Full Length Article
HEXACO personality predicts counterproductive work behavior and organizational citizenship behavior in low-stakes and job applicant contexts

https://doi.org/10.1016/j.jrp.2018.09.003Get rights and content

Highlights

  • Job applicants engage in moderate response distortion on HEXACO-PI-R.

  • Honesty-humility, extraversion, agreeableness, and conscientiousness predict lower CWB.

  • Honesty-humility, extraversion, agreeableness, and conscientiousness predict higher OCB.

  • Predictive validity shows small declines in applicant context.

Abstract

This study examined the degree to which the predictive validity of personality declines in job applicant settings. Participants completed the 200-item HEXACO Personality Inventory-Revised, either as part of confidential research (347 non-applicants) or an actual job application (260 job applicants). Approximately 18-months later, participants completed a confidential survey measuring organizational citizenship behavior (OCB) and counterproductive work behavior (CWB). There was evidence for a small drop in predictive validity among job applicants, however honesty-humility, extraversion, agreeableness, and conscientiousness predicted lower levels of CWB and higher levels of OCB in both job applicants and non-applicants. The study also informs the use of the HEXACO model of personality in selection settings, reporting typical levels of applicant faking and facet-level predictive validity.

Introduction

Personality testing for employee selection offers the potential to identify job applicants who will contribute positively to workplace culture and abstain from damaging behaviors such as fraud, theft, and harassment. However, job applicants are understandably motivated to make a positive impression on employers and tend to engage in response distortion when completing personality tests (Birkeland, Manson, Kisamore, Brannick, & Smith, 2006). This raises the question of whether job applicant response distortion reduces the predictive validity of personality measures in high stakes settings (Komar et al., 2008, Morgeson et al., 2007a, Morgeson et al., 2007b, Rothstein and Goffin, 2006).

Thus, the primary aim of the present study was to examine the effect of the job applicant context on the predictive validity of personality tests. A secondary aim was to examine the predictive validity of the broad and narrow traits of the HEXACO model of personality. To achieve these aims, we utilized two large existing databases of applicants and non-applicants that included responses for the HEXACO Personality Inventory - Revised. We engaged in follow-up measurement of these samples to measure their organizational citizenship behavior and counterproductive work behavior in a low-stakes confidential research setting. As will be discussed, this design overcomes some of the limitations of past literature on comparative predictive validity. In addition to the primary aim, the data also allowed for an examination of the predictive validity of the HEXACO model of personality, which is an increasingly popular alternative to the Big Five (Lee, Ashton, Morrison, Cordery, & Dunlop, 2008), particularly as it relates to unethical and deviant behavior (for a review see, Ashton, Lee, & de Vries, 2014). As we had data on the full 200-item measure of the HEXACO PI-R that provides reliable measurement of personality facets, the study also contributes to discussion about the relative merits of narrow traits in employee selection (Anglim et al., 2018, Ashton, 1998, Christiansen and Robie, 2011, Ones and Viswesvaran, 1996, Salgado et al., 2013, Tett et al., 2003).

The relationship between personality and employee behavior has generated considerable research in I/O psychology (for a review, see Barrick & Mount, 2012) with numerous meta-analyses that synthesize correlations between the Big Five and work performance (Barrick and Mount, 1991, Barrick et al., 2001, Hurtz and Donovan, 2000, Salgado, 1997, Tett et al., 1991). More recently, meta-analyses have focused on contextual performance outcomes such as organizational citizenship behavior (OCB), prosocial behavior which can benefit the organization or people within it, and counterproductive work behavior (CWB), behavior which can harm the organization or people within it. A meta-analysis of the correlations between self-rated Big Five personality and OCB obtained small but significant mean raw correlations for conscientiousness, agreeableness, emotional stability, extraversion and openness (Chiaburu, Oh, Berry, Li, & Gardner, 2011). With regard to CWB, a meta-analysis by Berry, Ones, and Sackett (2007) showed that low agreeableness and low conscientiousness were the traits most strongly related to overall CWB.

Researchers have also questioned whether the predictive validity of personality is retained when respondents are motivated and able to fake (Rothstein & Goffin, 2006). Respondents generally recognize which responses to personality test items are more socially desirable and do in fact respond in more socially desirable ways in job applicant settings (for a review, see Rothstein & Goffin, 2006). A meta-analysis by Birkeland et al. (2006) found that personality test scores were approximately half a standard deviation higher on conscientiousness and emotional stability in job applicant samples. A meta-analysis of instructed faking studies found average changes of around three-quarters of a standard deviation (Ones, Viswesvaran, & Schmidt, 1993). In addition to studies of instructed faking on HEXACO personality in the laboratory environment (Grieve and De Groot, 2011, MacCann, 2013), a recent study by Anglim, Morse, De Vries, MacCann, and Marty (2017) compared large samples of job applicants and non-applicants on the HEXACO-PI-R and found that job applicants scored higher on honesty-humility, extraversion, agreeableness, and conscientiousness.

