Barrick and Mount's meta-analysis on the relation between personality and job performance is considered a seminal article in the field of personnel selection. This study and others that followed it (1) have led to a general acceptance of the five-factor model of personality and its usefulness in predicting performance for all jobs. As a result, the use of personality tests as alternative selection devices in personnel selection has been advocated, specifically for selection of police officers. (2)
Meta-analysis has been the foundation for theory and conclusions regarding the predictive validity of personality tests. (3) Often overlooked, however, are the practicalities of translating meta-analytic results to the practice of personnel selection. We argue in this paper that practitioners must be cautious in applying meta-analytic results to actual applicant settings. General problems with meta-analysis have been discussed elsewhere, (4) and have been noted in other fields such as medicine. (5) In this paper, we specifically address problems with indiscriminately applying meta-analytic results to real-world settings, using law enforcement occupations as an example. This is especially timely since the Department of Justice (DOJ) has advocated the use of personality tests to reduce adverse impact. (6)
This paper has several goals. First, we examined the basis for the assertion that personality tests predict entry-level law enforcement officer performance. This discussion will focus on the difficulties associated with practical applications of meta-analytic results. In order to guide practitioners, we identify inclusion criteria that increase the applicability of meta-analysis results to an actual personnel selection setting. Our suggestions extend the work of Hurtz and Donovan, (7) who took an important initial step by addressing construct validity threats (i.e., how personality tests map onto the Big Five framework), treatment of the criterion domain, and the use of relevant samples in meta-analysis. Second, we conducted a selection-relevant meta-analysis using appropriate inclusion criteria to examine the relation between personality and performance in law enforcement occupations.
Why Entry-Level Law Enforcement Selection?
We focus on law enforcement because these occupations have many characteristics that make them particularly meaningful to examine. Starting in the mid 1970s (8) and continuing to the present time, (9) there has been a high propensity for litigation in law enforcement officer selection. Law enforcement selection batteries typically include cognitive ability tests, (10) which tend to result in adverse impact. (11) Employers have a genuine interest in finding alternatives for hiring entry-level law enforcement officers that will reduce or eliminate adverse impact while maintaining or increasing predictive validity. Since the sure techniques of race and gender norming are no longer viable options because of the Civil Rights Act (CRA) of 1991, interest has turned to examining the efficacy of including personality tests in selection batteries to reduce adverse impact. (12)
The use of personality tests to select law enforcement officers has become quite common. A noticeable increase in the use of such tests has occurred since the 1970s. (13) By far, the most commonly used personality test by municipal police departments is the Minnesota Multiphasic Personality Inventory (MMPI) (71.6 percent), followed by the Clinical Interview (57.4 percent) and the Personal History Questionnaire (52.9 percent). Furthermore, approximately 24.5 percent of municipal police departments use the California Psychological Inventory (CPI) and 18.7 percent use the 16 Personality Factor (16PF). (14) However, information regarding the use of these specific tests (e.g., given pre- or post-offer, compensatory or non-compensatory model) or the actual subscale or scales used was not available. This widespread use of personality tests for police selection, most notably the use of non-clinical personality tests, may be partially a result of meta-analytic results suggesting their usefulness. …