Vous êtes sur la page 1sur 20

INTERNATIONAL JOURNAL OF SELECTION AND ASSESSMENT

VOLUME 14 NUMBER 4 DECEMBER 2006

A Meta-Analytic Investigation of Job Applicant Faking on Personality Measures


Scott A. Birkeland* and Todd M. Manson
University of South Florida

Jennifer L. Kisamore
University of Oklahoma-Tulsa

Michael T. Brannick and Mark A. Smith


University of South Florida

This study investigates the extent to which job applicants fake their responses on personality tests. Thirty-three studies that compared job applicant and non-applicant personality scale scores were meta-analyzed. Across all job types, applicants scored significantly higher than non-applicants on extraversion (d 5 .11), emotional stability (d 5 .44), conscientiousness (d 5 .45), and openness (d 5 .13). For certain jobs (e.g., sales), however, the rank ordering of mean differences changed substantially suggesting that job applicants distort responses on personality dimensions that are viewed as particularly job relevant. Smaller mean differences were found in this study than those reported by Viswesvaran and Ones (Educational and Psychological Measurement, 59(2), 197210), who compared scores for induced fake-good vs. honest response conditions. Also, direct Big Five measures produced substantially larger differences than did indirect Big Five measures.

here has been a resurgence of interest in the use of personality variables in industrial/organizational psychology since the early 1990s. This resurgence is due in part to the emergence of the five-factor (Big Five) model of personality (Digman, 1990) and meta-analytic evidence that supports the usefulness of personality measures for the prediction of performance across a wide variety of jobs (e.g., Barrick & Mount, 1991; Hurtz & Donovan, 2000; Tett, Jackson, & Rothstein, 1991). The increased use of personality measures in personnel selection, however, has raised concerns about the extent to which applicants can

This paper is based on an earlier study presented at the 18th Annual Conference of the Society for Industrial and Organizational Psychology, Orlando, FL, April 2003. Special thanks to Yufan Liu for serving as a coder for this study. *Address for correspondence: Scott Birkeland, 1405 26th Ave NE, Minneapolis, MN 55418. E-mail: scott.birkeland@personneldecisions. com Scott Birkeland is now at Personnel Decisions International. Todd M. Manson is now at Allegheny College. Mark A. Smith is now at American Institutes for Research.

and do distort responses on such measures in an effort to improve their chances of being hired. Many studies have examined the extent to which participants who are instructed to fake responses can make themselves appear more socially desirable than participants who are instructed to respond honestly (see Viswesvaran & Ones, 1999). Such induced-faking studies usually include a fake-good condition whereby participants are instructed to respond in a manner that will make a good impression, make them look qualified for the job, increase their chances of getting the job, or ensure that they get the job. Viswesvaran and Ones (1999) meta-analyzed studies comparing mean fake-good and honest responses on personality measures that they classified into one of the Big Five dimensions. As expected, participants instructed to fake responses in a positive direction had higher means for each of the Big Five dimensions than did participants instructed to respond honestly. For studies that utilized between-subjects designs, the magnitude of faking was fairly consistent with standardized mean difference effect sizes across the Big Five dimensions ranging from .48 to

r 2006 The Authors Journal compilation r 2006 Blackwell Publishing Ltd, 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA

317

318 SCOTT A. BIRKELAND, TODD M. MANSON, JENNIFER L. KISAMORE, MICHAEL T. BRANNICK AND MARK A. SMITH

.65. The effect sizes for within-subjects studies were generally larger and ranged from .47 to .93. Thus, it is clear that when instructed to do so, participants are able to distort their responses on personality measures in the socially desirable direction. Results from individual studies following the inducedfaking paradigm suggest that participants tailor their fake-good responses on personality measures to their perception of the personality traits of employees in the target job. Furnham (1990) asked participants to present themselves as ideal candidates for three different jobs (advertising executive, banker, and librarian) and found that they produced quite different profiles for each job. Other studies have found that participants fake-good personality profiles closely matched the profiles the participants produced when describing an ideal employee (Martin, Bowen, & Hunt, 2002) or a typical employee (Mahar, Cologon, & Duck, 1995) in the target job. Research utilizing the induced-faking paradigm does not, however, address job applicant faking or response distortion on personality measures as part of a real-life selection process. Some researchers have suggested that there is relatively little faking by job applicants (e.g., Dunnette, McCartney, Carlson, & Kirchner, 1962; Hough, Eaton, Dunnette, Kamp, & McCloy, 1990). Other studies, however, have reported higher mean personality scale scores among job applicants compared with job incumbent or student groups (e.g., Griffin, 2002; Hough, 1998; Robie, Zickar, & Schmit, 2001; Rosse, Stecher, Miller, & Levin, 1998) and higher scores on social desirability or impression management scales among job applicants compared with non-applicants (Hough, 1998; Rosse et al., 1998). This recent research suggests that job applicants actually do distort their responses on personality measures, presenting themselves in a more desirable fashion. The current study examines faking or distortion that occurs when individuals take personality inventories as part of an employment application process by metaanalyzing studies comparing mean scale scores of job applicants with mean scale scores of non-applicants (e.g., job incumbents) on the same personality tests. This approach provides an opportunity to not only examine the overall extent of job applicant faking on personality measures but also provides an indication of the particular personality dimensions that applicants are most prone to distort, and whether this pattern varies for different types of jobs. The examination of the difference in means between applicants and non-applicants has been used in other studies as a procedure for detecting response distortion (e.g., Hough, 1998; Robie, 2001). The standardized mean difference (d, i.e., the standardized mean difference between applicant scale scores and non-applicant scale scores) has been the effect size of choice in such analyses. The authors position is that at least part (and perhaps the largest part) of the observed difference between applicant

and non-applicant scores is due to faking or response distortion. Because study participants were not assigned to condition, reasons for any observed difference other than faking cannot be eliminated. For example, it could be argued that the higher observed means for applicants compared with incumbents reflect true differences in personality. If there were true differences in personality among these groups, however, one would expect job incumbents to possess higher levels of socially desirable personality traits than do applicants because incumbents have previously been subjected to selection and have successfully retained their jobs (Robie, 2001). In addition to selection and attrition, the experience of working in a given environment could actually change a persons personality, presumably in the socially desirable direction. Such effects would, however, work against finding higher scores for applicants. An alternative explanation for higher applicant mean scores concerns item context (Robie, Schmit, Ryan, & Zickar, 2000; Schmit, Ryan, Stierwalt, & Powell, 1995). Robie (2001) suggests that non-applicants might consider their behavior across a wide range of life contexts when responding to a personality measure, whereas applicants may complete the measure focused specifically on how they behave at work. This frame-of-reference difference might cause applicants to score higher on a personality measure due to the finding that individuals who take a personality measure that is comprised of items contextualized to a work environment obtain higher scores than individuals taking a personality test that was comprised of general, non-contextualized items (Robie, 2001). Although this possibility certainly warrants future research, it seems doubtful that such context effects could account for all differences in personality scores between applicants and non-applicants. Consequently, we consider observed differences in means between applicants and non-applicants that favor the applicants to be consistent with the premise that applicants are faking.

Hypotheses
Similar to other personality meta-analyses (Barrick & Mount, 1991; Tett et al., 1991; Viswesvaran & Ones, 1999), scales from the measures used in studies included in this analysis were classified into one of the Big Five dimensions and analyzed separately. Thus, five different meta-analyses were conducted, one for each Big Five dimension. A number of hypotheses concerning these primary analyses were made. First, higher mean scores were hypothesized for job applicants compared with nonapplicants on all of the Big Five personality dimensions (Hypothesis 1a). This hypothesis is based on the assumption that job applicants want to present themselves in a favorable, socially desirable manner. Also, the effect sizes across dimensions were hypothesized to be smaller than those found by Viswesvaran and Ones (1999; Hypothesis

International Journal of Selection and Assessment

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

JOB APPLICANT FAKING ON PERSONALITY MEASURES

319

1b) because the Viswesvaran and Ones meta-analysis consisted of studies where participants were explicitly instructed to fake responses (i.e., induced faking) whereas the present study considered faking in real-life job applicant situations. Second, larger effect sizes were predicted for conscientiousness and emotional stability (Hypothesis 2). These larger effects were expected because these dimensions are likely perceived by applicants as being the most critical of the Big Five dimensions to portray in a socially desirable manner when applying for a job. It seems reasonable to expect that across jobs, applicants would try to present themselves as well-adjusted, dependable, and achievement oriented (Griffin, 2002). This logic is also supported by the meta-analytic finding of Ones, Viswesvaran, and Reiss (1996) that social desirability scales correlated most highly with emotional stability and conscientiousness. These two dimensions also had the largest effect sizes among the Big Five in the Viswesvaran and Ones (1999) meta-analysis described earlier. In addition to the primary analysis for each dimension, several moderator variables were examined. Job type was examined as a moderator variable in two ways. First, studies were classified into those involving sales or nonsales jobs (non-sales jobs included: managerial, clerical, manufacturing, public service, military, student, and computer programmer positions). We expected that this classification would be a significant moderator for extraversion such that sales jobs should show larger effect sizes on this dimension than non-sales jobs. More specifically, due to the nature of sales jobs (i.e., one must be able to quickly develop relationships with a wide variety of people), we expected that applicants for such roles would be motivated to present themselves as being highly extraverted, and as a result, make more of an effort to inflate their scores on the extraversion dimension than applicants for other roles (Hypothesis 3). Jobs were also classified as managerial or non-managerial. The supervisory nature of managerial jobs was expected to moderate the conscientiousness dimension, such that there would be a larger effect size for conscientiousness in managerial jobs than non-managerial jobs (Hypothesis 4). Because managerial jobs require planning and organizing skills, we expected that applicants for managerial jobs would be more motivated to present themselves as highly conscientious than would applicants in non-managerial jobs. Two other variables were examined as possible moderator variables with no a priori hypotheses. First, test type was examined by classifying studies into those that used direct measures of the Big Five dimensions and those that used indirect measures of the Big Five dimensions. An instrument was considered a direct measure of the Big Five if its scales were explicitly labeled as the Big Five dimensions. On the other hand, measures that assess personality characteristics closely related to some of the Big

