning and Valence When Predicting Faking as

Measured by the Openness Difference Score.

96

 

 

 

 

 

 

 

1 .2
1
0.8
0.6
0.4 +No Knowledge
()2 +Knowledge
0
-O.2
-0.4

 

No Warning Warning

 

 

 

Figure 3. Interaction Between Warning and Knowledge on Faking as Operationalized as

the Openness Difference Score.

97

 

 

+No Knowledge
+Know1edge

 

 

 

 

No Warning Warning

 

 

 

Figure 4. Interaction Between Warning and Knowledge on Faking as Operationalized by

the Conscientiousness Difference Score.

98

Next, Hypothesis 7 was tested by examining if knowledge of the constructs being
measured would moderate the relationship between the intention to fake and faking
behavior. The way in which knowledge affected the relationship between both pre- and
post-intentions and faking behavior was examined to determine if the time of the
administration of the intention measure would alter results. Additionally, both
operationalizations of faking were examined. Remember, faking behavior was assessed
in two ways: with scores on the lie scale, and with difference scores. Although self-
deception is not a measure of conscious distortion, both lie scale measures were used in
these analyses so that these relationships could be explored.

First, a moderated regression was run, with impression management scores (from
the applicant condition) as the dependent variable. The dichotomous knowledge variable
(coded as 0 or 1), pre-intentions, and their interaction served as independent variables.
As Table 11 shows, only pre-intentions signiﬁcantly predicted impression management
scale scores such that individuals with greater pre-intentions to fake scored lower on the
impression management scale.

Six more moderated regressions (using self-deception scores and each of the five
difference scores) were run with self-deception or the difference score of interest as the
dependent variable. The dichotomous knowledge variable (coded 0 or 1), pre-intentions
to fake, and their interaction, served as the independent variables. Table 11 shows that a
signiﬁcant interaction between pre-intentions and knowledge was only found when
predicting faking on the openness scale. This interaction supports the prediction that

those who had high intentions to fake and were given knowledge of what the test

99

measured would be able to increase their scores more than those who had similar
intentions but were not given this information.

The same seven analyses (ﬁve with the difference scores and with each of the two
social desirability scales) were conducted with post-intentions as an independent variable
and are shown in Table 12. These analyses yielded three signiﬁcant interactions. Faking
on neuroticism, openness, and conscientiousness (as indicated by the difference scores
for each of these scales) was signiﬁcantly predicted by the post-intention to fake x
knowledge interaction in the predicted direction. This means that those who had a high
intention to fake, and were given knowledge about what the personality test measured,
were able to increase their scores more through faking. these three measures than those
who had high intentions to fake but were not given this information. Therefore, the
interactions that were signiﬁcant support the hypothesis that knowledge would increase
the relationship between intentions to fake and faking behavior by making those with
greater intentions to fake more able to do so.

As described earlier, the knowledge manipulation was not successful (those
provided with knowledge of the constructs scored similar to those on the knowledge
scale, which was the manipulation check, as those who were not given this information).
Therefore, the same analyses just discussed that used the dichotomous knowledge
variable (received information versus did not receive information) were tested again with
the knowledge scale (the measure of how much participants knew about the personality
measure). Therefore, fourteen more regressions were run with the knowledge measure
from the post-questionnaire as one of the independent variables. Table 13 includes the

result of these analyses when the pre-intention to fake was used, while Table 19 displays

100

the results for post-intentions. None of the analyses that included the pre-intention
measure as an independent variable yielded signiﬁcant interactions. However, the post-
intentions by knowledge interaction did signiﬁcantly predict faking when impression
management scores and self-deception scores were used to assess faking behavior. As
expected, those who had high intentions to fake and had greater knowledge of what the
test assessed, were better able to fake compared to those who had high intentions to fake
and were not given the information. Therefore, again we ﬁnd that providing knowledge
of the constructs assessed by the test increases the relationship between the intention to
fake and faking behavior.

To assess the contribution of each of the manipulations (i.e., warning, knowledge,
and valance) on faking behavior, six AN OVAs were conducted with each of the measures
of faking as the dependent variable and warning, valence, and knowledge as the
independent variables. Results are presented in Table 15. As the Table shows, warning
had the largest inﬂuence on faking with those who were warned about the lie scale being
included on the test faking a lesser extent. Knowledge also predicted faking as measured
by the conscientiousness difference score. Across the six ANOVAs only three
interactions were signiﬁcant. Figures 2, 3, and 4 plot these interactions. As can be seen,
those who were offered an incentive and warned that a lie scale was included on the
measure faked much less than those who were warned, but were not offered the incentive.
Perhaps this occurred because those who were offered the incentive were told that they
would not receive the money if they were detected faking. Therefore, the individuals
who were warned realized the consequences for faking and therefore, faked to a lesser

extent. Figure 4 shows the interaction between warning and knowledge on the openness

101

difference score. As the Figure shows, those who were not given a warning and had no
knowledge of what the test was measuring were more likely to fake. However, keep in
mind that the openness difference score had a very low reliability. Also, there were no
signiﬁcant differences between the honest and applicant responses to the openness scale.
So, it appears that this scale was not faked. Therefore, these results are difﬁcult to
interpret. Figure 5 shows the interaction between warning and knowledge when the
conscientiousness difference score was used as the dependent variable. Those who were
given more knowledge of the construct being assessed faked more when there was no
warning. However the differences between those given knowledge and those that were
not are much smaller when a warning about the lie scale was given. Overall, these results
show that warning had the strongest effect on faking and valence generally had no effect.
Exploratogr Analyses

It has been suggested that faking may reduce the criterion-related validity of
personality measures (e. g., Dunnette et al., 1962; McDaniel et al., 1997; Paulhus, 1986).
Therefore, the relationship between the personality scales and performance criteria were
examined across the two experimental conditions (i.e., honest and applicant) to determine
how faking may inﬂuence the criterion-related validity of the measure.

As discussed earlier, a number of studies have found that conscientiousness is a
valid predictor of job performance (Barrick & Mount, 1991; Hough et al., 1990).
Research has also demonstrated that conscientiousness is a useful predictor of academic
achievement (Dollinger & Orf, 1991). Therefore, in the present study, participants’

Grade Point Average (GPA) served as the performance criterion.

102

The relationship between these criteria and the Conscientiousness scale across the
two administration conditions was explored. As Table 3 shows, conscientiousness
signiﬁcantly predicted GPA and this correlation was larger for the honest condition (_r =
.15) than the applicant condition (r = .10). However, further analyses revealed that this
difference was not signiﬁcantly different across the two manipulations. This results is
consistent with the preponderance of evidence suggests that faking has a small, negligible
effect on validity (e.g., Christiansen et al., 1994). However, just because the difference is
not signiﬁcant in the present study, a .05 difference in the validity estimate across the
honest and applicant condition may have practical effects. That is, selection decisions
may be altered such that the best candidate for a job (as indicated by the test under honest
conditions), will not be selected.

Summag

The theory of planned behavior predicted pre- and post-intentions to fake and
faking behavior. These ﬁndings are similar to other studies that have used the theory of
planned behavior in other contexts. However, valence toward doing well on the test and
warning of a lie scale did not act as moderators of the relationships between attitudes,
subjective norms, and perceived behavioral control and the intention to fake. Despite
this, concern with a lie scale being included on the test did have a main effect on the
intention to fake, with those who were more concerned about the lie scale having lower
intentions to fake the test. Valence toward doing well on the test had no such main
effect.

As the theory of planned behavior predicts, the intention to fake related to faking

behavior, however, there was little support for knowledge of the constructs the test

103

measured as a moderator of this relationship. Support for knowledge as a main effect on
faking behavior was also inconsistent.

Finally, faking did not adversely affect the relationship between conscientiousness
and GPA. Although the correlation between these two variables was higher in the honest
condition, no signiﬁcant differences were found.

Discussion

Several empirical studies have been conducted examining the antecedents of
faking behavior and the effects of faking on tests results, however, there is still no clear
consensus as to the causes of faking or what its effects are. This may be because no
attempt has been made to integrate the various empirical ﬁndings into any kind of
theoretical framework. Instead, each study stands in isolation and does not build upon
other research or explore possible interactions between different variables. This is the
ﬁrst study that attempts to examine multiple antecedents and effects on faking behavior
simultaneously. It is also the ﬁrst to use a theoretical framework to do so.

