Content uploaded by Deniz S. Ones
Author content
All content in this area was uploaded by Deniz S. Ones on May 14, 2021
Content may be subject to copyright.
A preview of the PDF is not available
The Effects of Social Desirability and Faking on Personality and Integrity Assessment for Personnel Selection
Abstract and Figures
A review of the extant literature and new empirical research suggests that social desirability is not much of a concern in personality and integrity testing for personnel selection. In particular, based on meta-analytically derived evidence, it appears that social desirability influences do not destroy the convergent and discriminant validity of the Big Five dimensions of personality (Emotional Stability, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness). We also present new empirical evidence regarding gender and age differences in socially desirable responding. Although social desirability predicts a number of important work variables such as job satisfaction, organizational commitment, and supervisor ratings of training success, social desirability does not seem to be a predictor of overall job performance and is only very weakly related to specific dimensions of job performance such as technical proficiency (r = -.07) and personal discipline (r = .05). Large sample investigations of the moderating influences of social desirability in actual work settings indicate that social desirability does not moderate the criterion-related validities of personality variables or integrity tests. The criterion-related validity of integrity tests for overall job performance with applicant samples in predictive studies is .41. Controlling for social desirability in integrity or personality test scores leaves the operational validities intact, thereby suggesting that social desirability functions neither as a mediator nor as a suppressor variable in personality-performance and integrity-performance relations. Theoretical explanations of why social desirability does not influence criterion-related validity are reviewed.
Figures - uploaded by Deniz S. Ones
Author content
All figure content in this area was uploaded by Deniz S. Ones
Content may be subject to copyright.
... In the context of conscientiousness, a study by Ones and Viswesvaran (1998) found that this trait is negatively related to unethical behavior in the workplace. This means that individuals with high levels of conscientiousness are less likely to engage in behavior that violates integrity. ...
This study examines the relationship between conscientiousness and agreeableness with integrity non-compliance among police officers in the Perlis Contingent from 2018 to 2021. It aims to understand how officers interpret integrity, how conscientiousness influences their handling of integrity issues, and how agreeableness impacts their experiences and decisions. Using a qualitative approach, data was collected through document analysis, in-depth interviews, and focus group discussions. The study provides valuable insights into the dynamics of personality traits and integrity in law enforcement, offering evidence-based recommendations to enhance integrity management. These findings are essential for strengthening ethical conduct within the Perlis Contingent and the broader Malaysian police force, contributing to a more accountable and just society.
... The motivation for this research stems from the extensive literature on differences in values and social attitudes between men and women. For example, women have been shown to be more trustworthy, public-spirited, and likely to exhibit 'helping' behavior (Eagly and Crowley, 1986), vote based on social issues (Goertzel, 1983), score better on 'integrity tests' (Ones and Viswesvaran, 1998), take stronger stances on ethical behavior (Glover et al., 1997;Reiss and Mitra, 1998) and behave more generously when faced with economic decisions (Eckel and Grossman, 1998). Thereby, one may ask to which extent these differences transmit to public policies in societies where women are better represented, either politically or socially. ...
There is anecdotal evidence that countries with female leadership in policymaking
are more efficient in combating the Covid-19 pandemic. This paper studies whether
countries with high female representativeness in political and social layers respond
differently to the Covid-19 outbreak. We explore patterns at a cross-country level,
which enables us to consider the variation of gender implicated institutions. Our
findings indicate that it is women’s social representation, rather than female
political leadership, that has the potential to capture cross-country variation in
Covid-19 policy responses. Our study confirms that well-functioning and effective
institutions are not established from the top-down but rather from the bottom-up.
... Purported effects of individual variables on SDR are often contradictory from one study to the next. For example, a meta-analysis suggests that age and gender are only modestly related to SDR(Ones and Viswesvaran, 1998; see alsoBelli et al., 2001;Mesmer-Magnus et al., 2006). However, other research finds no gender gap and suggests that age is confounded with education, which actually may show either a negative or a positive relationship with SDR(Klassen et al., 1975;Presser and Traugott, 1992;Bernstein et al., 2001;Heerwig and McCabe, 2009). ...