Five dimensions (e.g., 16 PF) but were not specifically developed to measure the five personality dimensions commonly accepted as the Big Five were classified as indirect measures.1 To our knowledge, no studies to date have examined differences in faking for direct vs. indirect Big Five measures. Several studies, however, have compared the criterion-related validities for direct vs. indirect Big Five measures (e.g., Hurtz & Donovan, 2000). Salgado (2003) conducted a meta-analysis of studies examining this issue and reported that Five Factor model (FFM)-based inventories displayed stronger criterion-related validities than non-FFM inventories for conscientiousness and emotional stability. Given these results, the present investigation was interested in comparing the effect sizes of direct vs. indirect Big Five measures. The degree of congruence or similarity between the applicant and non-applicant jobs in each study was also examined as a possible moderator. The purpose of this moderator analysis was to investigate whether studies that rigorously matched applicants and non-applicants produced effect sizes significantly different than studies that used less rigorous matching procedures. Studies deemed to have rigorous matching were those where members of the applicant group were applying for the same type of job currently held by non-applicant group members. Clearly, if one is to compare applicant test scores to non-applicant test scores, and then assume that the observed differences between these groups are due to intentional distortion, any known differences between these groups should be minimized as much as possible. Rigorous matching of groups, however, is not always feasible. Some studies, for example, included a mixture of similar jobs, such as various midlevel management positions in both the applicant and nonapplicant groups and considered the two groups as being equivalent for comparison purposes. Thus, the final moderator analysis was conducted with the intention of examining differences in effect sizes for studies that had a rigorous matching procedure vs. those that did not. This moderator analysis provides some evidence of the effects of methodological rigor (Lipsey & Wilson, 2001) on the outcomes of the primary studies.

Method
Inclusion Criteria
The present meta-analysis included studies that examined mean differences in personality test scores for persons applying for jobs (applicant group) vs. persons not applying for jobs (non-applicant group). Studies included in the metaanalysis met several criteria. First, individuals in both the applicant and non-applicant groups completed the same personality instrument. The participants completed an objective measure of at least one of the Big Five personality constructs or a construct related to one of the Big Five constructs. Second, participants in the applicant group

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

Volume 14 Number 4 December 2006

320 SCOTT A. BIRKELAND, TODD M. MANSON, JENNIFER L. KISAMORE, MICHAEL T. BRANNICK AND MARK A. SMITH

completed the measure as part of a real-life selection process. Finally, included studies were published or presented no earlier than 1950 and were available in English.

Literature Search
Several approaches were used to identify relevant studies. The first step was to conduct a computerized literature search using Psych Lit, Social Sciences Citation Index, and ABI Global Inform databases. Various combinations of the keywords personality, applicant, incumbent, nonapplicant, and Big Five were used in searches to identify published articles, dissertations, and book chapters relevant to our investigation. Second, a manual search of the 2000 through spring, 2003 editions of 13 key journals was conducted to find recent publications that may not yet have been included in the electronic databases. These journals included: Journal of Applied Psychology, Personnel Psychology, Human Performance, Organizational Behavior and Human Decision Processes, Journal of Personality and Social Psychology, Journal of Business and Psychology, Academy of Management Journal, Academy of Management Review, Journal of Management, Journal of Vocational Behavior, International Journal of Selection and Assessment, and Administrative Science Quarterly. Then, the reference sections from the articles selected were manually searched to identify any additional relevant articles. Finally, to avoid the file drawer problem,2 a manual search of the Society for IndustrialOrganizational Psychology (SIOP) program guide from 1998 to 2003 was conducted, several researchers in the area of personality assessment were contacted, and a request was posted on an I/O psychology internet bulletin board to locate any relevant unpublished studies. The search process identified a total of 36 articles, 11 of which were excluded upon further review. Six of the excluded articles were dropped because the studies were deemed to be laboratory induced faking studies (Bass, 1957; Jackson, Wroblewski, & Ashton, 2000; MuellerHanson, Heggestad, & Thornton, 2002; Rusmore, 1956; Schmit et al., 1995; and Vasilopoulos, Reilly, & Leaman, 2000). Three articles were dropped because of failure to include necessary effect size statistics (Orpen, 1971; Schmit & Ryan, 1993; Schwab & Packard, 1973), and two articles were dropped because the instruments used in the selection process were not personality measures (Baker, Dwight, & Jones, 2000; Weiner & Gibson, 2000). Thus, the literature search produced a total of 25 articles, containing 33 independent studies. Of the 33 independent studies, six were unpublished data. The codebook for the included studies is presented in Table 1.

Coding Process
The first step of the coding process was to group all scales into one of the Big Five dimensions using procedures

developed by Barrick and Mount (1991). To do so, three of the five authors and a fourth trained rater independently sorted the scales from instruments that were not direct Big Five measures (a total of 131 scales) into one of the Big Five dimensions. The results of this sorting process are displayed in the Appendix A. To conduct the sorting process, definitions of the scales from each study were reviewed and inter-correlation matrices of these scales were used when available to determine which scale to group together within a study. To classify each scale group into a Big Five dimension, the definition of each Big Five dimension (as reported in Costa & McCraes, 1989 NEO manual) was compared with the scale definitions reported in each study. All four raters agreed on the classification of 75 of the 131 scales (54%), at least three of the four raters agreed on the classification of 109 of the 131 scales (80%), and at least two of the four raters agreed on the classification of 130 of the 131 scales (96%). The overall proportion agreement for this process was .74 yielding a k of .67. Following this process, a consensus meeting was held to determine the final classification for scales that had less than unanimous agreement. As a result of the meeting, five scales were dropped from the analysis due to lack of rater agreement on a single classification for these scales. The next step was to code the moderator variable study characteristics. The moderator variables coded included: (1) whether or not the applicants job involved sales, (2) whether or not the applicants job was managerial, (3) whether the study used a direct or indirect Big Five measure (i.e., an instrument was coded as a direct measure of the Big Five if its scales were explicitly labeled as the Big Five dimensions), and (4) whether the applicant and nonapplicant subjects were matched in terms of job type (i.e., the applicant group was applying for the job for which nonapplicant members were incumbents). It should be noted that in some cases not enough information was provided to code the moderator variables, while in other cases the study contained a mixture of jobs in both groups, making coding the two job-type variables into a single category impossible. In such instances, the moderator variables were coded as not-available and excluded from the moderator analysis. The study characteristics coding was conducted by two of the five authors. To gauge the level of interrater agreement for this coding, k was calculated for each moderator variable. Results indicated perfect agreement (k 5 1.0) among the raters regarding whether a measure was a direct or indirect measure of the Big Five personality factors. k statistics for coding matching of job type, sales vs. non-sales jobs, and management vs. non-management jobs were .75, .77, and .77, respectively. Again, a consensus meeting was conducted to determine the final classification for each of the moderator variables. The last stage of the coding process was to record the sample sizes, means, and standard deviations for the applicant and non-applicant groups for each scale within

International Journal of Selection and Assessment

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

Table 1. Characteristics of studies included in the meta-analysis

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

JOB APPLICANT FAKING ON PERSONALITY MEASURES

Study ID Study useda Congruence Matched Matched Matched Matched Matched Not matched Not matched Matched Not matched Matched Not matched Matched Matched Matched Matched Matched Matched Not matched Matched Not matched Matched Matched 7947 1069 999 197 3079 370 2500 E, A, C, O, S E, A, C, O, S E, A, C, O, S E, A, C 1129 561 569 7,259 5096 946 796 73 701 406 2500 1003 225 152 2989 Indirect Indirect Direct Direct Direct Direct Direct Indirect Direct Direct Direct Not matched Not matched Matched Matched Not matched Not matched Not matched Not matched Not matched Matched Not matched Sales Sales Non-sales Non-sales Non-sales Non-sales Non-sales Non-sales Mixture Non-sales Mixture Non-sales Non-sales Non-sales Non-sales Sales Sales Sales Non-sales Non-sales Non-sales Non-sales Non-sales Non-sales Sales Non-sales Mixture Non-sales Mixture Mixture Mixture Non-sales Mixture Industrial sample Retail sample Banking sample Industrial sample Female sample Male sample Study 1 Study 2 Study 3 Industrial sample Retail sample b. com sample Branch manager CPI sample GPI sample 16 PF NEO (Greek vers.) PCI Weekly CPI GPI PPI Modified NEO TMPs HUGO HPI HPI E, A, C, O, S E, A, C, O, S E, C, O, S E, A, C, O, S E, A, C, O, S E, A, C, O, S E, A, C, O, S Adjective Checklist Adjective Checklist 16 PF 16 PF 16 PF 16 PF Gordon Pers. Prof. Guilford Martin NEO Long form Heron HPI Hough Hough Hough ABLE EPPS EPPS EPPS Matthew Temp. Matthew Temp. PEN OConnell/Hattrup E, A, C, O, S E, A, C, O, S E, A, C, O, S E, A, C, O, S E, S E, S E,C,S E, S C E,S E, A, C, O, S C E, C, O O, S E, A, C, S E, A, C, O E, A, C, O E, A, C, O E, C, S E, C, S E, S A, C 63 96 135 600 136 142 121 70 401 100 94 14442 25423 681 125 66 97 362 81 16 339 1524 50 70 72 190 298 84 88 45 260 100 288 963 508 270 113 49 69 760 139 61 225 374 Indirect Indirect Indirect Indirect Indirect Indirect Indirect Indirect Direct Indirect Direct Indirect Indirect Indirect Indirect Indirect Indirect Indirect Indirect Indirect Indirect Direct Test usedb