The present study was conducted to integrate empirical research regarding faking
on personality tests into a model of faking behavior. The theory of planned behavior
(Ajzen & Madden, 1986) was used as the framework to integrate these empirical
ﬁndings. Support was found for the theory of planned behavior in predicting faking
behavior. However, evidence for the moderating effects of valence, warning of a lie

scale, and knowledge was generally not found.

104

 

 

SSEom
maﬁa

 

@2332
wEom 83350
.«0 owuo_3o:M-

85 2 5:5

 

 

 

 

 

 

 

   
 

 

 

35
OH :Omucoucm

 

 

.mmEmcocﬂom cotoamsm .m 2:3”—

/1

 

 

28m
24 .8 wEEwB

 

 

 

963
@8309 one—333x44

 

 

 

ESZ 0383.6

 

 

 

 

_obcoU E5323—
wozoenﬁ

 

 

105

Figure 2 contains the relationships that were supported in the present study. The author is
not suggesting that this is a revised model of faking behavior, but simply wants to
summarize the ﬁndings in a clear way. The ﬁgure shows that, as predicted, attitude
toward faking, subjective norms toward faking, and perceived behavioral control all
related to the intention to fake. However, warning of a lie scale, which was predicted to
moderate these relationships, actually was found to have a main effect on the intention to
fake. Notice that valence toward doing well on the test is not included in the model
because it was not found to be a moderator or have a signiﬁcant main effect on the
intention to fake. As expected, the intention to fake related to faking behavior. The
arrows from knowledge of the constructs assessed to faking behavior and the line that
designates knowledge as a moderator between the intention to fake and faking behavior
are broken arrows because there is inconsistent evidence for these two relationships. The
results changed across operationalizations of faking and pre- and post-intentions. Below,
each of these findings will be discussed in more detail.
Inﬂuences on Intention to Fake

The theory of planned behavior predicts that if one’s attitude toward faking is
positive, subjective norms toward faking are positive (i.e., the individual perceives that
signiﬁcant others would approve of the behavior in question), and it is believed that one
could fake the test, the intention to fake will be higher than when these conditions are not
present. This study supports these predictions. Therefore, the theory of planned behavior
was successful at predicting the intention to fake the personality measure.

The correlations observed between attitudes, subjective norms, and perceived

behavioral control and the intention to fake are similar to those observed in other tests of

106

the theory. For example, Ajzen (1991) reviewed studies that had tested the theory of
planned behavior to explain everything from quitting smoking to theft. The correlations
observed between these three variables and the intention to fake in the present study are
very similar to those reported in the studies reviewed by Ajzen. Additionally, across the
studies that Ajzen reviewed, when attitudes, subjective norms, and perceived behavioral
control were used to predict intentions, the average multiple correlation was .71. A
regression was run with the current data with pre-intentions regressed onto attitudes,
subjective norms, and perceived behavioral control (see Table 7). The multiple
correlation was .76; very similar to the estimate provided by Ajzen (when post-intentions
were used the multiple correlation dropped to .26). Therefore, it seems that the results of
this study are consistent with other tests of the theory of planned behavior.

Another expected ﬁnding is that attitudes, subjective norms, and perceived
behavioral control related more strongly to pre-intentions than to post-intentions to fake.
This is to be expected because as time between the measurement of attitudes, subjective
norms, and perceived behavioral control and intentions increases, the weaker the
relationship will be (Ajzen, 1991). Events may occur over time to alter these
relationships. In the present study pre-intentions were measured the same time as
attitudes, subjective norms, and perceived behavioral control. However, generally one
week separated the administration of these measures and the post-intention measure.
These results are likely not a result of method bias because attitudes, subjective norms,
and perceived behavioral control did not correlate as strongly with each other as they did

with pre-intentions to fake.

107

Another reason pre- and post-intentions may have yielded different relationships
with the independent variables is that they use different referents. That is, while the pre-
intention measure refers to faking in general, the post-intention measure refers to faking
on the speciﬁc personality test the participants were administered. Therefore, post-
intentions were more speciﬁc than pre-intentions and this speciﬁcity may have altered
results.

Bagozzi (1992) suggests that the theory of planned behavior will not be successful
at explaining substantial variance in behavior until other variables are considered. He
suggests examining moderators of the relationships proposed by the theory. Two
moderators of the relationship between attitudes, subjective norms, and perceived
behavioral control of faking and intentions to fake, were examined. It was expected that
valence toward perforrnin g well on the test would moderate these relationships. Valence
toward doing well on the test was manipulated by offering $20 to participants who scored
in the top 15 % on the test. It was found that those who were offered the incentive had
higher valence toward doing well on the test. However, no support was found for the
moderation effect of valence. In fact, valence did not have any effect on either pre- or
post-intentions to fake (i.e., the main effects were not signiﬁcant). This ﬁnding
contradicts past research. Research suggests that the more desirable a job is the more
likely an individual is to fake to increase the chances of being hired (Schlenker, 1980;
Schlenker & Weigold, 1992).

The failure to ﬁnd that valence moderated these relationships may have occurred
because the manipulation was not successful at simulating the valence one would feel if

one really desired a job. That is, even though the manipulation did result in greater

108

valence toward doing well on the test, it did not create the same valence that would be
felt if an applicant really wanted the job for which he or she was applying. Given the
research that has proposed links between job desirability and faking (Fletcher, 1990;
Schlenker, 1980) and that this relationship makes rational sense, future research should
attempt to replicate these results using real applicants or using stronger valence
manipulations (e. g., more money).

It was also expected that warning applicants that a lie scale was included on the
personality measure would moderate the relationships between attitudes toward faking,
subjective norms, and perceived behavioral control, and the intention to fake. Although
no support was found for this relationship, a main effect for warning was found. Those
who were warned that a lie scale was included on the test were less likely to intend to
fake than those who were not warned about the lie scale. This ﬁnding is consistent with
past research that has found that when individuals are warned that a lie scale is included
in a measure, they fake less on the test (Doll, 1971; Kluger & Colella, 1993; Schrader &
Osbum, 1977; Wheeler et al., 1996).

So, overall it seems that the variables in the theory of planned behavior (attitudes
toward faking, subjective norms, and perceived behavioral control toward faking)
predicted the intention to fake. However, proposed moderators of these relationships
were not supported. Although Bagozzi (1991) has suggested that moderators would be
useful for increasing the explanatory power of the theory of planned behavior, Ajzen
(1991) has suggested that such moderators are generally not successful and does not
encourage their inclusion in tests of the theory (Ajzen & Madden, 1986). He cites

evidence that tests of the theory that have included such moderators are generally not

109

supported and the inclusion of them creates a less parsimonious model. The present
study conﬁrms this position as no moderating effects were found.
Predictors of Faking Behavior

Most faking research has used scores on a lie scale to operationalize faking.
Other studies have used difference scores. Differences in the operationalization and
measurement of faking across studies may account for some of the discrepant ﬁndings
regarding the relationship between faking and test validity. Therefore, the present study
Operationalized faking in two common ways by using a lie scale and difference scores.
By using these two methods, it could be determined if different operationalizations of
faking converged (i.e., drew the same conclusions).

The theory of planned behavior predicts that the intention to perform a behavior
will relate to the performance of that behavior. Therefore, it was anticipated that those
with greater intentions to fake the personality measure would be more likely to fake the
test. Two measures were used to operationalize faking: the ﬁve difference scores and
scores on the impression management and self-deception scales. Therefore, the
relationship between the intention to fake and faking behavior was examined across all
seven measures. Overall, ﬁndings were mixed and some were unexpected. It was found
that the difference scores were related to both pre— and post-intentions to fake the
measure, such that those with higher intentions to fake, faked a greater extent. However,
as would be expected, given their closeness in proximity (in terms of when they were
administered), the post-intention to fake was more strongly related to faking behavior.
Unexpectedly, impression management scores (and self-deception scores) were

negatively related to pre-intentions to fake. This suggests that those who indicated they

110

 

 

wou.
Ho“
imp!

inter

wh

adr

the
the
in:

for

 

Sir

 

would fake selection tests (on the pre-questionnaire) faked less on the personality test.
However, it was also found that post-intentions were strongly and positively related to
impression management and self-deception scores. That is, according to the post-
intention measure, those who indicated after they took the test that they had intended to
fake it had higher scores on the impression management scale. Again it was found that
post-intentions was more strongly related to impression management scores than were
pre-intentions. This is likely because of the time of measurement. The post-intention
measure was administered shortly after the participant took the test (with the impression
management scale included), whereas the pre-intention measure was administered about
one week before the impression management scale. Alternatively, it may be the case that
post-intentions, because it was more speciﬁc in referent, was a better predictor of faking
on the personality test. Keep in mind that the pre-intention measure was more general in
that it asked questions pertaining to the likelihood of faking a selection test in the future,
while the post-questionnaire asked about faking on the personality test that was just
administered.