Although there is relative consensus that response distortion occurs in applicant settings, there is less agreement about whether such settings also reduce the predictive validity of personality. Rothstein and Goffin (2006) concluded that while faking reduces the validity of personality testing, the predictive validity that remains is still sufficient to justify the use of personality testing. In contrast, Morgeson et al. (2007a) suggested that the evidence that the applicant context lowers predictive validity is inconclusive. They noted that relatively large samples are required to determine whether differences in predictive validity are significant, given the small observed correlations between personality and performance. Given that several studies have found that applicant personality predicts job performance (Barrick & Mount, 1996), a key issue is quantifying how much, if at all, predictive validity is reduced by applicant faking.

A range of empirical approaches have been taken to examine changes in predictive validity, and each has its limitations. First, early research showed that applying adjustments for faking based on impression management scales did not increase predictive validity (Barrick and Mount, 1996, Christiansen et al., 1994, Ones et al., 1996, Schmitt and Oswald, 2006). However, faking is better understood as the difference between a latent true score and the latent faked score (Peterson, Griffith, Isaacson, O'Connell, & Mangos, 2011). Impression management scales are contaminated with substantive variance and are not direct measures of the amount of faking, and therefore cannot be used to examine how predictive validity changes with faking (Anglim et al., 2017, De Vries et al., 2014, Mueller-Hanson et al., 2003, Uziel, 2010).

Second, several meta-analyses comparing the predictive validity of job applicant with non-applicant samples have shown that personality is predictive of performance outcomes in job applicant contexts. For example, Schmitt, Gooding, Noe, and Kirsch (1984) found average correlations between personality and supervisor performance ratings of 0.34 for concurrent designs using existing employees, and 0.30 for the predictive designs using applicants. Hough's (1997) meta-analysis indicated that predictive validities of personality were on average 0.07 higher in incumbent samples compared to applicant samples. In contrast, Tett et al. (1991) indicated that recruitment samples had a larger sample weighted validity correlation than incumbents (mean r of 0.20 versus 0.14), although 80% of the recruitment sample came from only one sample, the military Project A dataset (Campbell, 1990).

Finally, studies that have examined predictive validity by comparing applicants and non-applicants in a single study have had methodological limitations. Some research used laboratory designs (e.g., Douglas et al., 1996, Mueller-Hanson et al., 2003) where participants either answer a personality test honestly or under some form of instructed faking conditions. These studies often used university samples and used grade point average as an outcome for evaluating predictive validity. While job applicant role plays are a useful tool, researchers are generally most interested in how these results generalize to high-stakes settings, and for that, getting the responses of actual job applicants is preferred. Other studies compared predictive validity in actual job applicants with non-applicants such as incumbents (Anglim et al., 2018, Ellingson et al., 2007, Hough et al., 1990, Jeong et al., 2017, Mueller-Hanson et al., 2003). For example, Jeong et al. (2017) found that personality was less predictive of supervisor ratings in hired applicants than incumbents. Some studies comparing applicants and non-applicants have design confounds that compromise inferences about effects of the applicant context. For example, there has often been a time delay between applicant personality measurement and outcome measurement (i.e., predictive design) but no time delay for the non-applicants (i.e., incumbents, concurrent design). Differential range restriction between groups and other issues of group comparability represent additional challenges. Furthermore, many previous studies have had modest sample sizes and research on differential predictive validity across contexts benefits from large sample sizes. Thus, further research with larger samples that controls for these confounds is still needed.

Historically, personality research began with a proliferation of traits followed by the widespread acceptance that there were five broad personality traits (Costa and MacCrae, 1992, Goldberg, 1993). In addition to higher-order models (Digman, 1997, Musek, 2007, Van der Linden et al., 2010) and competing models of narrow facets (DeYoung et al., 2007, Paunonen and Ashton, 2001, Paunonen and Jackson, 2000, Roberts et al., 2005, Salgado et al., 2015), one of the biggest challenges to the Big Five comes from models that propose alternative broad traits (for a review, see Anglim & O’Connor, 2018). Prominently, Lee, Ashton, and colleagues developed the six factor HEXACO model based on extensive lexical studies in a range of languages (Ashton et al., 2014, Lee and Ashton, 2008). The HEXACO model largely retains extraversion, openness, and conscientiousness from the Big Five, but Big Five agreeableness and neuroticism are reconfigured and expanded into HEXACO emotionality, agreeableness, and honesty-humility. Initial studies using HEXACO personality to predict CWB have tended to identify Honesty–humility as a comparatively strong predictor of lower levels of CWB and other deviant behaviors (Ceschi et al., 2016, Chirumbolo, 2015, de Vries and van Gelder, 2015, Lee et al., 2005, Marcus et al., 2007, Oh et al., 2011).