Author(s)

Measured constructs

NonApplicant applicant ( N) (N) Test type

Job type 1c Job type 2 Non-mgmt Non-mgmt Mgmt Mgmt Non-mgmt Non-mgmt Non-mgmt Non-mgmt N/A Mgmt Mix Non-mgmt Non-mgmt Non-mgmt Non-mgmt Non-mgmt Non-mgmt Non-mgmt Non-mgmt Mgmt Non-mgmt Non-mgmt Mgmt Mgmt Mgmt Non-mgmt Mix Non-mgmt Mix Mix Mix Non-mgmt Non-mgmt

1.1 1.2 2.1 2.2 3.1 3.2 4 5 6 7 8 9.1 9.2 9.3 10 11.1 11.2 12 13.1 13.2 14 15

16.1 16.2 17 18 19 20 21

22 23 24 25

Dunnette et al. (1962) Dunnette et al. (1962) Elliott (1976) Elliott (1976) Elliott (1981) Elliott (1981) Gordon and Stapleton (1956) Green (1951) Griffin (2002) Heron (1956) Hogan et al. (1998) Hough (1998) Hough (1998) Hough (1998) Hough et al. (1990) Kirchner (1962) Kirchner (1962) Kirchner et al. (1960) Matthew (1986) Matthew (1986) Michaelis and Eysenck (1971) OConnell, Bott, Ramarkrishnan, and Doverspike (2003) Robie (2001) Robie (2001) Robie et al. (2001) Rosse et al. (1998) Smith (2001) Smith and Ellingson (2002) Smith, Hanges, and Dickson (2001) Stark et al. (2001) Tsaousis and Nikolaou (2001) Van Iddekinge et al. (2003) Weekley, Ployhart, and Harold (2003)

Volume 14 Number 4 December 2006

321

Indicates the studies that were used for analysis when article contained multiple studies. Non-sales jobs include managerial, clerical, manufacturing, public service, students, military, and computer programmers. E, extraversion; A, agreeableness; C, conscientiousness; O, openness; S, emotional stability (opposite of Neuroticism); Mgmt, management. b 16 PF, 16 personality factors; Gordon Pers. Prof., Gordon personal profile inventory; Guilford Martin, Guilford Martin inventory; Heron, Heron personality inventory; NEO, neuroticism, extraversion, openness; HPI, Hogan personality inventory; ABLE, assessment of background and life experiences; EPPS, Edwards personal preference schedule; PEN, psychoticism, extraversion, neuroticism; CPI, California psychological inventory; GPI, global personality inventory; PCI, personal change indicator.

322 SCOTT A. BIRKELAND, TODD M. MANSON, JENNIFER L. KISAMORE, MICHAEL T. BRANNICK AND MARK A. SMITH

each study. Two of the five authors recorded these statistics for all studies. Intra-class correlations (ICC; Shrout & Fleiss, 1979) were computed to examine the interrater reliability for each of the six statistics. For both the applicant and non-applicant sample sizes, the raters had perfect reliability [ICC(3, 2) 5 1.0]. Intraclass correlations [ICC(3, 2)] of .99 were obtained for both the applicant and non-applicant means. For the applicant standard deviations, the raters had perfect reliability [ICC(3, 2) 5 1.0]. For the non-applicant standard deviations, an ICC(3, 2) of .99 was obtained.

Analyses
The first step of the analysis was to compute d, the standardized mean difference for each of the scales included in each study. To avoid dependent effect sizes for studies providing more than one effect size within a Big Five dimension, effect sizes within a Big Five dimension were averaged within studies. Separate meta-analyses were computed for the extraversion, agreeableness, conscientiousness, openness to experience, and emotional stability dimensions. Because it was assumed that the effect sizes from each study were likely to differ in ways too numerous and complex to capture with a few simple study characteristics, a random effects meta-analytic model was chosen. The random-effects meta-analytic procedures adopted for this study were developed by Hedges and colleagues (e.g., Cooper & Hedges, 1994; Hedges & Olkin, 1985; Hedges & Vevea, 1998). Software for such methods is now widely available and accessible descriptions of the algorithms can be found in textbooks such as Lipsey and Wilson (2001). The method used for the current study is essentially equivalent to the Hunter and Schmidt (1990) bare bones analysis. Both methods (Hedges & colleagues; Hunter & Schmidt, 1990) compute a weighted mean effect size and estimate the variability of effect sizes after accounting for sampling error. There are minor technical differences in the estimation methods between the two algorithms. One can expect similar but not identical results from the two methods. Although the HunterSchmidt method can be used to account for sampling error alone, it is typically used to account for other artifacts, most often measurement error (reliability) and range restriction. Range restriction did not seem particularly applicable to this study. The applicant samples were not subject to selection, and if the incumbent samples were so restricted, it is likely that the restriction would work in opposition to our hypotheses, as noted earlier. Regarding corrections for the reliability of measurement, nine of the studies (36%) presented reliability data based on the measures taken in that study. The weighted mean internal consistency estimates were .77, .68, .74, .77, and .77 for extraversion, agreeableness, conscientiousness,

openness, and emotional stability, respectively. We used the reliability estimates to adjust the mean weighted effect sizes to show the standardized mean difference that would be expected if applicant and comparison groups responded to instruments with perfect internal consistency. To examine the heterogeneity of the five effect-size distributions, the Q value (i.e., the weighted sums of squares of the effect sizes) was computed. To test the significance of these Q values, the obtained Q value was compared with the critical value for a w2distribution with k1 degrees of freedom, where k is the number of studies. A statistically significant Q indicates a heterogeneous distribution (Lipsey & Wilson, 2001). If this Q statistic was significant (po.05), the proposed moderator variables were examined. To examine each proposed moderator variable, the analog to the ANOVA procedure described by Lipsey and Wilson (2001; pp.135138) was used. The mean effect size for each subgroup was computed. Then, the between and within-group Qs from the ANOVA were computed. Significance was set at p4.05. The total within-group Q and the within-level Qs were also examined to determine how much heterogeneity existed in the data after accounting for each moderator.

Results
The number of effect sizes obtained for each analysis is shown in Table 2. It can be seen that these frequencies differ from cell to cell. Overall, the agreeableness and openness dimensions contained fewer effect sizes than extraversion, conscientiousness, and emotional stability. In terms of the moderator variables, the job-type moderator variables (i.e., sales vs. non-sales and management vs. non-management) generally contained fewer effect sizes than the congruence and test-type moderator variables. The only cells that contained fewer than five effect sizes were the agreeableness dimension for the management job condition and the emotional stability dimension for the sales job condition.

Primary Analyses
Table 3 presents the results of the primary meta-analyses for each of the five personality dimensions. Results show that applicants scored significantly higher than nonapplicants across four of five dimensions, as evidenced by the positive lower bounds of the 95% confidence intervals. These results partially support Hypothesis 1a, which stated that applicants inflate their scores more than nonapplicants across all dimensions. As shown in Table 4, Hypothesis 1b was also supported. The effect sizes obtained from this study were significantly smaller than the effect sizes obtained from the Viswesvaran and Ones (1999) induced-faking study (p-values for t-tests were

International Journal of Selection and Assessment

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

JOB APPLICANT FAKING ON PERSONALITY MEASURES

323

less than or equal to .05)3 on all but the conscientiousness dimension. Comparing the uncorrected effect sizes of the five dimensions, conscientiousness and emotional stability, as shown in Table 3, produced larger effect sizes (d 5 .45 and .44), respectively, than extraversion, agreeableness, and openness (d 5 .11, .16, and .13), respectively, thus supporting Hypothesis 2. An examination of the 95% confidence intervals for the emotional stability dimension showed no overlap with the 95% confidence intervals for extraversion and openness. While there is a slight overlap between conscientiousness and agreeableness, there is no overlap in the 95% confidence intervals between conscientiousness and either extraversion or openness. Next, the heterogeneity of the five effect size distributions was computed. As shown in Table 3, the Q values for all five distributions were significant (po.01), indicating that the distribution of effect sizes for each dimension was

heterogeneous. Therefore, the proposed moderator variables were examined.

Moderator Analyses
Table 5 presents the correlations among the four moderator variables. As shown, no statistically significant correlations were observed among any of the moderator variables. Owing to the small k values (k ranged from 14 to 29 for the moderator correlation analyses), however, the power of the analysis was low. Tables 69 present the results of the moderator analyses. Sales Jobs vs. Non-Sales Jobs. The first study characteristics variable examined was the sales vs. non-sales job-type moderator. It was hypothesized that the extraversion effect sizes would be larger for sales jobs than for nonsales jobs (Hypothesis 3), indicating that applicants for sales jobs inflate their extraversion scores more than do applicants for non-sales jobs. As shown in Table 6, the between-groups Q for the extraversion dimension was significant (po.01) indicating the presence of a moderator. An examination of the sales group vs. non-sales group effect sizes indicates that the sales group demonstrated larger mean differences than did the non-sales group. Thus, Hypothesis 3 was supported. The between groups Q for the agreeableness dimension was also significant (po.05). Examination of the mean effect sizes indicate that sales applicants tended to deflate their scores on agreeableness, while non-sales applicants tended to inflate their scores along this dimension. Management Jobs vs. Non-Management Jobs. Hyothesis 4 stated that the conscientiousness effect sizes would be larger for studies consisting of management jobs than studies consisting of non-management jobs. As indicated by the between-groups Q value in Table 7, the results did not support this hypothesis. In fact, an examination of the mean effect size for non-management sample studies was larger than the mean effect size for management sample studies. Additionally, none of the other four Big Five dimensions were moderated by this variable.