Social desirability is a response bias affecting all interviewing methods, whereby people
express opinions which do not correspond to their true preferences, in order to avoid
being associated with socially undesirable opinions and behaviors. In this contribution,
we argue that people are more likely to exhibit social desirability to the extent that
they are ambivalent or have little involvement toward the issue at hand. To test these
propositions we analyze the data from a List Experiment, where Swiss students were
asked for their attitudes about welfare state policies. A List Experiment is an astute
experimental design in which some subjects are induced to speak their mind freely,
without fear of being blamed for having "socially deviant" opinions, while other
subjects are left exposed to social normative influence. Simple between-group
comparisons and recent refinements of the method allow to estimate the aggregate
i Maître d’enseignement et de recherche, Quartier UNIL-Mouline, Bâtiment Géopolis, CH–1015
Lausanne, lionel.marquis@unil.ch
4
proportion and the socio-political profile of "liars". The results of the experiment
indicate that between 8 and 18 percent of subjects probably "lied" when answering
about three items: about welfare entitlement (i.e., foreigners having the same rights to
social services as Swiss citizens), about job nondiscrimination (i.e., of foreign against
Swiss job seekers), and, most evidently, about tax fairness (i.e., more equity in tax
policy, "even if I or my family were to be taxed more heavily"). Biases of this
magnitude are relatively modest in comparison to other research (e.g., racial attitudes
in the US). However, our findings indicate that social desirability responding (SDR) is
more widespread among some categories represented in our sample of subjects. Most
notably, SDR exceeded 30 or 40 percent of individuals with little political knowledge, of
younger individuals, as well as of political science students, which is broadly consistent
with our hypotheses. Additional analyses suggest that subjects tend to have a limited
tolerance for lying; when confronted with a list of items, they seem to consider that
"one little lie" is acceptable, but not more than that. Promises and limitations of our
approach are discussed in the conclusion.
We compare modern machine learning techniques to ordinary least squares regression on out-of-sample operational validity, adverse impact, and dropped predictor counts within a common selection scenario: the prediction of job performance from a battery of diverse psychometrically-validated tests. In total, scores from 1.2 billion validation study participants were simulated to describe outcomes across 31,752 combinations selection system design and scoring decisions. The most consistently valuable improvement from adopting modern machine learning over traditional regression was from dropping predictors rather than by improving prediction. On average, modern machine learning improved prediction of performance from psychometric scale composites only when the ratio of sample size to scale count was less than approximately 3, although algorithm choice, predictor count, and selection ratio affected outcomes as well. We also simulated the effects of design choices when combining item scores, which showed consistent, superior predictive accuracy for several modern machine learning algorithms, especially elastic net and random forest, over traditional regression. Given these results, we suggest the potential of machine learning for employee selection is unlikely to be realized in selection systems focusing on the combination of scale composites from previously validated psychometric tests. Instead, it will be realized in unconventional design scenarios, such as the use of individual items to make multiple trait inferences, or with novel data formats like text, image, audio, video, and behavioral traces. We therefore recommend researchers focus on the potential value of MML in future selection contexts rather than continuing to focus on the current value of MML in current selection contexts.
Die Leistungsfähigkeit einer jeden Organisation hängt u. a. davon ab, dass professionelle eignungsdiagnostische Methoden zum Einsatz kommen. Sie sorgen dafür, dass die richtigen Menschen eingestellt und innerhalb des Unternehmens entsprechend ihrer Eignung platziert werden. Darüber hinaus unterstützen sie die Personalentwicklung. Der vorliegende Beitrag informiert über grundlegende Methoden der Eignungsdiagnostik und gibt dabei konkrete Hinweise zur Optimierung der eingesetzten Methoden. Überdies wird in einem eigenen Abschnitt der potenzielle Nutzen künstlicher Intelligenz in der Eignungsdiagnostik diskutiert.
Many psychologists still regard correlations with social desirability (SD) scales as evidence of the invalidity of measures, despite 20 years of research showing that this interpretation is usually unjustified. Although items or scales may be characterized as high or low in SD, there is little evidence that individuals differentially respond to this property when completing self-report questionnaires under normal instructional conditions. In an attempt to separate substance from style in SD scales, self-reports from 215 adult men and women were compared to the external criterion of spouse ratings on a range of personality traits in the domains of neuroticism, extraversion, and openness to experience. When self-reports were 'corrected' using scores from the Eysenck Personality Inventory Lie scale and the Marlowe-Crowne SD scale, validity coefficients decreased, rather than increased, in most cases. Both scales were shown to be substantively related to neuroticism and, to a lesser degree, to extraversion and closedness. These results suggest that correlations with SD scales should be given substantive rather than artifactual interpretations and that the widespread practice of correcting scores for lying, defensiveness, or SD should be questioned.
Three experiments examined impression-management responses to the Reid Report Integrity Attitude Inventory. Subjects encouraged to score high on the honesty test attained higher scores than a control group, but no higher than a group of job applicants. Study 2 offered money for high scores, and provided information to use concernign the first, second, or both factors of the Reid Report. Subjects in the three information conditions scored higher than those in a control condition, but again were no higher than job applicants. In a third study, subjects were asked to respond to the Reid Report and to several other measures as if they seriously wanted a job. After the test, each subject was overpaid for the participation. Reid Report scores were significantly correlated with returning versus retaining the money. These results suggest that integrity tests possess predictive validity despite spite some impression-management response distortion. Relations with other personality measures also are presented.