Other research has examined whether narrow facets provide additional benefits in predicting workplace behavior (Anglim and Grant, 2014, Anglim et al., 2018, Cronbach and Gleser, 1965, Ones and Viswesvaran, 1996). In addition, several more recent studies examined incremental prediction of facets (de Vries et al., 2011, Jenkins and Griffith, 2004, Judge et al., 2013, Salgado et al., 2013, Salgado et al., 2015, Tett et al., 2003, Vinchur et al., 1998, Ziegler et al., 2014) and novel traits outside the five-factor model (Ashton, 1998, Conte and Gintoft, 2005, Crant, 1995, Lounsbury et al., 2004). Many of these studies suggest that inclusion of facets can increase prediction by about a third to a half. Nonetheless, existing research examining correlations between personality facets and OCB/CWB has rarely used the HEXACO model, and has instead used other personality frameworks such as the NEO-PI-R. Given the increased practitioner interest in the HEXACO framework, research is needed to understand how the narrow facets of the HEXACO model predict OCB and CWB. In particular, such research can guide decisions about how employers should weight the broad and narrow traits of the HEXACO model in selection systems. In may also enable the development of shorter and more efficient measures. From a theoretical perspective, understanding the relative importance of facets can provide a richer understanding of how predictive validity operates at the level of broad traits by highlighting particular narrow facets that are more or less important.

On the backdrop of the above, the main aim of the current study was to examine how the predictive validity of personality tests varies between job applicant and non-applicant settings. A secondary aim was to examine the predictive validity of HEXACO personality for OCB and CWB. Respondents were either job applicants or research participants who had completed the 200-item HEXACO-PI-R with six factors and 25 facets (for information on this large baseline sample, see Anglim et al., 2017). Respondents also completed a follow-up survey, which included measures of OCB and CWB, in a low-stakes testing environment (confidential research), approximately 18 months after initial testing. By having an extended period between initial personality assessment and the follow-up survey, we were able to ensure that applicants and non-applicants answered follow-up questions about OCB and CWB equally honestly. The design of the current study also overcomes some limitations with previous research. In particular, much previous research confounds the setting (applicant versus non-applicant) with the design (predictive versus concurrent). For example, non-applicants are commonly incumbents where outcomes are measured concurrently. In our design, all outcomes are measured in a time-lagged fashion.

Section snippets

Participants and procedure

The sample consisted of 607 participants (347 non-applicants; 260 job applicants). Note that we use the terms “non-applicants” and “job applicants” to distinguish how baseline personality was measured, even though both groups completed follow-up measurement in a non-applicant context. The time between original and follow-up measurement was generally over one year and similar for applicants (M = 1.6 years, SD = 1.1, range: 0.04–4.06) and non-applicants (M = 1.9 years, SD = 1.0, range: 0.1–3.81).

Preliminary validity checks

Before examining differences in predictive validity between job applicants and non-applicants and to attribute differences in predictive validity to the effect of the job applicant context, we verified whether responses by job applicants at follow-up were of approximately equivalent honesty as those of non-applicants. Applicants and non-applicants showed minimal differences on personality at follow-up, where mean absolute Cohen's d was 0.14 at follow-up compared to 0.50 at baseline. There were

Discussion

The current study assessed the degree to which the predictive validity of personality is attenuated when measured in a job applicant context, and provided a general assessment of the ability of the HEXACO model of personality to predict OCB and CWB. Results showed that when personality measures were completed for employee selection purposes, applicants responded in a more socially desirable manner. This was reflected in elevated means, smaller standard deviations, smaller correlations with

Open practices

Data, data analysis scripts, supplementary materials, and item-level information are provided at https://osf.io/wa6yj. The study was not preregistered.

Declarations of interest

The fifth author (AM) uses the HEXACO model of personality in a consulting context. The other authors have no interests to declare.

Contribution note

JA, LE, and AM were involved in study design and data collection. JA performed the data analysis. All authors contributed to writing of the manuscript.

References (74)

  • J. Anglim et al.

    Comparing job applicants to non-applicants using an item-level bifactor model on the HEXACO personality inventory

    European Journal of Personality

    (2017)
  • J. Anglim et al.

    Measurement and research using the Big Five, HEXACO, and narrow traits: A primer for researchers and practitioners

    Australian Journal of Psychology.

    (2018)
  • M.C. Ashton

    Personality and job performance: The importance of narrow traits

    Journal of Organizational Behavior

    (1998)
  • M.C. Ashton et al.