Table 2. Number of effect sizes used in primary and moderator analyses E Total Sales Sales Non-sales Management Management Non-management Test type Direct Big Five Indirect Big Five Congruence Matched Unmatched 29 6 17 7 17 11 18 16 13 A 20 5 9 4 11 9 11 10 10 C 27 6 14 6 15 12 15 14 13 O 20 6 9 5 10 8 12 11 9 S 25 3 17 7 13 10 15 14 11

E, extraversion; A, agreeableness; C, conscientiousness; O, openness; S, emotional stability.

Table 3. Primary meta-analysis results for Big Five personality dimensions Construct Extraversion Agreeableness Conscientiousness Openness Emotional stability
*

K 29 20 27 20 25

Applicant (N) 53,745 27,842 69,325 46,037 21,219

Non-applicant (N) 18,096 16,126 18,941 14,224 13,991

Mean d .11* .16 .45** .13* .44**

95% Confidence Interval .006.217 .010 to .324 .303.591 .014.243 .278.593

^ d .13 .19 .52 .15 .50

REVC .0709 .1356 .1332 .0585 .1464

Q 667.7** 1057.5** 1337.1** 409.41** 935.0**

po.05,**po.01.REVC, random effects variance component; Q, w2 test of effect size heterogeneity; ^ d, mean d adjusted for reliability.

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

Volume 14 Number 4 December 2006

324 SCOTT A. BIRKELAND, TODD M. MANSON, JENNIFER L. KISAMORE, MICHAEL T. BRANNICK AND MARK A. SMITH

Table 4. Comparison of present meta-analysis and Viswesvaran and Ones (1999) meta-analyses Present investigation effect sizes Extraversion Agreeableness Conscientiousness Openness Emotional stability
a

Viswesvaran and Ones (1999) BSD effect sizes .63 .48 .60 .65 .64

Viswesvaran and Ones (1999) WSD effect sizes .54 .47 .89 .76 .93

Analog to t-test comparisona o .01 o .05 .10 o .05 .05

.111 .157 .446 .128 .435

p-values of analog to t-test comparison between present investigation effect sizes and Viswesvaran and Ones (1999) between-subject effect sizes. See endnote for explanation of this comparison. BSD, between-subject design; WSD, within-subject design.

Table 5. Moderator variable inter-correlations 1 Extraversion 1. Sales vs. Non-sales 2. Mgmt vs. Non-mgmt 3. Test type 4. Congruence Agreeableness 1. Sales vs. Non-sales 2. Mgmt vs. Non-mgmt 3. Test type 4. Congruence Conscientiousness 1. Sales vs. Non-sales 2. Mgmt vs. Non-mgmt 3. Test type 4. Congruence Openness 1. Sales vs. Non-sales 2. Mgmt vs. Non-mgmt 3. Test type 4. Congruence Emotional stability 1. Sales vs. Non-sales 2. Mgmt vs. Non-mgmt 3. Test type 4. Congruence
*

Table 6. Results of sales vs. non-sales moderator analysis Overall Total Between- Withink groups Q groups Q REVC 23 14 20 15 20 7.6** 5.1* 2.6 .0 2.1 36.6* 14.3 30.6* 22.0 18.6 .0483 .1195 .1025 .0341 .1940 Withinlevel Q 2.73 33.88** .95 13.35* 6.43 24.19* 6.63 15.34* 3.06 15.42

.18 .13 .18 .47 .47 .14 .19 .19 .19 .29 .19 .19 .02 .03 .28

.21 .13

.30

Extraversion Agreeableness Conscientiousness Openness Emotional stability

.43 .21

.30

By level Extraversion Sales position Non-sales position

k 6 17 5 9 6 14 6 9 3 17

Mean .347* .011 .204 .266* .125 .397** .161 .155 .009 .410

SE .106 .061 .170 .119 .143 .091 .093 .067 .267 .111

.22 .22

.33

.11 .21

.29

Agreeableness Sales position Non-sales position Conscientiousness Sales position Non-sales position Openness Sales position Non-sales position Emotional stability Sales position Non-sales position
*

.25 .21

.26

Indicates significant at po.05,**Indicates significant at po.01. Mgmt, management.

Indicates significant at po.05**Indicates significant at po.01. REVC, random effects variance component.
agreeableness, conscientiousness, openness, and emotional stability. The between-group Q values for these dimensions were all significant at the po.05 level. An examination of the mean effect sizes for each level of the moderator indicates that the studies using direct Big Five measures

Test Type. The third moderator variable examined whether direct measures of the Big Five personality dimensions (e.g., NEO) produced different effect sizes than did indirect measures. The results, shown in Table 8, indicate that test type was a significant moderator for

International Journal of Selection and Assessment

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

JOB APPLICANT FAKING ON PERSONALITY MEASURES

325

Table 7. Results of management vs. non-management moderator analysis Overall Extraversion Agreeableness Conscientiousness Openness Emotional stability Total Between- Withink groups Q groups Q REVC 24 15 21 15 20 .1 1.9 2.3 .0 .1 28.5 18.5 37.2 18.9 21.8 .0948 .1195 .0909 .0414 .1811 Withinlevel Q 2.44 26.04* 3.22 15.31 11.8 25.39* 4.94 13.95 5.40 16.44
*

Table 8. Results of test-type moderator analysis Overall Extraversion Agreeableness Conscientiousness Openness Emotional stability Total Between- Withink groups Q groups Q REVC 29 20 27 20 25 .86 16.9** 32.5** 4.0* 9.7** 43.7* 14.0 36.5 15.6 28.2 .0760 .114 .0730 .0668 .1318 Withinlevel Q 17.90 25.79 7.89 6.08 14.32 22.14 11.68 3.96 9.68 18.51

By level Extraversion Direct Big Five Indirect Big Five Agreeableness Direct Big Five Indirect Big Five Conscientiousness Direct Big Five Indirect Big Five Openness Direct Big Five Indirect Big Five Emotional stability Direct Big Five Indirect Big Five
*

k 11 18 9 11 12 15 8 12 10 15

Mean .175 .069 .506** .145 .794** .154 .275* .024 .722** .236*

SE .088 .071 .116 .108 .082 .077 .096 .081 .120 .010

By Level Extraversion Mgmt position Non-mgmt position Agreeableness Mgmt position Non-mgmt position Conscientiousness Mgmt position Non-mgmt position Openness Mgmt position Non-mgmt position Emotional stability Mgmt position Non-mgmt position
*

k 7 17 4 11 6 15 5 10 7 13

Mean .156 .107 .072 .217 .179 .415** .167 .150 .384 .329*
** *

SE .125 .081 .178 .111 .132 .084 .097 .074 .168 .124

Indicates significant at po.05, Indicates significant at po.01. REVC, random effects variance component; mgmt, management.

Indicates significant at po.05, **Indicates significant at po.01. REVC, random effects variance component.
ness than did unmatched groups. The between-group Q values for extraversion, agreeableness, emotional stability, and openness were extremely small, indicating that this study characteristic was not a significant moderator.

produced mean effect sizes that were significantly larger than the mean effect sizes for studies using indirect Big Five measures. The results also indicate that four of the five dimensions (agreeableness, conscientiousness, emotional stability, and openness) did not have significant within-group Q values (p4.05), indicating that, after accounting for test type, the distribution of effect sizes for these dimensions was homogeneous. Given the relatively small number of studies included in each of these cells, however, this result could be due to insufficient power. Applicant and Non-Applicant Congruence. The last moderator variable examined whether studies that had successfully matched subjects from both applicant and non-applicant groups produced different effect sizes than studies that did not match these subjects. Results from this analysis, shown in Table 9, indicate that the only dimension with a significant between-group Q value was conscientiousness (po.01). This indicated that matched groups demonstrated smaller mean differences on conscientious-

Discussion
Overall, the results of this meta-analysis are consistent with the hypothesis that individuals applying for jobs distort their scores on measures of personality to portray themselves in a positive manner. The degree to which they distort their scores, however, is (a) less than the degree that they distort scores when instructed to fake and (b) depends on the personality dimension being measured, the type of job, and the type of test.

Real-Life Faking vs. Induced Faking


To provide a context for interpreting the results of this investigation, it is helpful to compare the results obtained from this analysis with the results obtained from Viswes-

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

Volume 14 Number 4 December 2006

326 SCOTT A. BIRKELAND, TODD M. MANSON, JENNIFER L. KISAMORE, MICHAEL T. BRANNICK AND MARK A. SMITH

Table 9. Results of congruence moderator analysis Overall Extraversion Agreeableness Conscientiousness Openness Emotional stability Total Between- Withink groups Q groups Q REVC 29 20 27 20 25 .1 .4 7.2** 1.1 2.9 45.1* 25.3 30.7 20.0 29.7 .075 .1391 .1451 .0617 .1563 Withinlevel Q 20.9 24.2* 15.3 10.0 19.3 11.4 11.4 8.7 14.2 15.5

By level

Mean .129 .092 .102 .206 .244 .654 .188 .065 .309 .591

SE .0758 .0802 .1258 .1195 .1070 .1090 .0836 .0853 .1111 .1238

Extraversion Matched 16 Unmatched 13 Agreeableness Matched 10 Unmatched 10 Conscientiousness Matched 14 Unmatched 13 Openness Matched 11 Unmatched 9 Emotional stability Matched 14 Unmatched 11
*

Viswesvaran and Ones (1999) study found somewhat similar levels of faking across dimensions, especially in the between-subjects comparisons. In the within-subjects comparisons, conscientiousness and emotional stability demonstrated the largest effect sizes. In comparison, the present investigation found noticeably different levels of faking across dimensions, with applicants inflating their scores to a much larger degree on the conscientiousness and emotional stability dimensions. Therefore, it appears that in real-life settings respondents who have a motivation to present themselves in a desirable manner tend to be most concerned with inflating their scores on the conscientiousness and emotional stability dimensions. This suggests that job applicants might view these constructs as being particularly desirable by employers and, thus, focus their attention on inflating their scores on these dimensions. Future research should examine this issue more directly by investigating applicant perceptions of important personality dimensions in the job application process (e.g., Martin et al., 2002).