Knowledge of constructs measured by the test was examined as a moderator of
the relationship between the intention to fake and faking behavior. Limited support for
the moderating effect of knowledge was found. However, these ﬁndings should be
interpreted in light of the fact that the knowledge manipulation was not successful. It was
found that those who were given knowledge of what constructs the test assessed scored
similarly on the knowledge scale as those who were not given this information.
Exploratory analyses did not yield any reason for this lack of effect. For example, the

data was searched for possible outliers, but none were found. The knowledge scale was

111

also split into speciﬁc and general knowledge scales and still no effects were found
between the two groups. Additionally, past experience with personality tests was
controlled for and this did not inﬂuence the manipulation. However, there were
signiﬁcant differences in personality scale scores between those given knowledge and
those not given knowledge. Therefore, it seems that although the knowledge
manipulation may have failed, providing this information to participants manipulated
something.

There are at least two possible explanations for this ﬁnding. First, maybe the
knowledge scale did not adequately tap knowledge of the constructs measured. That is,
the manipulation check was ﬂawed, rather than the manipulation itself. Therefore, the
manipulation actually worked, but the manipulation check was not adequate to measure
this. Perhaps knowledge should have been measured in a different way. For example,
after taking the test participants could have been asked to read a list of the personality
items measuring the same construct (grouped together one right after the other) and try to
explain what the group of items measured. This could have been done for each construct.
This may have provided us with a more precise measure of knowledge.

Another alternative is that the knowledge manipulation was not successful at
manipulating knowledge of the constructs assessed, but instead inﬂuenced variables that
were not directly examined in the present study but do have an impact on faking
behavior. For example, research has shown that perceptions of face validity indirectly
inﬂuences test performance, with those who have greater face validity being more
motivated and therefore performing better on the test (Chan, Schmitt, DeShon, Clause,

Delbridge, 1997). It may be that those who were given knowledge of what constructs the

112

test assessed felt the test was more face valid (because they had a better understanding of
why the test may relate to performance) and therefore, these individuals scored higher
because they were more motivated. Clearly, this is speculation. The author is using this
as an illustration to demonstrate how the manipulation may have effected faking behavior
(as it clearly did) without affecting the knowledge participants had about the test.

Although unexpected in the present study, the results pertaining to knowledge are
consistent with the work of others who have examined ability as a moderator in the
theory of planned behavior. In the present study it was anticipated that those who were
given knowledge of the constructs assessed by the test would be better able to fake. It
was expected that knowledge would increase the strength of the relationship between the
intention to fake and faking behavior. Ajzen and Madden (1986) suggested that ability
has rarely been found to operate as a moderator in the theory of planned behavior.
Although the results of the present study conﬁrm Ajzen and Madden’s position on
moderators, it would be inappropriate to assume that knowledge (or ability) does not
moderate the relationship between the intention to fake and faking behavior because the
manipulation may have been unsuccessful.

Also of interest is the relationship between the two social desirability scales:
impression management and self-deception. Paulhus (1984; 1986) suggests that self-
deception is the component of socially desirable responding that reﬂects the tendency of
individuals to see themselves more positively than is actually the case. Psychologically
healthy individuals tend to see themselves more positively than they really are, so self-
deception taps just how positive our opinions are of ourselves. Therefore, self-deception

is related to true personality variance. Paulhus suggests that self-deception should not

113

 

 

0':

vi

C).

bt

tie

it

re

in

 

change across experimental manipulations. Unlike impression management, self-
deception should remain constant because different situations should not alter how we
view ourselves. Impression management, on the other hand, refers to conscious attempts
to distort the way one is perceived. Paulhus proposes that impression management
should change across experimental manipulations because people will want to be
perceived in different ways across situations and will manipulate the impressions they
make to ensure this occurs.

The present results do not support the construct validity of the self-deception or
impression management scales. Self-deception and impression management were
signiﬁcantly correlated in the honest condition (.56) and the applicant condition (.71).
Additionally, the difference scores for these two variables were signiﬁcantly correlated at
.67. These are substantial correlations indicating that there is no discriminant validity
evidence for these two scales.

Additionally, self-deception scores changed across experimental manipulations.
Not only did they change across the eight conditions (as demonstrated by the ANOVAs),
but the relationship between self-deception and other variables changed across the honest
and applicant conditions. Additionally, although not always the same strength, self-
deception always changed in a similar manner to impression management. For example,
it was found that, like impression management, self-deception was negatively related to
pre-intentions to fake, but positively related to post-intentions to fake (but the
relationship between self-deception and intentions was slightly smaller than it was with

impression management).

114

These results also provide little construct validity evidence for the impression
management scale. Although impression management scores were signiﬁcantly
correlated with the difference scores, they were not as high as one would expect.
However, it is not clear if the impression management scale failed to tap faking, if the
difference scores were not successful at isolating faking, or both. Either way, it seems
these two indicators of faking (impression management and difference scores) were
measuring different things as they often resulted in different conclusions (they did not
correlate similarly with other variables).

It should be noted that this study is not the ﬁrst to ﬁnd that self-deception and
impression management are highly related and relate in similar ways to other constructs.
In their study with truck driver applicants, Barrick and Mount (1996) found that the
correlation between self-deception and impression management was above .40 in both
samples examined. Additionally, they found that the two scales had similar relationships
with the personality scales.

As discussed earlier, most of the research that has explored faking and the effects
of faking on test results have used scores on social desirability scales as the
operationalization of faking. These studies have rarely questioned the construct validity
of these measures. However, the present study provides very little support that these
measures are construct valid. This lack of construct validity may explain why ﬁndings
within the faking literature are so discrepant. For example, different social desirability
scales are used across these studies and these scales may tap self-deception and
impression management to different extents. Clearly, more time and effort needs to be

spent providing construct validity for these social desirability measures and developing

115

more distinct and valid measures of these two constructs (assuming these two constructs
exist and are distinct). It is critical that a construct valid measure of faking behavior be
developed. Only when we can precisely measure faking will we be able to determine
when faking actually occurs and how it inﬂuences the results of non-cognitive selection
tests.

Effect of Faking on Criterion-Related Validity

Much of the research in the faking literature has examined how applicant faking
inﬂuences the validity of non-cognitive or personality measures (Dunnette et al., 1962;
Hough et al., 1990; Ones et al., 1996). Generally, the way in which the relationship
between the test scores and performance criteria changes as a result of faking is
examined. In the present study, there is substantial evidence that faking was more
prevalent in the applicant condition. As the paired samples t—tests in Table 6
demonstrate, participants did signiﬁcantly increase their scores when asked to behave like
applicants versus when they were asked to answer honestly. Additionally, impression
management scores were also signiﬁcantly higher in the applicant condition. Therefore,
it was examined if this faking had an adverse effect on the criterion-related validity of the
personality measure.

Results of this analysis demonstrated that the relationship between
conscientiousness scores and GPA was not signiﬁcantly different across the honest and
applicant condition. Therefore, it seems that faking did not have an effect on the validity
of conscientiousness in this study. However, there was still a difference of .05 between
the validity of conscientiousness in the honest and in the applicant condition. Such a

difference might still have a practical effect in selection contexts.