    The HEXACO honesty-humility, agreeableness, and emotionality factors a review of research and theory

    Personality and Social Psychology Review

    (2014)
  • M.R. Barrick et al.

    The big five personality dimensions and job performance: A meta-analysis

    Personnel Psychology

    (1991)
  • M.R. Barrick et al.

    Effects of impression management and self-deception on the predictive validity of personality constructs

    Journal of Applied Psychology

    (1996)
  • M.R. Barrick et al.

    Nature and use of personality in selection

  • M.R. Barrick et al.

    Personality and performance at the beginning of the new millennium: What do we know and where do we go next?

    International Journal of Selection and Assessment

    (2001)
  • R.J. Bennett et al.

    Development of a measure of workplace deviance

    Journal of Applied Psychology

    (2000)
  • C.M. Berry et al.

    Do other-reports of counterproductive work behavior provide an incremental contribution over self-reports? A meta-analytic comparison

    Journal of Applied Psychology

    (2012)
  • C.M. Berry et al.

    Interpersonal deviance, organizational deviance, and their common correlates: A review and meta-analysis

    Journal of Applied Psychology

    (2007)
  • S.A. Birkeland et al.

    A meta-analytic investigation of job applicant faking on personality measures

    International Journal of Selection and Assessment

    (2006)
  • J.P. Campbell

    An overview of the army selection and classification project (Project A)

    Personnel Psychology

    (1990)
  • A. Ceschi et al.

    Grit or honesty-humility? New insights into the moderating role of personality between the health impairment process and counterproductive work behavior

    Frontiers in psychology

    (2016)
  • D.S. Chiaburu et al.

    The five-factor model of personality traits and organizational citizenship behaviors: A meta-analysis

    Journal of Applied Psychology

    (2011)
  • A. Chirumbolo

    The impact of job insecurity on counterproductive work behaviors: The moderating role of honesty–humility personality trait

    The Journal of psychology

    (2015)
  • N.D. Christiansen et al.

    Correcting the 16pf for faking: Effects on criterion-related validity and individual hiring decisions

    Personnel Psychology

    (1994)
  • N.D. Christiansen et al.

    Further consideration of the use of narrow trait scales

    Canadian Journal of Behavioural Science

    (2011)
  • J.M. Conte et al.

    Polychronicity, Big Five personality dimensions, and sales performance

    Human Performance

    (2005)
  • Costa, P. T., & MacCrae, R. R. (1992). Revised NEO personality inventory (NEO PI-R) and NEO five-factor inventory (NEO...
  • J.M. Crant

    The Proactive Personality Scale and objective job performance among real estate agents

    Journal of Applied Psychology

    (1995)
  • L.J. Cronbach et al.

    Psychological tests and personnel decisions

    (1965)
  • A. de Vries et al.

    Broad versus narrow traits: Conscientiousness and honesty-humility as predictors of academic criteria

    European Journal of Personality

    (2011)
  • R.E. De Vries et al.

    Rethinking trait conceptions of social desirability scales: Impression management as an expression of honesty-humility

    Assessment

    (2014)
  • C.G. DeYoung et al.

    Between facets and domains: 10 aspects of the Big Five

    Journal of Personality and Social Psychology

    (2007)
  • J.M. Digman

    Higher-order factors of the Big Five

    Journal of Personality and Social Psychology

    (1997)
  • Douglas, E. F., McDaniel, M. A., & Snell, A. F. (1996). The validity of non-cognitive measures decays when applicants...
  • Cited by (42)

    • Identifying social courage antecedents and mediating effects: Applying the HEXACO and approach/avoidance frameworks

      2021, Personality and Individual Differences
      Citation Excerpt :

      These six dimensions are: honesty-humility, which includes the facets of sincerity, fairness, greed avoidance, and modesty; emotionality, which includes the facets of fearfulness, anxiety, dependence, and sentimentality; extraversion, which includes the facets of social self-esteem, social boldness, sociability, and liveliness; agreeableness, which includes the facets of forgivingness, gentleness, flexibility, and patience; conscientiousness, which includes the facets of organization, diligence, perfectionism, and prudence; and openness, which includes the facets of aesthetic appreciation, inquisitiveness, creativity, and unconventionality. Since its creation, many studies have supported the validity of the HEXACO framework and showed that the HEXACO dimensions uniquely predict outcomes when all are simultaneously included in analyses (Anglim et al., 2018; Ashton & Lee, 2005, 2007, 2013; Lee & Ashton, 2014; Lee, Ashton, et al., 2005; Lee et al., 2019; Louw et al., 2016). These outcomes include core performance, organizational citizenship, and voice behaviors; however, no researcher has linked the HEXACO framework with social courage.

    View all citing articles on Scopus
    View full text