Job-Type Moderator Analyses


The results of the sales vs. non-sales job moderator analysis further support the assumption that job applicants may respond to personality measures in a way that increases their chances of getting a job. It was predicted that individuals applying for sales jobs would attempt to inflate their scores on extraversion scales, due to the skills perceived to be associated with sales success. This hypothesis was supported by the pattern of results. The mean effect size for the extraversion dimension was significantly higher for sales jobs than for non-sales jobs. In other words, it appears that the dimensions that sales applicants chose to distort were different from the dimensions that non-sales applicants chose to distort. It is also interesting to note that the agreeableness dimension was also moderated by job type. Here, applicants for sales jobs appeared to deflate their scores on the agreeableness dimension. At first glance, this result might seem counterintuitive in that sales people should want to come across as highly agreeable. A closer look at the agreeableness construct, however, reveals that scoring low on certain components of this construct might be viewed positively by someone applying for a sales job. For example, scoring low on the agreeableness dimension is related to being assertive; such a quality may prove helpful in closing a sale, and as a result, be perceived as a positive for someone who is applying for a sales job. In fact, a recent investigation by Zhao and Seibert (2006) found that entrepreneurs tend to score lower on measures of agreeableness than members of other occupations. Zhao and Seibert (2006) explained this finding by describing how high levels of agreeableness may inhibit ones willingness to drive hard bargains, look out for ones own self-interest, and influence or manipulate others for ones own advantage (p. 261). Given this, it is

Indicates significant at po.05,**Indicates significant at po.01. REVC, random effects variance component.

varan and Ones (1999) induced faking meta-analysis. As described in the introduction, the Viswesvaran and Ones study investigated the magnitude of the effect of faking by calculating the observed scale score differences between participants who were given instructions to fake-good compared with participants who were instructed to respond honestly. Thus, in a sense, the effect sizes found in the Viswesvaran and Ones meta-analysis represent the maximum degree of faking or intentional distortion one would expect on the Big-Five dimensions in a job application context. Conversely, the current investigation examined the degree of response distortion that occurs when participants are applying for a job and are not instructed to fake responses. In other words, our metaanalysis examined intentional distortion that occurs under realistic, employee selection conditions. Table 4 presents the overall uncorrected effect sizes from both metaanalyses. As expected, the uncorrected effect sizes from the current investigation were substantially less than those obtained from the induced faking study, especially for the extraversion, openness, and agreeableness dimensions. It is also interesting to compare the patterns of effect sizes across dimensions from the two studies. Overall, the

International Journal of Selection and Assessment

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

JOB APPLICANT FAKING ON PERSONALITY MEASURES

327

not surprising those sales applicants, who are in many respects similar to entrepreneurs, tended to deflate their scores on the agreeableness dimension. A similar moderator analysis was conducted for management jobs vs. non-management jobs. It was hypothesized that individuals applying for management jobs would attempt to inflate their scores on the conscientiousness dimension to a greater degree than individuals applying for non-management positions. The results of this moderator analysis, however, did not support the hypothesis. One possible explanation for this finding could be that conscientiousness is a trait that is desirable for persons in nearly all positions, not just management jobs. Overall, the results of the job-type moderator analyses, along with the results of the primary analysis, suggest that applicants appear to be distorting their responses on personality dimensions they view as particularly relevant to the specific job for which they are applying. Across most jobs, these dimensions appear to be conscientiousness and emotional stability. In some cases, however, such as sales jobs, the dimension might be extraversion, or even agreeableness. Thus, the results of this investigation are consistent with results from induced-faking studies (Furnham, 1990; Mahar et al., 1995; Martin et al., 2002) and with the position that applicants identify the personality characteristics that appear especially relevant to the job for which they are applying, and adjust their responses accordingly. Future research should examine this tendency in more types of jobs than was possible in this study. For example, applicants for customer service jobs may be particularly motivated to distort their responses on the agreeableness dimension.

Design Characteristic Moderator Analyses


In addition to the job-type analyses, two moderator analyses examining research design characteristics were also conducted. The first analysis examined the impact of study rigor. To operationalize study rigor, the congruence or match between the applicant and non-applicant groups was evaluated for each study. Results from this analysis suggest that, with one exception (conscientiousness), study rigor did not significantly moderate the standardized mean difference between the applicant and non-applicant group. This result is viewed positively because it suggests that the decision to include studies that may not have precisely matched subjects in the applicant and non-applicant conditions did not dramatically impact the overall results of the meta-analyses. The second research design moderator that we investigated was test type. Here, studies that used direct Big Five measures were compared with studies that used indirect measures of the Big Five. Studies using direct Big Five measures produced significantly higher mean effect sizes (po.05) on four of the five Big Five dimensions; the only dimension that did not show significance was extraversion.

Applicants appeared to inflate their scores substantially more on the direct Big Five measures than on the indirect Big Five measures. The finding that the direct Big Five measures show larger mean difference between applicants and non-applicants raises a number of important issues. First, it is interesting to note that even though direct Big Five measures appeared to show a higher prevalence of faking, as noted earlier, work by Salgado (2003) suggests that FFM instruments display stronger criterion-related validities than non-FFM instruments. One might anticipate lower validities for instruments that display higher levels of faking; however, results from this investigation are not consistent with such conjecture. Another possible explanation for this finding relates to the procedure we used for categorizing and analyzing the indirect Big Five scale scores. As previously stated, to categorize the indirect Big Five scale scores into the Big Five framework, we sorted each indirect Big Five scale score into one of the Big Five dimensions. If a particular study contained multiple indirect scale scores that mapped to the same Big Five construct, we took the average effect size for these indirect scale scores, following Barrick and Mount (1991). In essence, this procedure assumes that the constructs measured by indirect Big Five instruments are equivalent to the constructs measured by the direct Big Five instruments. This may be a faulty assumption. For example, it may be that a scale score from one of the indirect Big Five instruments is only measuring a specific facet of the particular Big Five dimension to which it was categorized. If this were true, then constructs we assumed to be equivalent were in fact not. At best, by assuming this equivalence, we and others who have used this procedure (e.g., Barrick & Mount, 1991; Viswesvaran and Ones, 1999) may have compared scores measuring constructs of different bandwidth. At worst, by assuming this equivalence, we may have compared scores that were, in fact, measuring completely different constructs or different facets of a construct. Whatever the explanation, this is a question that warrants further research. Regardless of the cause of these differences, our results, at a minimum, question the often-used practice of including both direct and indirect Big Five measures into the same analysis. Given that so much of the previous personality research has used this practice, it would be interesting to see how this issue impacts the results of these studies. One way to investigate this issue would be to sort the narrow bandwidth indirect Big Five scale scores into one of the Big Five facet dimensions, and then analyze the data using the Big Five facet constructs instead of the traditional Big Five dimensions. This procedure would help to guard against comparing narrow bandwidth scales with broad, Big Five constructs. The problem with this procedure would be obtaining an adequate sample size for each of the Big Five facet dimensions. As a start, however, it might be useful to only include studies that used direct Big Five measures and reported scores for each facet,

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

Volume 14 Number 4 December 2006

328 SCOTT A. BIRKELAND, TODD M. MANSON, JENNIFER L. KISAMORE, MICHAEL T. BRANNICK AND MARK A. SMITH

and investigate how specific facet dimensions within the same Big Five construct operate differently.