116

Implications and Directions for Future Research

The results of this study have several implications for the theory of planned
behavior and the faking literature. First, this study extends the theory of planned
behavior to a selection context. As far as the author is aware, there is no published study
that has applied this theory to explain faking behavior in selection settings. Thus, the
theory itself was supported. Attitudes, subjective norms, and perceived behavioral
control all related to the intention to fake and the intention to fake related to faking
behavior. However, the proposed moderators of these relationships were generally
unsuccessful. As discussed, Ajzen and Madden (1986) suggest that such moderators
should not be included because it destroys the parsimony of the theory and little support
has been provided for moderators when they are included.

It is true that the inclusion of moderators and other variables will make the theory
more complex. The fact remains, however, that although the theory is a good predictor of
faking behavior there is still considerable variance that it does not account for.
Therefore, other variables should be examined and integrated in the theory as both
moderators and direct effects on the intention to fake and faking behavior. Researchers
should explore the literature to ﬁnd other variables that have been linked to faking
behavior and determine if these variables can successfully be integrated into the theory to
better predict faking.

Second, this study demonstrates that the theory of planned behavior can
successfully predict faking behavior and may therefore be useful for altering faking
behavior. For example, as the theory of planned behavior predicts, positive attitudes,

subjective norms, and perceived behavioral control toward faking were all related to the

117

intention to fake. In turn, the intention to fake related to faking behavior. We should
now use this information and determine how we can decrease applicant faking by altering
attitudes, subjective norms, and perceived behavioral control over faking. For instance,
we may be able to create negative attitudes toward faking by describing to applicants
(prior to the test administration) why faking may be bad, unpleasant, or foolish. By doing
this we may be able to create negative (or at least less positive) attitudes toward faking,
which should decrease the intention to fake. The test administrator can explain, prior to
administering the personality test (or non-cognitive measure), reasons why it is unwise to
fake on such tests. For example, the administrator can explain to applicants that
performance on such tests is linked to job performance and that individuals who score
well on these tests do well on the job they were hired for. When individuals lie on these
tests, they may be selected but may not perform well on the job and may then not stay
with the organization. Such an explanation may alter an applicant’s attitude toward
faking by making the applicant recognize that it may be unwise and unnecessary to fake
the test.

A similar procedure could be followed to alter subjective norms. This could be
done by explaining to applicants that lying on such tests is typically not perceived well by
others and is generally frowned upon. Additionally, perceptions of behavioral control
may be altered by explaining to applicants that lying on the test is probably not as easy as
one may think and that people are likely to decrease their scores by faking. Applicants
may also be told that a lie scale is included in the measure. Although not examined in the
present study, such warnings may result in decreased perceived behavioral control over

faking, thereby resulting in lower intentions to fake.

118

However, it would probably be easier to alter attitudes toward faking and
perceived behavioral control over faking than subjective norms. After all, the test
administrator would have a difﬁcult time convincing an applicant that the people who
matter to him or her would frown upon faking if the applicant knows this is not the case.
Future research should attempt to explore if such procedures would result in a signiﬁcant
decrease in faking. Such questions can easily be addressed in both the ﬁeld and in lab
studies.

Third, these ﬁndings may help explain some of the discrepancies in the literature
with regards to how faking inﬂuences the results of personality tests and, more generally,
non-cognitive measures. In this study, two operationalizations of faking were used. The
personality difference scores served as one measure of faking and scores on the social
desirability scales served as another. These two operationalizations are the most
frequently used in the literature. It was found that these two measures of faking rarely
converged. For example, it was found that impression management scores correlated
negatively with pre-intentions to fake, while the difference scores had positive
relationships.

If the results using these two operationalizations of faking differed so drastically
within the same study, it is not surprising that the literature is plagued by such
discrepancies when different procedures and measures of faking are used across these
studies. Future research should attempt to determine why these differences exist. Earlier
it was discussed how there is little construct validity evidence for the social desirability
scale. Perhaps the reason the impression management scores and difference scores led to

different conclusions is because the impression management scale is not isolating

119

conscious distortion. Future work should work towards isolating the reasons for these
differences in results.

These results also demonstrate that Paulhus’ self-deception scale and impression
management scale overlap a great deal. Although the impression management scale was
generally more predictive of faking behavior, the two scales were highly correlated and
related to other variables in a similar manner (e. g., both were negatively correlated with
pre-intentions to fake, but positively correlated with post-intentions). Therefore, it is not
clear if these two constructs are distinct. Very little work has been done determining the
construct validity of these scales. Future research should attempt to determine the
construct validity of these measures.

An alternative measure of faking that may more precisely measure faking is
response latencies. It has been found that when individuals fake their responses, it takes
longer for them to respond (Dwight & Alli ger, 1997). However, when they respond
honestly it is found that response times decrease substantially. Future work should
examine how response latencies can be used to assess faking behavior. Using this
measure of faking we may be able to better assess the antecedents and consequences of
faking behavior.

Finally, these results have implications for other non-cognitive selection
measures. Generally, results that have been obtained in the faking literature, using
personality tests, have been replicated with other non-cognitive measures (e.g., biodata
forms, integrity tests) and vice versa. For instance, it has been found that when asked to
fake, people can successfully increase their scores on these measures by about a standard

deviation (Dunnette et al., 1962; McDaniel et al., 1996; McFarland & Ryan, in press).

120

Therefore, research should be directed toward applying the theory of planned behavior to
explaining faking on these other tests and also determining if the procedures described
above for deterring faking (e. g., altering attitudes) decrease faking on these other non-
cognitive measures.

Limitations

Although this is the first study to integrate empirical ﬁndings within the faking
literature into a theoretical framework, the are some limitations of this study that should
be pointed out. First, some may question the generalizability of these results because this
study was conducted in the lab rather than the ﬁeld. However, a lab study was used to
provide the necessary control needed to test these relationships. To increase the external
validity of the ﬁndings, participants were asked to pretend that they were applicants for a
job they really wanted and some were even offered an incentive that was based on their
performance. However, it is suggested that these ﬁndings be replicated within an actual
ﬁeld setting.

Second, three of the difference scores had very low reliability. This likely
occurred because, for some of the scales, there was little variance in the extent to which
peOple faked or there was little change in scores from the honest to the applicant
condition. For instance, scores on the openness scale were not signiﬁcantly different
across the two administrations. Therefore, it seems that people either could not or simply
did not fake this measure. Since there was little change across the two administrations
for this variable, the difference score was not reliable.

Third, this study examined a number of different variables that effect faking in

order to gain a deeper understanding of the antecedents and effects of faking behavior.

121

Clearly, however, all the variables that effect faking could not be considered within the
context of this study. Therefore, future research should explore how the theory of
planned behavior may be combined with other variables that have been found to
inﬂuence faking behavior. Attempts should be made to examine as many of these
different effects on faking behavior simultaneously. Only then will we have a true
understanding of what the antecedents of faking behavior are and the effect faking has on
test results.

Finally, it is important to note that the model of faking behavior presented in this
study does not address the process of faking. That is, the thought processes and
strategies people use when faking was not examined. Future work should be done to
identifying the process of faking behavior. Such work may utilize a verbal protocol to
assess the thoughts that respondents have when they fake on a non-cognitive test. This
assessment method may also give us some ideas about how to best measure faking
behavior.

Conclusion

The present study used the theory of planned behavior to integrate and organize
empirical ﬁndings from the faking literature. Although the hypothesized moderators
were generally not supported, it was found that the theory itself did predict faking
behavior. Therefore, it seems that the theory of planned behavior can be used to predict
when faking is likely to occur and also suggests a number of ways to deal with faking
(e. g., altering attitudes toward faking). Future attempts should use this theory to predict
faking behavior while simultaneously considering other empirical ﬁndings from the

faking literature that may have direct effects on the intention to fake, faking behavior, or

122

may moderate the relationships in the theory. Only when we consider the multiple
inﬂuences on faking behavior will we be able to truly understand when faking is likely to
occur, how it may inﬂuence the results of our tests, and how we may deal with faking on
personality measures to ensure that these tests provide useful information from which

selection decisions can be based.

123

REFERENCES

124

REFERENCES

Abrahams, N. M., Neumann, I. & Githens, W. H. (1971). Faking vocational
interests: Simulated versus real life motivation. Personnel Psycholagv. 24, 5-12.

Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In J. Kuhl &

J. Beckman (Eds), Action-Control: From Cognition to Behavior (pp. 11-39).
Heidelberg: Springer.

Ajzen, I. (1991). Theory of planned behavior. Organizational BehavioL and Human_
Decision Processes 50, 179-211.

