Using personality traits to predict police officer performance

The Authors

Beth A. Sanders, Thomas More College, Sociology and Criminal Justice, Crestview Hills, Kentucky, USA

Abstract

Purpose – The purpose of this study is to evaluate the usefulness of personality traits, namely the Big Five, as a means of selection in good police officers.

Design/methodology/approach – The current study examines a sample of 96 police officers from eight non-urban police departments.

Findings – Age and attitude were found to be better predictors of job performance measures than were personality traits. A cynical work attitude was negatively related to ratings of job performance. Officer age was found to have a non-linear relationship to job performance.

Research limitations/implications – Difficulties in measuring police performance are discussed, as is the relative importance of individual officer personality versus organizational culture.

Originality/value – The study extends the research on police officer selection and issues of job performance and measurement.

Article Type:

Research paper

Keyword(s):

Police; Personality; Performance management; Selection; Human resource management.

Journal:

Policing: An International Journal of Police Strategies & Management

Volume:

Number:

Year:

2008

pp:

129-147

Copyright ©

Emerald Group Publishing Limited

ISSN:

1363-951X

Introduction

The importance of the police selection process is well understood. Potential consequences of selection errors include both social and economic costs. The job of policing is unique in the amount of power and authority its entry-level employees are given. Thus, police agencies are expected to have hiring procedures in place that protect the community from unstable or incompetent police recruits. Policing personnel systems are fairly closed; almost all police promotions are done from within (Pynes, 1994). Therefore, failing to select the best officers has important long-term personnel ramifications for police departments. Finally, the selection process is important financially for departments because they incur substantial up-front costs when hiring new officers (Cascio, 1999; Malouff and Schutte, 1986).

Selection procedures incorporate many steps such as a written test, background check, medical exam, psychological test, and oral interview (Reaves and Hickman, 2004). This multiple-hurdle approach is not always based on choosing the best applicants, but rather on keeping out the most incompetent (Gaines and Falkenberg, 1998). This negative selection process is referred to as “weeding out”. While the desirability of “selecting in” is recognized by most police executives, many researchers would agree that current selection methods are based on eliminating the unfavorable candidates rather finding recruits with certain positive qualities. Selecting in has thus far proven to be considerably more difficult (Burbeck and Furnham, 1984). As with many occupations, it is often easier to recognize bad performance than good.

For such an important topic, relatively few studies have been conducted and the bulk of them are over a decade old. The usefulness of psychological tests for screening out extremely unsuitable and emotionally unstable applicants is generally agreed upon (Burbeck and Furnham, 1984; Burkhart, 1980; Hiatt and Hargrave, 1988a). However, the value of psychological tests for predicting successful job performance is less substantiated.

Part of the difficulty in tying psychological traits to job performance is the lack of consensus surrounding the identification of desirable traits of police officers. Certain personality characteristics related to good police work do appear more frequently than others in the police literature. Traits such as intelligence, dependability, and common sense have considerable support in the literature. In contrast, a host of other characteristics and skills such as sensitivity, empathy, and flexibility seem to garner more inconsistent results. Defining what makes a good officer can be difficult and often is just the absence of doing something wrong (James et al., 1984).

Much of the research that attempts to link personality characteristics of officers with job performance has relied on outcome measures such as academy test scores or academy attrition, rather than on the job performance. This assumes that academy performance is a good indicator of field performance. There is little evidence to suggest that is the case (Burkhart, 1980). This paper seeks to address these issues in police selection. Specifically, this study examines the question of how good policing is defined, and attempts to identify personality traits which predict successful field performance.

Goals of police selection

Seemingly, the purpose of the selection process is to hire the best candidate. In practice however, police selection goals are really two-fold: weeding out and selecting in. The selection process should both eliminate bad applicants who are not fit for policing, as well as identify those remaining candidates who would make successful officers (Alpert, 1991). Both researchers and police executives would agree that selecting in would be particularly useful; but it is also considered more difficult (Alpert, 1991; Burbeck and Furnham, 1984).