Implications, Future Research, and Limitations


The results of this meta-analysis have several implications for the use of personality testing in organizations. The finding that job applicants appear to purposely inflate their responses on personality measures suggests that more research is needed on the possible magnitude of and impact of such faking. Some research examining the influence of faking among job applicants on the criterion validity of personality measures has concluded that faking does not alter the criterion validities of the Big Five dimensions (Hough, 1998; Ones et al., 1996). Further research addressing this question is needed, however, because previous research has shown that faking may change the rank order of individual scores and, thus, hiring decisions (Hough, 1998; Rosse et al., 1998). One traditional method that has been used to address applicant faking involves the use of score corrections based on social desirability scales. Proper research studies have shown no evidence, however, that these score corrections can improve validity or accurately account for individual faking (Piedmont, McCrae, Riemann, & Angleitner, 2000). It is also important to note that researchers have recently begun to use within-subjects designs to measure individuallevel faking in the real world. This involves measuring the personality of individuals while they are job applicants and then measuring the same people while they are job incumbents. Presumably, the only difference between these situations is the pressure to present oneself positively. Some research has shown that the situation has little effect (Ellingson & Sackett, in press), while other research shows a detrimental effect of the testing situation on the construct validity of the scores (Smith & Canger, 2005). In addition, applicant faking may affect the factor structure of personality inventories another method of addressing construct validity. Recent research findings on this issue have shown conflicting results. Some have reported differences in the factor structure of Big Five measures from job applicants compared with non-applicants (e.g., Schmit & Ryan, 1993), while others have found a stable factor structure (e.g., Smith, Hanges, & Dickson, 2001). Similar studies using item-response theory (IRT) analyses of applicant and non-applicant responses have also reported similar conflicting results (Robie et al., 2001; Stark, Chernyshenko, Chan, Lee, & Drasgow, 2001). Another implication of our results is that personality test developers and/or organizations need to continue investigating ways to reduce applicant score distortion. One possibility might involve developing less transparent test items so that it is less obvious that faking an item will help increase their own attractiveness to an organization (e.g., Jackson et al., 2000; James, 1998). However, other research has essentially suggested the opposite using

more contextualized personality items (transparently related to work) increases the validity of personality measures (e.g., Bing, Whanger, Davidson, & Van Hook, 2004). Another possibility for reducing applicant distortion is warning applicants that faking can be identified in some way (McFarland, 2003). Recent research suggests that such warnings decrease the likelihood of applicant faking (Dwight & Donovan, 2003; Vasiliopoulos, Cucina, & McElreath, 2005; Vasilopoulos, McFarland, Cucina, & Ingerick, 2002). In addition, forced-choice measures of personality have the potential for simply not allowing an applicant to give an overly positive view of him/herself (Jackson et al., 2000; Martin et al., 2002). While some research has suggested that this approach minimizes the effects of applicant faking (e.g., Stark, Chernyshenko, & Drasgow, 2006), other research contradicts such finding (e.g., Heggestad, Morrison, Reeve, & McCloy, 2006). Thus, further research is clearly warranted. Also, other research has suggested that using the employee interview to measure personality can reduce the effects of faking (Van Iddekinge, Raymark, & Roth, 2005), although, again, further research is warranted in this area (Roth, Van Iddekinge, & Huffcutt, 2005). We must note that this meta-analysis utilized existing data and reviewed past research findings. Mainly, these studies have investigated applicant faking of personality measures at the group level; that is, looking at differences between groups of applicants and non-applicants. Response distortion, however, really is an individual level issue; that is, we are actually concerned with the extent to which individuals alter their scores between testing situations. Accordingly, McFarland and Ryan (2000) have suggested that multiple factors affect an individual applicants decision to fake. It may be somewhat misleading to assume that faking is constant across individuals in the applicant situation. In light of the current findings, it may be more accurate to assume that the Viswesvaran and Ones (1999) meta-analytic findings represent the extent of faking when everyone is equally motivated to fake. On the other hand, the current findings may represent the faking of only a subset of individuals. In addition, it was not possible in this meta-analysis to examine differential effects of self-deception and impression management as separate dimensions of social desirability (Li & Bagger, 2005; Paulhus, 1984). However, because there is no reason to expect that applicant and nonapplicant groups would differ on average in self-deception, it is assumed that the mean differences in personality test scores found in this study reflect the influence of impression management. Other limitations of this study also deserve mentioning. First, some levels of the moderator variables contained a small number of studies from which we based our results. Clearly, as more research is conducted in this area, it might be useful to re-analyze these data using additional studies. Second, for many of the indirect Big Five instruments, we

International Journal of Selection and Assessment

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

JOB APPLICANT FAKING ON PERSONALITY MEASURES

329

were forced to categorize scales into a Big Five construct based on the scale descriptions only, having no intercorrelation data to help in the sorting process. In the future, it might be useful to conduct an analysis using only direct Big Five measures.

Notes
1. In some cases, an instruments content might have been based on the Five Factor model of personality, but included scales that were not explicitly labeled as one of the Big Five personality dimensions. These measures were classified as indirect measures of the Big Five. 2. The file drawer problem refers to likelihood that some relevant studies may be excluded from a metaanalysis simply because those studies were not published or presented. Lack of publication is more likely to be the case for studies that did not find significant effects. 3. To determine if the effect sizes from our study were significantly smaller than the effect sizes from the Viswesvaran and Ones (1999) study, we conducted an analog of the t-test comparison for each Big Five dimension. To perform these comparisons, we first estimated the standard error for each of the Viswesvaran and Ones p effect sizes using the following formula: ObservedSD= k. Then to compute the standard error of the difference, we used the following formula: SEdiff q 2 SE2 1 SE2 :

With SE1 being the standard error of our effect size and SE2 being the standard error derived from the Viswesvaran and Ones (1999) study. The observed difference between these effect sizes was then divided by the SEdiff. The resulting value was then compared with the unit normal distribution to determine statistical significance.

References
Baker, T., Dwight, S.A. and Jones, M.W. (2000) Assessing differences in applicant and incumbent performance on a biodata test for sales positions. Presented at the 15th annual conference of the Society for Industrial Organizational Psychologists, New Orleans, U.S. Barrick, M.R. and Mount, M.K. (1991) The Big Five personality dimensions and job performance: A meta-analysis. Personnel Psychology, 44, 126. Bass, B.M. (1957) Faking by sales applicants of a forced choice personality inventory. Journal of Applied Psychology, 41, 403404. Bing, M.N., Whanger, J.C., Davidson, H.K. and Van Hook, J.B. (2004) Incremental validity of the frame-of-reference effect in personality scale scores: A replication and extension. Journal of Applied Psychology, 89, 150157. Cooper, H. and Hedges, L.V. (Eds). (1994) The handbook of research synthesis. New York: Russell Sage Foundation.

Costa, P.T. and McCrae, R.R. (1989) The NEO personality inventory manual. Odessa, FL: Psychological Assessment Resources. Digman, J.M. (1990) Personality structure: Emergence of the fivefactor model. Annual Review of Psychology, 41, 417440. * Dunnette, M.D., McCartney, J., Carlson, H.C. and Kirchner, W.K. (1962) A study of faking behavior on a forced-choice selfdescription checklist. Personnel Psychology, 15, 1324. Dwight, S.A. and Donovan, J.J. (2003) Do warnings not to fake reduce faking? Human Performance, 16, 123. Ellingson, J.E. and Sackett, P.R. (in press). Cross-situational consistency of personality scale scores in organizations. Journal of Applied Psychology, forthcoming. * Elliott A.P. (1976) Fakers: A study of managers responses on a personality test. Personality Review, 5, 3236. * Elliott, A.G.P. (1981) Some implications of lie scale scores in reallife selection. Journal of Occupational Psychology, 54, 916. Furnham, A. (1990) Faking personality questionnaires: Fabricating different profiles for different purposes. Current Psychology: Research and Reviews, 9, 4655. * Gordon, L.V. and Stapleton, E.S. (1956) Fakability of a forcedchoice personality test under realistic high school employment conditions. Journal of Applied Psychology, 40, 258262. * Green, R.F. (1951) Does a selection situation induce testees to bias their answers on interest and temperament tests? Educational and Psychological Measurement, 11, 503515. * Griffin, B. (2002). Applicants faking good: Item bias in the NEOPI conscientiousness scale. Poster presented at 17th annual conference of the Society for Industrial Organizational Psychology, Toronto, Canada. Hedges, L.V. and Olkin, I. (1985) Statistical methods for metaanalysis. Orlando, FL: Academic Press. Hedges, L.V. and Vevea, J.L. (1998) Fixed- and random-effects models in meta-analysis. Psychological Methods, 3, 486504. Heggestad, E.D., Morrison, M., Reeve, C.L. and McCloy, R.A. (2006) Forced-choice assessments of personality for selection: Evaluating issues of normative assessment and faking resistance. Journal of Applied Psychology, 91, 924. * Heron A. (1956) The effects of real-life motivation on questionnaire response. Journal of Applied Psychology, 40, 6568. * Hogan, J., Rybicki, S.L., Motowidlo, S.J. and Borman, W.C. (1998) Relations between contextual performance, personality, and occupational advancement. Human Performance, 11, 189207. * Hough, L.M. (1998) Effects of intentional distortion in personality measurement and evaluation of suggested palliatives. Human Performance, 11, 209244. * Hough, L.M., Eaton, N.K., Dunnette, M.D., Kamp, J.D. and McCloy, R.A. (1990) Criterion-related validities of personality constructs and the effect of response distortion on those validities. Journal of Applied Psychology Monograph, 75, 581595. Hunter, J.E. and Schmidt, F.L. (1990) Methods of meta-analysis: Correcting error and bias in research findings. Newbury Park, CA: Sage. Hurtz, G.M. and Donovan, J.J. (2000) Personality and job performance: The Big Five revisited. Journal of Applied Psychology, 85, 869879. Jackson, D.N., Wroblewski, V.R. and Ashton, M.C. (2000) The impact of faking on employment tests: Does forced choice offer a solution. Human Performance, 13, 371388. James, L.R. (1998) Measurement of personality via conditional reasoning. Organizational Research Methods, 1, 131163. * Kirchner, W.K. (1962) Real-life faking on the Edwards personal preference schedule by sales applicants. Journal of Applied Psychology, 46, 128130. * Kirchner, W.K., Dunnette, M.D. and Mousley, N. (1960) Use of the Edwards personal preference schedule in the selection of salesmen. Personnel Psychology, 13, 421424.