 

Ajzen, I. & Fishbein, M. (1980). Understanding attitudes and predicting social behavior.
Englewood Cliffs, NJ: Prentice-Hall.

Ajzen, I. & Madden, T. J. (1986). Prediction of goal-directed behavior:
Attitudes, intentions, and perceived behavioral control. Journal of Experimental Social
Psychology, 22, 453-474.

Anastasi, A., & Urbina, S. (1997). Psychological Testing. Upper Saddle River,
NJ: Prentice Hall.

Bagozzi, R. P. (1992). The self-regulation of attitudes, intentions, and behavior.
Social PsychologyOuarterly. 55, 178-204.

Bandura, A. (1986). Social foundations of thought and action: A social cogpitive
theory. Englewood Cliffs, NJ: Prentice-Hall.

Barrick, M. R., & Mount, M. K. (1991). The big ﬁve personality dimensions and
job performance: A meta analysis. Personnel Psychology, 44, 1-26.

Barrick, M. R., & Mount, M. K. (1996). Effects of impression management and
self-deception on the predictive validity of personality constructs. J oumal of Applied
Psychology. 81 , 261-272.

Bass, B. M. (1957). Faking by sales applicants of a forced choice personality
inventory. Journal of Applied Psychology, 41, 403-404.

Beale, D. A., & Manstead, A. S. R. (1991). Predicting mothers’ intentions to
limit frequency of infants’ sugar intake: Testing the theory of planned behavior. J oumal

of Applied Social Psychology, 21, 409-431.

 

Beck, L., & Ajzen, I. (1991). Predicting dishonest actions using the theory of
planned behavior. Journal of Research in Personality. 25, 285-301.

125

Becker, T. E. & Colquitt, A. L. (1992). Potential versus actual faking of a biodata
form: An analysis along several dimensions of item type. Personnel Psychology. 45.
389-406.

Bobko, P., Roth, P. L., & Potosky, D. (1999). Derivation and implications of a
meta-analytic matrix incorporating cognitive ability, alternative predictors, and job
performance. Personnel Psychology. 52, 561-589.

Boldero, J. (1995). The prediction of household recycling of newspapers: The
role of attitudes, intentions, and situational factors. Journal of Applied Social
Psychology, 25, 440-462.

Boldero, J ., Moore, S., & Rosenthal, D. (1992). Intention, context, and safe sex:
Australian adolescents’ responses to aids. J oumal of Applied Social Psychology, 22,
1373-1396.

Borislow, B. (1958). The Edwards Personal Preference Schedule (EPPS) and
fakability. Journal of Applied Psychology, 42, 22—27.

Cellar, D. F., Miller, M. L., Doverspike, D. D., & Klawsky, J. D. (1996).
Comparison of factor structures and criterion-related validity coefﬁcients for two

measures of personality based on the ﬁve factor model. Journal of Applied Psychology,
§_1_ , 694-704.

Chan, D., Schmitt, N ., DeShon, R. P., Clause, C., & Delbridge, K. (1997).
Reactions to cognitive ability tests: The relationship between race, test performance, face
validity perceptions, and test-taking motivation. Journal of Applied Psychology, 82, 300-
310.

Christiansen, N. D., Gofﬁn, R. D., Johnston, N. G., & Rothstein, M. G. (1994).
Correcting the 16PF for faking: Effects on criterion-related validity and individual hiring
decisions. Personnel Psychology, 47, 847-860.

Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159.

Cohen, J. (1988). ﬂatistical power alalysis for the behavioral sciences. (2“d ed.).
Hillsdale, NJ: Erlbaum.

Cohen, J. & Leﬂcowitz, J. (1974). Development of a biographical inventory blank
to predict faking on personality tests. Journal of Applied Psychology, 59, 404-405.

Collins, J. M., and Gleaves, D. H. (1998). Race, job applicants, and the five-
factor model of personality: Implications for Black psychology,
Industrial/Organizational Psychology, and the Five-Factor Theory. Journal of Applied

Psychology, 83, 531-544.

126

Costa, P. T., & McCrae, R. R. (1985). The NEO Personaligr Inventogr manual.
Odessa, FL: Psychological Assessment Resources.

Costa, P. T., & McCrae, R. R. (1989). The NEO PI/FFI manual supplement.
Odessa, FL: Psychological Assessment Resources.

Cunningham, M. R., Wong, D. T., & Barbee, A. P. (1994). Self-presentation
dynamics on overt integrity tests: Experimental studies of the Reid report. Journal of

Applied Psychology, 79, 643-658.

Doll, R. E. (1971). Item susceptibility to attempted faking as related to item
characteristic and adopted fake set. Journal of Psychology, 77. 9-16.

Dollinger, S. J ., & Orf, L. A. (1991). Personality and performance in
“personality”: Conscientiousness and Openness. Journal of Research in Personality. 25 ,
276-284.

Douglas, E. F., McDaniel, M. A., & Snell, A. F. (1996). The validiry of non-

cogpitive measures decays when applicants fake. Paper presented at the annual
conference of the Academy of Management, Cincinnati, OH.

Dunnette, M. D., McCartney, J ., Carlson, H. C., & Kirchner, W. K. (1962). A
study of faking behavior on a forced-choice self-description checklist. Personnel

Psychology, 15, 13-24.

Dwight, S. A. & Alliger, G. M. (1997). Using response latencies to identify overt
integrity test dissimulation. Paper presented at the 12th annual conference of the Society
for Industrial and Organizational Psychology, St. Louis, April, 1997.

Ellingson, J. E., Sackett, P. R., & Hough, L. M. (1999). Social desirability
corrections in personality measurement: Issues of applicant comparison and construct

validity. Journal of Applied Psychology, 84, 155-166.

Ellingson, J. E., Smith, D. B., & Sackett, P. R. (1999). Investigating the inﬂuence
of social desirability on parsonality factor structure. Paper presented at the 14‘h annual
conference of the Society of Industrial and Organizational Psychology, Atlanta, May,
1999.

Feldman, J. M., & Lynch, J. G. (1988). Self-generated validity and other effects
of measurement on belief, attitude, intention, and behavior. Journal of Applied

Psychology, 73, 421-435.

Fishbein, M. & Ajzen, I. (1975). Beliefs. attitude. intention. and behavior: An
introduction to theory and research. Reading, MA: Addison-Wesley.

127

Fishbein, M., & Jaccard, J. J. (1973). Theoretical and methodological
considerations in the prediction of family planning intentions and behavior.

Representative Research in Social Psychology, 4, 37-51.

Fletcher, C. (1990). The relationships between candidate personality, self-
presentation strategies, and interviewer assessments in selection interviews: An
empirical study. Human Relations. 43, 739-749.

Godin, G., Valois, P., Lepage, L., & Desharnais, R. (1993). Predictors of
smoking behavior: An application of Ajzen’s theory of planned behavior. British
Journal of Addiction. 87, 1335-1343.

 

Goldman, J. A. & Olczak, P. V. (1976). Effect of knowledge about self-
actualization on faking the personal orientation inventory. Journal of Consulting and

Clinical Psychology, 44, 680.

Hiscock, C. K., Layman, L. B., & Hiscock, M. (1994). Cross-validation of two
measures for assessing feigned mental incompetence in male prison inmates. Criminal
Justice and Behavior. 21, 443-453.

Holden, R. R., & Jackson, D. N. (1981). Subtlety, information, and faking effects
in personality assessment. Journal of Clinigrl Psychology. 37, 379-388.

Hollinger, R. C. (1989). A manager’s gpide to preventing employee theft. Park
Ridge, IL: London House Press.

Hough, L. M. (1998). Effects of intentional distortion in personality measurement
and evaluation of suggested palliatives. Human Perforrrrance. 11, 209-244.

Hough, L. M., Eaton, N. K., Dunnette, M. D., Kamp, J. D., & McCloy, R. A.
(1990). Criterion-related validities of personality constructs and the effect of response
distortion on those validities. Journal of Applied Psychology, 75, 581-595.

Hunt, H. (1948). The effect of deliberate deception on Minnesota Multiphasic
Personality Inventory Performance. Journal of Consulting Psychology, 12, 396-402.