Research over the past few decades has demonstrated the effectiveness of the selection process in weeding out bad candidates. Early selection research identified several background characteristics that predicted which recruits would fail out or later be fired. These characteristics included a bad driving record, previous job firings, and prior divorce (Malouff and Schutte, 1986). Subsequent research focused on psychological testing of police applicants. Many articles examined the predictive validity of the MMPI (Minnesota Multiphasic Personality Inventory) as it had become the most commonly used psychological test in police selection. Several researchers in the 1970s and 1980s found that certain questions or subscales of the MMPI were decent predictors of unsatisfactory policing (Bartol, 1982; Bartol, 1991; Daniels and King, 2002; Hiatt and Hargrave, 1988a, 1988b; Hogan and Kurtines, 1975; Winters, 1992). Poor performance has been measured in a variety of ways (such as attrition through firing or resignation). Many studies with the goal of validating police selection steps have focused on academy performance rather than on field performance (e.g. Hogan and Kurtines, 1975; McQuilkin et al., 1990; Topp and Kardash, 1986).

Although psychological testing has been shown to be fairly effective at identifying unsuitable candidates, departments have had less success in finding ways to predict which recruits will turn into good officers. Burkhart (1980) uses the analogy of IQ scores to explain the difficulty of selecting in good candidates after poor candidates have been removed from the applicant pool. Low IQ scores are fairly effective at predicting poor school performance. However, high IQ scores do not strongly correlate with high academic achievement.

The oral interview, among the most commonly used tools in nearly all occupations to make final selections among job candidates, has been criticized for its lack of validity and reliability. Police agencies routinely use oral interviews as a screen in step near the end of the selection process in order to choose from the final applicant pool. However, this commonly used selection step has not been shown to predict job performance of police officers (Falkenberg et al., 1991). While police evaluation tools have progressed beyond simply counting the number of arrests and tickets, recognizing good policing is still difficult due to the complex nature of the job (Fyfe, 1999).

Measuring officer performance and identifying the traits of good officers

There are two elements necessary in the search for a valid method of screening in the best officers. The first is related to police performance evaluation. Before one can discuss what personality traits might be correlated with good performance, there has to be a way to measure good performance. No clear consensus exists on how to best operationalize police performance (Bartol, 1991). Second, after good officers are identified, the next step is to determine which character traits are able to discriminate the best officers from the rest. The following section will discuss the issues of officer evaluation methods and identification of personality traits of good officers.

Measuring officer performance

While a great deal has been written on police performance in the last few decades, much of that research has focused on organizational effectiveness rather than individual officer job performance (see Langworthy (1999) for a partial review). Within the police performance appraisal literature, there is disagreement about how officers should be evaluated and by whom (Falkenberg et al., 1991). Measures of performance vary widely between agencies and between research studies (Falkenberg et al., 1991). Early performance evaluation of police officers often focused on number of arrests and tickets. However, it is widely recognized that policing is more of a craft than a science, and good policing is not neatly measured by counting arrests (Bayley and Bittner, 1984; Fyfe, 1999; Muir, 1977). If this is the case, good performance may be subjective, and thus difficult to measure reliably.

A number of studies attempting to validate selection steps have used academy performance (e.g. class rank, exam scores, or graduation) or job survival (not quitting or being fired) as outcome measures (Bartol, 1991; Carpenter and Raza, 1987; Hogan and Kurtines, 1975; Lorr and Strack, 1994; McQuilkin et al., 1990; Mullins and McMains, 1995; Topp and Kardash, 1986). Unfortunately, there is no consistent evidence that shows that these indicators can be generalized to on the job performance. Burkhart (1980) argues that academic performance and field performance may not be linked at all. Job survival studies can be useful, because of their cost savings to police agencies. However, they do not identify the characteristics of a good police officer. According to Fyfe (1999), good policing is difficult to measure because the police role and function are much more elusive than other public service occupations such as firefighting. He further contends that the absence of clear and specific standards for measuring good policing can greatly harm departments.

What are the traits of a good cop?