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

Volume 14 Number 4 December 2006

330 SCOTT A. BIRKELAND, TODD M. MANSON, JENNIFER L. KISAMORE, MICHAEL T. BRANNICK AND MARK A. SMITH

Li, A. and Bagger, J. (2005) Using the BIDR to distinguish the effects of impression management and self-deception on the criterion validity of personality measures: a meta-analysis. Paper presented at the Annual Conference of the Society for Industrial and Organizational Psychology, Los Angeles, U.S. Lipsey, M.W. and Wilson, D.B. (2001) Practical meta-analysis. London: Sage. Mahar, D., Cologon, J. and Duck, J. (1995) Response strategies when faking personality questionnaires in a vocational selection setting. Personality and Individual Differences, 18, 605609. Martin, B.A., Bowen, C.C. and Hunt, S.T. (2002) How effective are people at faking on personality questionnaires? Personality and Individual Differences, 32, 247256. * Matthew G.V. (1986) Effect of selection situations on performance in personality inventory. Indian Journal of Applied Psychology, 23, 3538. McFarland, L.A. (2003) Warning against faking on a personality test: Effects on applicant reactions and personality test scores. International Journal of Selection and Assesssment, 11, 265276. McFarland, L.A. and Ryan, A.M. (2000) Variance in faking across noncognitive measures. Journal of Applied Psychology, 85, 812821. * Michaelis, W. and Eysenck, H.J. (1971) The determination of personality inventory factor patterns and intercorrelations by changes in real-life motivation. The Journal of Genetic Psychology, 118, 223234. Mueller-Hanson, R.A., Heggestad, E.D. and Thornton, G.C. (2002). The impact of faking on selection decisions. Poster Presented at the 17th annual conference of the Society for Industrial Organizational Psychology, Toronto, Canada, April. * OConnell, M.S., Bott, J.P., Ramarkrishnan, M. and Doverspike, D. (2003) Practical implications of applicant versus incumbent differences on personality scales. Poster Presented at the 18th annual conference of the Society for Industrial Organizational Psychology, Orlando, U.S., April. Ones, D.S., Viswesvaran, C. and Reiss, A.D. (1996) Role of social desirability in personality testing for personnel selection: The red herring. Journal of Applied Psychology, 81, 660679. Orpen, C. (1971) Fakability of the Edwards Personnel Preference Schedule in personnel selection. Personnel Psychology, 24, 14. Paulhus, D.L. (1984) Two-component models of socially desirable responding. Journal of Personality and Social Psychology, 46, 598609. Piedmont, R.L., McCrae, R.R., Riemann, R. and Angleitner, A. (2000) On the invalidity of validity scales: Evidence from selfreports and observer ratings in volunteer samples. Journal of Personality and Social Psychology, 78, 582593. * Robie, C. (2001) Faking and personality measurement. Unpublished Technical Report, Personnel Decisions International. Robie, C., Schmit, M.J., Ryan, A.M. and Zickar, M.J. (2000) Effects of item context specificity on the measurement equivalence of a personality inventory. Organizational Research Methods, 3, 348365. * Robie, C., Zickar, M.J. and Schmit, M.J. (2001) Measurement equivalence between applicant and incumbent groups: An IRT analysis of personality scales. Human Performance, 14, 187 207. * Rosse, J.G., Stecher, M.D., Miller, J.L. and Levin, R.A. (1998) The impact of response distortion on preemployment personality testing and hiring decisions. Human Performance, 83, 634644. Roth, P.L., Van Iddekinge, C.H. and Huffcutt, A.I. (2005) Personality saturation in structured interviews. International Journal of Selection and Assessment, 13, 261273. Rusmore, J.T. (1956) Fakability of the Gordon Personal Profile. Journal of Applied Psychology, 40, 175177.

Salgado, J.F. (2003) Predicting job performance using FFM and nonFFM personality measures. Journal of Occupational and Organizational Psychology, 76, 323346. Schmit, M.J. and Ryan, A.M. (1993) The Big Five in personnel selection: Factor structure in applicant and non-applicant populations. Journal of Applied Psychology, 78, 966974. Schmit, M.J., Ryan, A.M., Stierwalt, S.L. and Powell, A.B. (1995) Frame-of-reference effects on personality scale scores and criterion-related validity. Journal of Applied Psychology, 80, 607620. Schwab, D.P. and Packard, G.L. (1973) Response distortion on the Gordon Personal Inventory and the Gordon Personal Profile in a selection context: Some implications for predicting employee tenure. Journal of Applied Psychology, 58, 372374. Shrout, P.E. and Fleiss, J.L. (1979) Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin, 86, 420428. * Smith, M. (2001). Incumbent versus applicant scores on the CFI Personality Inventory. Unpublished Technical Report, TMP Worldwide. Smith, M.A. and Canger, J.M. (2005) Social desirability scales and faking: A within-person, longitudinal study. In D.S. Ones and C. Viswesvaran (Eds), Usefulness of social desirability scales, faking scores, and potential alternatives. Symposium conducted at the 20th Annual Conference of the Society for Industrial and Organizational Psychology. * Smith, D.B. and Ellingson, J.E. (2002) Substance versus style: A new look at social desirability in motivating contexts. Journal of Applied Psychology, 87, 211219. * Smith, B.D., Hanges, P.J. and Dickson, M.W. (2001) Personnel selection and the five-factor model: Reexamining the effects of applicants frame of reference. Journal of Applied Psychology, 86, 304315. Stark, S., Chernyshenko, O. and Drasgow, F. (2006) Toward increased utilization of multidimensional forced choice measures in personality assessment. In Symposium (Chair: Chernyshenko), Innovative response formats in personality assessments: Psychometric and validity investigations. Presented at 21st Annual SIOP Conference: Dallas, U.S., May. Stark, S., Chernyshenko, O.E, Chan, K.L, Lee, W.C. and Drasgow, F. (2001) Effects of the testing situation on item responding: Cause for concern. Journal of Applied Psychology, 86, 943953. Tett, R.P., Jackson, D.N. and Rothstein, M. (1991) Personality measures as predictors of job performance: A meta-analytic review. Personnel Psychology, 44, 703742. * Tsaousis, I. and Nikolaou, I.E. (2001) The stability of the FiveFactor model of personality in personnel selection and assessment in Greece. International Journal of Selection and Assessment, 9, 290301. * Van Iddekinge, C.H., Raymark, P.H., Eidson, C.E. and Putka, D.J. (2003) Applicantincumbent differences on personality, integrity, and customer service measures. Poster Presented at the 18th annual conference of the Society for Industrial Organizational Psychology, Orlando, U.S. Van Iddekinge, C.H., Raymark, P.H. and Roth, P.L. (2005) Assessing personality with a structured employment interview: Construct-related validity and susceptibility to response inflation. Journal of Applied Psychology, 90, 536552. Vasiliopoulos, N.L., Cucina, J.M. and McElreath, J.M. (2005) Do warnings of response verification moderate the relationship between personality and cognitive ability? Journal of Applied Psychology, 90, 306322. Vasilopoulos, N.L., McFarland, L.A., Cucina, J.M. and Ingerick, M. (2002) A comparison of measures of faking across types of warnings and levels of job familiarity. Paper presented at the 17th annual conference of the Society for Industrial/Organizational Psychology, Toronto, Canada.

International Journal of Selection and Assessment

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

JOB APPLICANT FAKING ON PERSONALITY MEASURES

331

Vasilopoulos, N.L., Reilly, R.R. and Leaman, J.A. (2000) The influence of job familiarity and impression management on selfreport measure scale scores and response latencies. Journal of Applied Psychology, 85, 5064. Viswesvaran, C. and Ones, D.S. (1999) Meta-analysis of fakability estimates: Implications for personality measurement. Educational and Psychological Measurement, 59, 197210. * Weekley, J.A., Ployhart, R.E. and Harold, C.M. (2003) Personality and situational judgment tests across applicant and incumbent

settings. Poster Presented at 18th annual conference of the Society for Industrial Organizational Psychology, Orlando, U.S., April. Weiner, J.A. and Gibson, W.M. (2000) Practical effects of faking on job applicant attitude test scores. Poster Presented at the 15th annual conference of the Society for Industrial Organizational Psychology, New Orleans, U.S. Zhao, H. and Seibert, S.E. (2006) The Big Five personality dimensions and entreprenurial status: A meta-analytical review. Journal of Applied Psychology, 91, 259271.

Appendix A
Study ID Author Inventory Adjective checklist Adjective checklist 16 PF Scalea Assertiveness Assertiveness Reserved Submissive Sober Shy Reserved Submissive Sober Shy Extraversion Extraversion Sociability Ascendancy G S T A R Sociability Sociability Influence Dominance Energy level Exhibition Affiliation Dominance Exhibition Affiliation Dominance Exhibition Affiliation Dominance Gregariousness Gregariousness E Sociability Social presence Dominance Capacity for status Empathy Self-acceptance Extraversion Test type Indirect Indirect Indirect

Extraversion 1.1 Dunnette et al. (1962) 1.2 Dunnette et al. (1962) 2.1 Elliott (1976)

2.2

Elliott (1976)

16 PF

Indirect

3.1 3.2 4 5

Elliott (1981)b Elliott (1981) Gordon and Stapleton (1956) Green (1951)

16 PF 16 PF Gordon Per. Prof Guilford Martin

Indirect Indirect Indirect Indirect

7 8 9.2 10 11.1

Heron (1956) Hogan et al. (1998) Hough (1998) Hough et al. (1990) Kirchner (1962)

Heron HPI Hough ABLE EPPS

Indirect Direct Indirect Indirect Indirect

11.2

Kirchner (1962)

EPPS

Indirect

12

Kirchner et al. (1960)

EPPS

Indirect

13.1 13.2 14 16.1

Matthew (1986) Matthew (1986) Michaelis and Eysenck (1971) Robie (2001)

Matthew temp Matthew temp PEN CPI

Indirect Indirect Indirect Indirect

16.2

Robie (2001)