Hurtz, G. M. & Bohon, L. M. (1997). Impression management of personality-
b_ased employee selection tests. Paper presented at the 12th annual conference of the
Society for Industrial and Organizational Psychology, St. Louis, April, 1997.

Kelly, B. L., Miles, C. C., & Terman, L. M. (1936). Ability to inﬂuence one’s

score on a typical pencil and paper test of personality. Character and Personality, 4, 206-
215.

Klein, S. P. & Owens, W. A. (1965). Faking of a scored life history blank as a
function of criterion objectivity. Journal of Applied Psychology. 49, 452-454.

128

Kluger, A. N. & Colella, A. (1993). Beyond the mean bias: The effect of
warning against faking on biodata item variances. Personnel Psychology, 46, 763-780.

Kluger. A. N, Reilly, R. R., & Russell, C. J. (1991). Faking biodata tests: Are
option-keyed instruments more resistant? Journal of Applied Psychology. 76, 889-896.

Krug, S. E. (1978). Further evidence on the 16PF distortion scales. Journal of
Personality Assessment. 42, 515-518.

Lautenschlager, G. J. (1994). Accuracy and faking of background data In G.S.
Stokes, M. D. Mumford, & W.A. Owens (Eds), Biodata Handbook (pp. 391-419). Palo
Alto, California: Consulting Psychologists Press.

Leary, M. R., & Kowalski, R. M. (1990). Impression management: A literature
review and two component model. Psychological Bulletin, 107, 34-47.

Longstaff, H. P.. (1948). Fakability of the Strong Interest Blank and the Kuder
Preference Record. Journal of Applied Psychology, 32, 360-369.

Mabe, P. A., & West, S. G. (1982). Validity of self-evaluation of ability : A
review and meta-analysis. Journal of Applied Psychology. 67. 280-296.

Mael, F. A. (1991). A conceptual rationale for the domain and attributes of
biodata items. Personnel Psychology, 44, 763-791.

Marcy, S. M., & Brown, M. A. (1983). Residential energy conservation: The
role of past experience in repetitive household behavior. Environment and Behavior. 15,
123-141.

McCrae, R. M. & Costa, P. T. (1983). Social desirability scales: More substance
than style. Journal of Consulting and Clinical Psychology. 51, 882—888.

McDaniel, M. A., Douglas, E. F., and Snell, A. F. (1997). A survey of deception
among job seekers. Paper presented at the 12th Annual Conference of the Society for
Industrial and Organizational Psychology, St. Louis, April.

McFarland, L. A., & Kabin, M. (in progress). Strategies to reduce faking on
personality tests: Effects on applicant reactions.

McFarland, L. A. & Ryan, A. M. (in press). Variance in faking across non-
cognitive measures. Journal of Applied Psychology.

129

McFarland, L. A., Ryan, A. M. & Ellis, A. (2000). Effect of item placement on
faking a p_ersonality meagre: Does randomizing items lead to less faking? Paper
presented at the 15Lh annual conference for the Society of Industrial and Organizational
Psychology, New Orleans, LA.

 

Mumford, M. D. & Stokes, G. S. (1992). Developmental determinants of
individuals action: Theory and practice in applying background measures. In M. D.

Dunnette & L. M. Hough (Eds), Handbook of industrial and organizational psychology
(2“d ed., vol. 3, pp. 61-138). Palo Alto, CA: Consulting Psychologists Press.

Murphy, K. R. (1993). Honesy in the worlgplace. Belmont, CA: Brooks/Cole.

Napier, J. D. (1979). Effects of knowledge of cognitive-moral development and
request to fake on deﬁning issues test p-scores. Journal of Psychology, 101, 45-52.

Nias, D. K. B. (1972). The effects of providing a warning about the lie scale in a
personality inventory. British Journal of Educational Psychology, 42, 308-312.

Norman, P., & Conner, M. (1996). Predicting health-check attendance among
prior attenders and nonattenders: The role of prior behavior in the theory of planned
behavior. Journal of Applied Social Psychology, 26, 1010-1026.

O’Bannon, R. M., Goldinger, L. A., & Appleby, J. D. (1989). Honesty and
integg'ty testing: A practical gpide. Atlanta: Applied Information Resources.

Ones, D. S. & Viswesvaran, C. (1998). The effects of social desirability and
faking on personality and integrity assessment for personnel selection. Human
Performance. 11, 245-269.

 

Ones, D. S., Viswesvaran, C., & Reiss, A. D. (1996). Role of social desirability
in personality testing for personnel selection: The red herring. Journal of Applied

Psychology, 81, 660-679.

Pandey, J ., & Rastagi, R. (1979). Machiavellianism and ingratiation. Journal of
Social Psychology. 108, 221-225.

Pannone, R. D. (1984). Predicting test performance: A content valid approach to
screening applicants. Personnel Psychology. 37, 507-514.

Paulhus, D. L. (1984). Two-component models of socially desirable responding.
Journal of personality and Social PsvcholagL46, 598-609.

Paulhus, D. L. (1986). Self-deception and impression management in test

responses. In A. Angleiner & J. S. Wiggins (Eds), Personality Assessment Via
Questionnaire New York: Springer.

130

Paulhus, D. L. (1991). Measurement and control of response bias. In Measures
of Personality and Social Psychological Attitudes. Academic Press, Inc.

Reinecke, J ., Schmidt, P., & Ajzen, I. (1996). Application of the theory of
planned behavior to adolescents’ condom use: A panel study. Journal of Applied Social
Psychology, 26, 749-772.

Rogosa, D., Brandt, D., & Zimowski, M. (1982). A growth curve approach to the
measurement of change. Psychological Bulletin. 92, 726—748.

Rosse, J. G., Stecher, M. D., Miller, J. L., & Levin, R. A. (1998). The impact of
response distortion on preemployment personality testing and hiring decisions. J oumal
of Applied Psychology. 83, 634-644.

Ryan, A. M. & Sackett, P. R. (1987). Pre-employment honest testing: Fakability,

reactions of test takers, and company image. J oumal of Business and Psychology, 1,
248-256.

Schifter, D. B. & Ajzen, I (1985). Intention, perceived control, and weight loss:
An application of the theory of planned behavior. Journal of Personality and Social
Psychology, 49, 843-851.

Schlenker, B. R. (1980). Impression Management. Monterey, CA: Brooks/
Cole.

Schlenker, B. R. & Weigold, M. F. (1992). Interpersonal processes involving
impression regulation and management. Annual Review of Psychology. 43. 133-168.

Schmit, M. J. & Ryan, A. M. (1993). The big ﬁve in personnel selection: Factor

structure in applicant and nonapplicant populations. Journal of Applied Psychology, 78,
966-974.

Schrader, A. D. & Osbum, H. G. (1977). Biodata faking: Effects of induced
subtlety and position speciﬁcity. Personnel Psychology, 30, 395-404.

Shaffer, G. S., Saunders, V., & Owens, W. A. (1986). Additional evidence for the
accuracy of biographical data: Long-term retest and observer ratings. Personnel

Psychology, 39, 791-809.

Stokes, G. S., Hogan, J. B., & Snell, A. F. (1993). Comparability of incumbent
and applicant samples for the development of biodata keys: The inﬂuence of social
desirability. Personnel Psychology. 46, 739-762.

131

Tett, R. R, Jackson, D. N ., & Rothstein, M. (1991). Personality measures as
predictors of job performance: A meta-analytic review. Personnel Psychology, 44, 703-
742.

Theodorakis, Y. (1992). Prediction of athletic participation: A test of planned
behavior theory. Perceptual and Motor Skills. 74, 371-379.

Tisak, J. & Smith, C. S. (1994). Defending and extending difference score
methods. Journal of Management, 20, 675-682.

Wheeler, J. K., Hamill, L. S., & Tippins, N. T. (1996). Warning against candidate
misrgpresentation: Do they work? Paper presented at the 11th Annual Conference of the
Society of Industrial and Organizational Psychology, San Diego, April.

Zerbe, W. J. & Paulhus, D. L. (1987). Socially desirable responding in
organizational behavior: A reconception. Academy of Management Review. 12, 250-
264.

Zickar, M. (1997). Computer simulation of faking on a mrsonality test. Paper
presented at the 12th Annual Conference of the Society for Industrial and Organizational
Psychology, St. Louis, April.