Three models are developed for the effects of social desirability (SD) on organizational behavior research results. SD can act as (a) an unmeasured variable that produces spurious correlations between study variables, (b) a suppressor variable that hides relationships, or (c) a moderator variable that conditions the relationship between two other variable.
The authors examined whether individuals can fake their responses to a personality inventory if instructed to do so. Between-subjects and within-subject designs were metaanalyzed separately. Across 51 studies, fakability did not vary by personality dimension; all the Big Five factors were equally fakable. Faking produced the largest distortions in social desirability scales. Instructions to fake good produced lower effect sizes compared with instructions to fake bad. Comparing meta-analytic results from within-subjects and between-subjects designs, we conclude, based on statistical and methodological considerations, that within-subjects designs produce more accurate estimates. Between-subjects designs may distort estimates due to Subject × Treatment interactions and low statistical power.
-30 social scientists and 50 undergraduate students responded to the Phase I1 Profile, a widely used honesty test, with separate instructions to (1) answer candidly and, after two weeks had elapsed, co (2) answer as if the test needed to be passed to obtain a desirable job. Both groups changed a substantial number of responses suggesting that the test may be susceptible to "faking good." In recent years, paper-and-pencil tests of honesty have emerged as a common preemployment screening device (Sackett & Harris, 1984; Sackett, Burris, & Callahan, 1989). In general, these tests attribute dishonest behaviors by employees to antisocial personality traits, yet define the construct of honesty or integrity as little more than conventional attitudes about the theft of property. Furthermore, these tests do not take into account external influences on employeesy behavior such as the work environment or the organizational culture (Moore & Stewart, 1989). This study extends the finding that 78% of undergraduates sampled by Logan, Koettel, and Moore (1986) failed the Phase I1 Profile test of honesty when it was administered solely for their own information in a setting free of rewards or threats. The vahdity and reliability correlations reported by the test developer were the highest included in the review by Sackett and Harris (1784). They were obtained from job applicants who were tested and later admitted to theft or other wrongdoing during interrogations using a polygraph. Yet, Logan, et al. (1986) were unable to find an association between scores on this profile and 16 PF factors that would ordinarily be associated with honest and dishonest behaviors. The national cut-off for failure on the profile was set at 45 to 50% based on the test developer's experience as a criminologist and polygrapher (Lousig-Nont & Associates, 1982). By comparison, the 78% failure rate by college undergraduates with work experience and no incentive to fake their responses (75 of 96 test takers: x,2 = 30, p< .001) in the Logan, et al. study suggests that the cut-off point is unduly high. Even if the test met rigorous standards of v&dity, the presumption that 50% of the populace is too dishonest to undertake gainful employment casts doubt on the test's ability to predict efficiently honest and dishonest behaviors on the job. While it is
Personality researchers are converging on the “Big Five?’ personality constructs as a taxonomy for describing the basic dimensions of personality. If, however, prediction rather than description is important, the Big Five may not be an adequate taxonomy. This article presents data that suggest the Big Five is not an adequate taxonomy of personality variables for predicting important criteria. The Big Five constructs are too heterogeneous and incomplete. A nine-factor taxonomy is presented along with correlational evidence demonstrating differential prediction of criteria for each of the nine personality constructs.
This study investigated the relation of the "Big Five" personality di- mensions (Extraversion, Emotional Stability, Agreeableness, Consci- entiousness, and Openness to Experience) to three job performance criteria (job proficiency, training proficiency, and personnel data) for five occupational groups (professionals, police, managers, sales, and skilled/semi-skilled). Results indicated that one dimension of person- ality. Conscientiousness, showed consistent relations with all job per- formance criteria for all occupational groups. For the remaining per- sonality dimensions, the estimated true score correlations varied by occupational group and criterion type. Extraversion was a valid pre- dictor for two occupations involving social interaction, managers and sales (across criterion types). Also, both Openness to Experience and Extraversion were valid predictors of the training proficiency criterion (across occupations). Other personality dimensions were also found to be valid predictors for some occupations and some criterion types, but the magnitude of the estimated true score correlations was small (p < .10). Overall, the results illustrate the benefits of using the 5- factor model of personality to accumulate and communicate empirical findings. The findings have numerous implications for research and practice in personnel psychology, especially in the subfields of person- nel selection, training and development, and performance appraisal.