GPI

Indirect

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

Volume 14 Number 4 December 2006

332 SCOTT A. BIRKELAND, TODD M. MANSON, JENNIFER L. KISAMORE, MICHAEL T. BRANNICK AND MARK A. SMITH

Appendix. Continued
Study ID Author Inventory PPI Modified NEO Scalea Influence Warmth Gregariousness Excitement seeking Positive emotions Extraversion Likes parties Entertaining Experience setting Likes crowds Likes parties Entertaining Experience setting Likes crowds Dominance Liveliness Social boldness Privateness Extraversion Extraversion Extraversion Cooperativeness Cooperativeness Tough minded Natural Tough minded Natural Likeability Cooperativeness Deference Abasement Succorance Nurturance Deference Abasement Succorance Nurturance Nurturance Succorance Abasement Agreeableness Tolerance Agreeableness Trust Straightforwardness Altruism Compliance Tendermindedness Agreeableness Easy to live with Caring Sensitive Likes people Test type Direct Direct

Extraversion continued 17 Robie et al. (2001) 18 Rosse et al. (1998)

19 20

Smith (2001) Smith and Ellingson (2002)

TMPS HUGO HPI

Direct Direct

21

Smith, Hanges, and Dickson (2001)

HPI

Direct

22

Stark et al. (2001)

16 PF

Indirect

23 24 25

Tsaousis and Nikolaou (2001) Van Iddekinge et al. (2003) Weekley, Ployhart, and Harold (2003)

NEO (Greek ver.) PCI Weekly Adjective checklist Adjective checklist 16 PF 16 PF HPI ABLE EPPS

Direct Direct Direct Indirect Indirect Indirect Indirect Indirect Indirect Indirect

Agreeableness 1.1 Dunnette et al. (1962) 1.2 Dunnette et al. (1962) 2.1 Elliott (1976) 2.2 8 10 11.1 Elliott (1976) Hogan et al. (1998) Hough et al. (1990) Kirchner (1962)

11.2

Kirchner (1962)

EPPS

Indirect

12

Kirchner et al. (1960)

EPPS

Indirect

15 16.1 16.2 18

OConnell et al. (2003) Robie (2001) Robie (2001) Rosse et al. (1998)

OConnell/Hattrup CPI GPI Modified NEO

Direct Indirect Indirect Direct

19 20

Smith (2001) Smith and Ellingson (2002)

TMPs HUGO HPI

Direct Direct

International Journal of Selection and Assessment

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

JOB APPLICANT FAKING ON PERSONALITY MEASURES

333

Appendix. Continued
Study ID Author HPI Inventory Scalea Easy to live with Caring Sensitive Likes people Warmth Sensitivity Agreeableness Agreeableness Agreeableness Imaginativeness Imaginativeness Less intelligent Practical Conservative Less intelligent Practical Conservative Intellectance Adaptability Openness Change Change Change Intellectual efficiency Openness to change Creativity Actions Openness Reading Education Culture Good memory Reading Education Culture Good memory Openness to change Abstractness Openness to change Openness Calmness Calmness Emotionally unstable Trusting Self-assured Relaxed Emotionally unstable Trusting Self-assured Relaxed Test type Direct

Agreeableness continued 21 Smith, Hanges, and Dickson (2001)

22 23 24 25 Openness 1.1 1.2 2.1

Stark et al. (2001) Tsaousis and Nikolaou (2001) Van Iddekinge et al. (2003) Weekley, Ployhart, and Harold (2003) Dunnette et al. (1962) Dunnette et al. (1962) Elliott (1976)

16 PF NEO (Greek vers) PCI Weekly Adjective checklist Adjective checklist 16 PF

Indirect Direct Direct Direct Indirect Indirect Indirect

2.2

Elliott (1976)

16 PF

Indirect

8 9.2 9.3 11.1 11.2 12 16.1 16.2 17 18 19 20

Hogan et al. (1998) Hough (1998) Hough (1998) Kirchner (1962) Kirchner (1962) Kirchner et al. (1960) Robie (2001) Robie (2001) Robie et al. (2001) Rosse et al. (1998) Smith (2001) Smith and Ellingson (2002)

HPI Hough Hough EPPS EPPS EPPS CPI GPI PPI Modified NEO TMPs HUGO HPI

Direct Indirect Indirect Indirect Indirect Indirect Indirect Indirect Indirect Direct Direct Direct

21

Smith, Hanges, and Dickson (2001)

HPI

Direct

22 23 24

Stark et al. (2001) Tsaousis and Nikolaou (2001) Van Iddekinge et al. (2003)

16 PF NEO (Greek vers) PCI Adjective checklist Adjective checklist 16 PF

Indirect Direct Direct Indirect Indirect Indirect

Emotional stability 1.1 Dunnette et al. (1962) 1.2 Dunnette et al. (1962) 2.1 Elliott (1976)

2.2

Elliott (1976)

16 PF

Indirect

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

Volume 14 Number 4 December 2006

334 SCOTT A. BIRKELAND, TODD M. MANSON, JENNIFER L. KISAMORE, MICHAEL T. BRANNICK AND MARK A. SMITH

Appendix. Continued
Study ID Author Inventory 16 PF 16 PF Gordon Pers. Prof Guilford Martin Scalea NPF NPF Emotional stability I N D C M Emotional maladjustment Adjustment Resilient self-esteem Self-esteem Emotional stability Maladjustment Maladjustment N Well-being Independence Self-control Neuroticism Emotional control Stress tolerance Optimism Angry hostility Depression Impulsiveness Emotional stability Not anxious No depression No guilt Self-confidence Not anxious No depression No guilt Self-confidence Emotional stability Tension Apprehension Neuroticism Emotional stability Conscientiousness Conscientiousness Expedient Poorly integrated Expedient Poorly integrated Responsibility Conscientiousness Ambition School success Prudence Dependability Test type Indirect Indirect Indirect Indirect

Emotional stability continued 3.1 Elliott (1981) 3.2 Elliott (1981) 4 Gordon and Stapleton (1956) 5 Green (1951)

7 8 9.3 10 13.1 13.2 14 16.1

Heron (1956) Hogan et al. (1998) Hough (1998) Hough et al. (1990) Matthew (1986) Matthew (1986) Michaelis and Eysenck (1971) Robie (2001)

Heron HPI Hough ABLE Matthew temp Matthew temp PEN CPI

Indirect Direct Indirect Indirect Indirect Indirect Indirect Indirect

17

Robie et al. (2001)

PPI

Direct

18

Rosse et al. (1998)

Modified NEO

Direct

19 20

Smith (2001) Smith and Ellingson (2002)

TMPs HUGO HPI

Direct Direct

21

Smith, Hanges, and Dickson (2001)

HPI

Direct

22

Stark et al. (2001)

16 PF

Indirect

23 24

Tsaousis and Nikolaou (2001) Van Iddekinge et al. (2003)

NEO (Greek vers.) PCI Adjective checklist Adjective checklist 16 PF 16 PF Gordon Pers. Prof. NEO long form HPI

Direct Direct Indirect Indirect Indirect Indirect Indirect Direct Direct

Conscientiousness 1.1 Dunnette et al. (1962) 1.2 Dunnette et al. (1962) 2.1 Elliott (1976) 2.2 4 6 8 Elliott (1976) Gordon and Stapleton (1956) Griffin (2002) Hogan et al. (1998)

9.1

Hough (1998)

Hough

Indirect

International Journal of Selection and Assessment

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

JOB APPLICANT FAKING ON PERSONALITY MEASURES

335

Appendix. Continued
Study ID Author Inventory Hough Hough ABLE EPPS Scalea Achievement Detail mindedness Work orientation Conscientiousness Endurance Achievement Order Endurance Achievement Order Achievement Order Consistency Endurance Thoughtfulness Thoughtfulness Conscientiousness Responsibility Ach via conformance Socialization Flexibility Conscientiousness Work focus Competence Order Dutifulness Achievement striving Self-discipline Deliberation Conscientiousness Moralistic Virtuous No hostility Mastery Moralistic Virtuous No hostility Mastery Rule-conscientiousness Vigilance Perfectionism Conscientiousness Conscientiousness Conscientiousness Test type Indirect Indirect Indirect Indirect

Conscientiousness continued 9.2 Hough (1998) 9.3 Hough (1998) 10 Hough et al. (1990) 11.1 Kirchner (1962)

11.2

Kirchner (1962)

EPPS

Indirect

12

Kirchner et al. (1960)

EPPS

Indirect

13.1 13.2 15 16.1

Matthew (1986) Matthew (1986) OConnell et al. (2003) Robie (2001)

Matthew temp Matthew temp OConnell/Hattrup CPI

Indirect Indirect Direct Indirect

16.2 17 18

Robie (2001) Robie et al. (2001) Rosse et al. (1998)

GPI PPI Modified NEO

Indirect Direct Direct

19 20

Smith (2001) Smith and Ellingson (2002)

TMPs HUGO HPI

Direct Direct

21

Smith, Hanges, and Dickson (2001)

HPI

Direct

22

Stark et al. (2001)

16 PF

Indirect

23 24 25
a

Tsaousis and Nikolaou (2001) Van Iddekinge et al. (2003) Weekley, Ployhart, and Harold (2003)

NEO (Greek vers.) PCI Weekly

Direct Direct Direct

For Inventory abbreviations, see Table 1 footnote. When multiple scales are indicated, scales were combined. Elliott used the 16 PF in both 1976 and 1981 but used Bipolar scales in 1976and the Global Factors in 1981.

r 2006 The Authors Journal compilation r Blackwell Publishing Ltd. 2006

Volume 14 Number 4 December 2006

Vous aimerez peut-être aussi