Zickar, M., Rosse, J ., & Levin, R. (1996). Modeling faking on a personality test.

Paper presented at the 11th Annual Conference of the Society for Industrial and
Organizational PsychologY, San Diego, April.

132

APPENDICES

133

APPENDIX A

Pre-Questionnaire

Please respond to the following questions as honestly as you can. Several of the items
below ask about your feelings toward lying on selection tests. Please note that by the
term “lying” it is meant any response that is not completely accurate. Therefore,
providing a response on a selection test that you think is an exaggeration of the truth is
considered a lie, as is a response that is completely false. Please use this deﬁnition of
“lying” as you respond to the items below.

Please indicate the extent to which you agree or disagree with items 1 to 17 using the
following 5 point scale:

1 = Strongly Disagree

2 = Disagree
3 = Neither Agree nor disagree
4 = Agree

5 = Strongly Agree

1. If I lied on a selection test, most of the people who are important
to me would disapprove. [SN]

2. No one who is important to me thinks it is OK to lie on a selection test. [SN]

3. Most people who are important to me will look down on me if I lie on a selection

test. [SN]
4. It would be easy for me to lie on a selection test. [PBC]
5. If I want to, I can lie on a selection test. [PBC]

6. I can imagine times when I might lie on a selection test, even if I had not
planned to. [P1]

7. Even if I had a good reason, I could NOT bring myself to lie on a selection test.[PI]

8. If I had the opportunity, I would lie on a selection test. [PI]
9. I would NEVER lie on a selection test. [PI]
10. I would exaggerate the truth on a selection test. [PI]

134

11. I may lie on a selection test in the future. [PI]

12. I would make myself look as good as I possibly could on a selection test. [PI]
13. I would NOT exaggerate the truth on a selection test. [PI]
14. My parents would approve of me lying on a selection test. [SN]
15. Lying on a selection test would be easy for me. [PBC]
16. I would try to make myself look as good as possible on a selection test. [PI]
17. I would lie on a selection test. [PI]

18. I think that lying on a selection test is (Circle the response that best describes how

you feel): [A]
1 2 3 4 5
Bad Good
1 2 3 4 5
Unpleasant Peasant
1 2 3 4 5
Foolish Wise
1 2 3 4 5

Useless Useful
1 2 3 4 5
Unattractive Attractive

A=Attitude toward faking, SN=Subjective norms toward faking, PBC=Perceived
behavioral control, PI=Pre-intentions to fake.

135

APPENDIX B

Post-Questionnaire

Please respond to the following questions as honestly as you can. Several of the items
below ask about your feelings toward lying (or faking) on the test. Please note that by the
terms “lying” and “faking” it is meant any response that is not completely accurate.
Therefore, providing a response on a selection test that you think is an exaggeration of
the truth is considered a lie, as is a response that is completely false. Please use this
deﬁnition of “lying” and “faking” as you respond to the items below.

Answer each of the following statements using the ﬁve point scale below.

Strongly Disigree Neither3 Agree Agree Strongly Agree
Disagree or Disagree

1. It was usually clear to me which personality trait an item was measuring. [K]
2. I intended to give some inaccurate answers on the test. [I]
3. I would like to score in the top 15% on this test. [V]
4. It is very desirable to me to get a high score on the personality test. [V]
5 . I was concerned that A lie scale would identify me as a liar. [C]
6. Some items measured how reliable and responsible a person is. [K]
7. I attempted to fake my responses on the test. [I]
8. I think some items were on the test to detect lying. [C]
9. I intended to respond to all items honestly. [I]
10. It was obvious which personality traits the test was measuring. [K]
11. I was NOT worried that I would be identiﬁed as lying on the test. [C]
12. I intended to make myself look as good as possible on the test. [I]
13. I think there is a way the experimenter can tell if I lied on the test. [C]
14. Some items measured how friendly and agreeable a person is. [K]
15. I tried to be honest so that I would not be identiﬁed as lying on the test. [C]
16. It was very important to me to do well on the personality test. [V]
17. I intended to make myself look very good on the selection test. [I]

136

18. I do NOT think there is any way the experimenter can tell if I lied or not. C]

19. I really wanted to score high on the test. [V]
20. I could NOT tell what the personality test was measuring. [K]
18. While I took the test, I was concerned that my test scores may be thrown

out because I lied. [C]
22. I attempted to lie a lot on the test. [I]

23. Some items measured how concerned people are with how others view them. [K]

24. Doing well on the personality test was NOT desirable to me. [V]
25. Some items measured whether a person prefers to be alone or enjoys being around
others. [K]
26. I intended to lie on some items to increase my test score. [I]
27. Some items measured how open a person is to new experiences. [K]

28. I have heard that tests like the one I just took often include items to detect
when an applicant is lying. [C]
29. As I took the test I was concerned that there may be a lie scale included that

would identify those who lie on the test. [C]

K=Knowledge of construct being measured, I =Intention to fake, C = Concern with being
identified as a faker, V: Valence toward doing well on the test.

137

APPENDIX C

Demographic Questionnaire

Please circle or fill in the appropriate response:

PID

1.

. Age:

 

Year:

a. Freshman
b. S0phomore
c. Junior

(1. Senior

e. Other

.Sex: a. Male b. Female

. GPA (If you are a freshman, give the cumulative

GPA you had when you graduated high school):

. Race:

a. White

b. African American
c. Hispanic

(1. Asian

e. Native American
f. Other

. How many times have you taken the type of personality test you just took:

Never

Once or twice
Three or four times
More than 4 times

kWh-D"

7. Where you thinking of any particular job as you answered the personality test?

8. If the answer to question 7 was yes, what type of job were you thinking about?

 

138

APPENDIX D

Consent Form

Selection Testing
Purpose of Study

The following study has two parts. You will complete a personality test that is typically
used to hire job applicants, two times. You will also be asked to answer some survey
questions. The purpose is to obtain a better understanding of how individuals respond to
such tests. The entire experiment will take no longer than one and a half hours (1.5
hours). You will receive 3 credits for your participation

Informed Consent

The nature and general purpose of this study has been explained to me and the attached
statement has been read to me by Lynn McFarland, from the Department of Psychology.
I understand that my participation is voluntary and that all information is conﬁdential and
my identity will not be revealed. I am free to withdraw consent and discontinue
participation in the project at any time. Any questions I have about the project will be
answered by the researcher.

On the basis of the above statements, I agree to participate in this project.

 

 

Participant’s Signature Date

 

Participant’s Name (Printed)

139

APPENDIX E

Protocol

Hello, my name is Lynn McFarland and I will be your experimenter today. The
purpose of this study is to see how individuals respond to a personality test that is
frequently used to select applicants into a variety of different jobs.

In this experiment you will take a 60 item personality test twice, given different
instructions each time. You will also be asked to ﬁll out a survey that will ask you for
some demographic information and another survey that will ask some other questions
relating to the test. This entire experiment should last one and a half hours (1.5 hours).
You will receive one credit for your class per half hour. However, even if this
experiment takes less than one and a half hours you will receive the full three credits if
you complete all parts of the experiment.

[hand out consent form]

The consent form I am handing out states that you are participating in this study
voluntarily and that all of your responses will remain completely conﬁdential and
anonymous. You can leave this experiment any time you want. However, you will only
receive credit for the amount of time you are here.
[wait until all participants have signed the consent form and answer any questions they
might have. Then collect the forms)
[hand out scantron answer sheets]

I am now handing out an answer sheet. In the section of your answer sheet that is

marked PID write your mother’s birth date as your identiﬁcation number and place that

140

number in the section of your questionnaire that is marked PID. For example, if your
mother’s birth date is June 13, 1950, you should record “061350’ in the section of your
questionnaire that is marked PID. If your mother was born on September 6, 1955, you
should record 090655 in the section of your questionnaire that is marked PID. If you do
not know or want to use your mother’s birth date, then use some other number that you
will remember because you will have to use this PID number throughout the experiment.
For the honest condition

For the knowledge of construct condition
[hand out description of constructs measured on the test]

The sheet that I am handing out describes the ﬁve constructs (personality traits)
the personality test you are about to take [or about to take again] measures. The
personality test was created to measure ﬁve personality traits. It measures
Conscientiousness, Neuroticism, Openness to New Experiences, Agreeableness, and
Extraversion. [hand out sheet that describes each construct and have them read along as
you read the following] Conscientiousness is how reliable a person is. Those high in
conscientiousness are typically rule abiding and reliable while those low on this trait are
typically late and fail to meet deadlines. Neuroticism measures how worried a person
typically is. Those high in neuroticism generally are very concerned with how others
perceive them and they are very concerned with detail. Those low on this trait are less
concerned with how others perceive them and they tend to worry much less. Openness to
new experiences is how open minded a person is. Those high on this trait are typically
very willing to try new things and are open to others’ Opinions and points of views, while

those low on this trait tend to be closed minded. Agreeableness is how agreeable a

141

person is. Those high on this trait generally avoid confrontations or arguments and are
very friendly. Those low on this trait tend to enjoy conﬂict and often disagree with
others. Finally, extraversion measures how outgoing and friendly an individual is. Those
high in extraversion generally speak more quickly, have no trouble speaking to new
people, and enjoy being around others. Those low on this trait generally prefer to be
alone and are more likely to get anxious when they are around people they do not know.
For all honest conditions

Before I hand out the personality test I want to remind you that the test you are
about to take is one that is frequently used by employers to select employees for all types
of jobs (e. g., managerial jobs, salesperson, police ofﬁcer, etc). However, keep in mind
that within this context the results of this test has no bearing on you. Therefore, please
answer the following questions on the personality test as honestly as possible. Your
answers will remain completely anonymous. I have no way of connecting the test back to
you. Your answers will be used for research purposes only, and will not be used to
evaluate you in any way. The nature of the project requires that you answer the
following questions as honestly as possible, so please provide as accurate answers as you
can.

For warning of lie scale condition

[half of the participants will also be given the following instructions]

Sometimes people try to make themselves out to be better than they really are.
Therefore, this test includes a scale to check on this. Such scales are used to identify
individuals who give inaccurate or false responses. Please keep this fact in mind as you

take this selection test.

142

[hand out personality test]

There is no time limit on this test. You may begin.
[wait till everyone appears to be done]

Has everyone completed the test? [if everyone has ﬁnished, proceed to next
test/measure (this will change depending on if the test is taken under the honest condition
ﬁrst or after the applicant condition)].

Applicant Conditions
[hand out scantron answer sheets]
I am now handing out an answer sheet. In the section of your answer sheet that is
marked PID write the same PID that you wrote on the other answer sheet.
For Knowledge of constructs measured condition
[if the test was already administered under the honest condition, then participants will
simply be reminded that the test measures ﬁve constructs and given the names of these
ﬁve constructs again. If the participants are given the personality test under the applicant
condition first, then they will be given the same instructions as described above]
For all applicant conditions

Again I would like to remind you that the test you are about to take is one that is
frequently used by employers to select employees for all types of jobs (e. g., managerial
jobs, salesperson, police ofﬁcer, etc). When answering the questions on the personality
test imagine that you are a job applicant. Please answer as you would if you were really
taking the test to get a job. Keep in mind that your answers will be kept completely
anonymous.

For incentive condition

143

[in addition to the general applicant instructions given directly above, participants in this
condition will also be told the following]

To make this situation more like an applicant situation, we are offering an

incentive. Those of you who score in the top 15% on this test will receive $20.

For warning of lie scale condition
[those in this condition will also be given the general applicant instructions above. In
addition half will also be in the incentive condition)

Sometimes people try to make themselves out to be better than they really are.
Therefore, this test includes a scale to check on this. Such scales are used to identify
individuals who give inaccurate or false responses. Individuals who have a high score
on this scale (i.e., those individuals that the lie scale identiﬁes as faking responses) will
be eliminated from the selection process (and from consideration for the incentive if
they are in that condition). Please keep this fact in mind as you take this selection test.

For all applicant conditions

[hand out personality test]
There is no time limit on this test. You may begin.
[wait until it appears that everyone has completed the test]

Has everyone completed the personality test? [if everyone is done hand out

another scantron form]

Once again, I am handing out an answer sheet. In the section of your answer

sheet that is marked PID write the same PID that you wrote on the other answer sheets.
Please check to ensure that the PID you put on this answer sheet is the same one you

placed on the others.

144

[hand out post-questionnaire]

The survey I am handing out [the post-questionnaire] asks you some questions
regarding your orientation and frame of mind while you were taking the test. [Read
instructions out loud as they read along] Please answer honestly. Again, I have no way
to connect your responses back to you. Responses will be kept anonymous, so please
respond honestly.

[when it seems like everyone is done]

Has everyone completed the survey? [If everyone is done proceed to hand out the
demographic form].

[Hand out demographic questionnaire]

The questionnaire I am handing out asks you some demographic information.
Please completely ﬁll out this questionnaire. Where it says PID please write the same ID
you put on all scantrons and on the post-questionnaire.

[when everyone is done ﬁlling out the demographic form, hand out debrieﬁng form]

I am now handing out the debrieﬁng form which will describe the purpose of this
study. [I will read the debrieﬁng form to them as they read it themselves]. Are there any
questions? If anyone has any questions regarding this study feel free to call me at the

number listed on the debrieﬁng form. Thank you very much for your participation.

145

APPENDIX F

Debrieﬁng Form

Research has indicated that personality tests (like the one you just took) can be faked.
For example, when asked to increase scores on these types of tests, individuals can
increase their scores considerably. Research has also shown that applicant faking on
personality tests may reduce the useful of these tests. For example, it has been shown
that when applicants fake these tests, the tests are not as valid (i.e., the test does a poor
job of identifying those who would do well on the job). Therefore, researchers and
employers have tried to determine ways to reduce applicant faking on these tests within
selection contexts.

Considerable research has shown that warning applicants that there is a lie scale on the
test that identify fakers may reduce applicant faking. Additionally, it has been suggested
that the more information applicants have about a test, the more likely they are to fake it
(because they are better able to fake). Research also suggests that the more desirable the
outcome of doing well is, the more applicants fake. Therefore, if an applicant ﬁnds a job
very desirable, he or she is considerably more likely to fake responses in an effort to get
as high a score as possible in the hopes that this will get him or her the job.

This study attempts to examine a number of the factors that may reduce applicant faking
in selection contexts. For example, some participants were given information about the
constructs (personality traits) that the test was designed to measure. We believe that we
will ﬁnd that those of you given such information will have higher scores (when you
were asked to take the test as an applicant) as compared to those who were not given this
information.

At this point I would like to urge you to remember (write down) the ID number you
have used throughout this experiment. Some of the participants in this study were given
an incentive of $20 when they took the personality test under instructions to behave like
an applicant. These individuals were told that those who scored in the top 15% on the
test would receive the money. Therefore, those who were in this condition should email
or call me (see below) after December 15, 1999. When you contact me, be sure to have
your ID number handy. I will use this ID number to look up your score and tell you if
you were in the top 15% (note that a lie score will not be used to eliminate any
individuals from consideration for the money). If you do not remember your ID
number, I will have no way of knowing how you scored on the test and there is no
way you can receive the money. If you do qualify for the money (i.e., you are in the top
15%) I will tell you how and when to pick up the $20.

To be fair, those of you who were offered an incentive and were told that a lie scale was

included on the measure will be considered separately from those who were not informed
that a lie scale was included in the measure (as those of you warned about the lie scale

146

would likely have not scored as high as those who were not). Also, those not offered an
incentive will also be eligible for $20 as well. Of the approximately 160 individuals who
take part in the experiment who were not given such an incentive, 15% will be randomly
selected to receive the $20. Contact me by phone or email (see below) after December
15, 1999 to ﬁnd out if you were selected as one of the 15 % to receive the money. Again,
make sure you have your ID number. If you do not know your ID number I will
have no way of knowing if you were one of the people selected to get the money.

Thank you very much for your participation. If you have any questions regarding the

study please feel free to contact me, Lynn McFarland, by phone or email
(M CFARLI 8 @pilot.msu. edu or 355-2171).

147

HICHIGQN STQTE UNIV. LIBRQRIES
l||||llllllllllllllllllllllllllllllllll1111111111
31293020486530