Even after many decades of study, it is not always clear what constitutes successful policing (Bartol, 1991). Several studies have attempted to identify the characteristics of good officers (Bartol, 1982; Hogan, 1971; Hogue et al., 1994; Pugh, 1986). An early study by Hogan (1971) found that highly rated policemen were masculine and natural, as well as assertive, dependable, and uncomplicated. A similar study found that highly rated Chicago officers were those who married younger, came from stable backgrounds, and were well-socialized (Hogan, 1971). Officers categorized as good in these early studies seem to exhibit conforming behavior, rather than sharing a cluster of personality traits. A portion of police selection research has focused on identifying and measuring a set of personality traits, or core attributes, of good officers. While a certain group of characteristics such as dependability, honesty, and intelligence appear frequently in the literature, there is little agreement or consistency on the usefulness of personality testing in police selection (Ho, 1999; Sanders, 2003). For example, Hogue et al. (1994) surveyed departments regarding essential characteristics of officers. While they found that police agencies tended to value honesty, good character, truthfulness and emotional stability, their findings measured agency preference, not actual officer performance. Personality traits can be good predictors of job performance. People who are dependable, goal-oriented, persistent, and organized tend to do well at almost any job. Those who are careless, irresponsible, and low achievers do poorly at most jobs (Mount and Barrick, 1998).

The Big Five personality inventory

The empirical literature suggests that personality scales may prove useful in predicting good police performance. In particular, the Big Five personality inventory has been found to be related to job performance in a variety of occupations, including policing (Black, 2000; Cortina et al., 1992; Mount and Barrick, 1998).

The Big Five refers to five broad personality constructs that personality psychologists believe can be used to encapsulate an individual's entire personality (Table I). The five factors are:

extroversion;
neuroticism;
agreeableness;
conscientiousness; and
openness.

While the idea of statistically condensing personality is not new, only recently did a solidified and consistent model emerge (Digman, 1990). There is now a near-consensus that the Big Five can describe the most salient aspects of personality (Heller et al., 2002; John and Srivistava, 1999).

The first trait, extroversion, is sometimes referred to as assertiveness and is in contrast to introversion (Digman, 1990). Extroversion is typically seen as a positive quality and somewhat related to positive response to stimuli and general sociability and happiness (Heller et al., 2002). Extraverts had higher salaries, more promotions, and were more satisfied with their careers (Heller et al., 2002).

The second trait of neuroticism is the opposite of emotional stability. Neuroticism is associated with tension, irritability, and higher anxiety levels. Those with high neuroticism scores are more likely to be discontented and moody (John and Srivistava, 1999). According to Heller et al. (2002), those who are more neurotic pay more attention to negative events and therefore seem to experience more of them. While neuroticism is thought be useful for some very detail-oriented jobs, generally neuroticism is seen as a negative trait. Neuroticism scores have been found to predict unhappiness with career (Seibert and Kraimer, 2001).

The third trait, agreeableness, has also been described as likeability or compliance. Agreeable people are described as courteous, flexible, trusting, good-natured, cooperative, forgiving, tolerant, and good-hearted (Barrick and Mount, 1991; Digman, 1990). Seibert and Kraimer (2001) refer to the construct as a continuum ranging from soft-hearted, good-natured, and trusting, at one extreme to cynical, rude, and suspicious at the other. The literature on agreeableness and job performance is inconsistent (Seibert and Kraimer, 2001). More agreeable workers, while good team players, often are not ruthless enough to succeed in the workplace. They are less able than other employees to shirk blame for bad performance and take credit for good performance (Seibert and Kraimer, 2001). However, several policing studies suggest that less agreeable officers engage in more workplace misconduct (Cuttler and Muchinsky, 2006). Black (2000) found agreeability to be the only Big Five trait that was not related to police training performance.

Fourth, those who possess high levels of openness are described as imaginative, curious, original, broad-minded, and artistically sensitive (Barrick and Mount, 1991). According to Barrick and Mount (1991), openness has the strongest relationship to intelligence, although not all psychologists agree. Those with high scores have an inquiring intellect, while those with low scores are closed to new experiences (Digman, 1990). Barrick and Mount's (1991) meta-analysis discovered openness to be unrelated to police performance.

Finally, conscientiousness indicates the individual's degree of organization, persistence, and motivation in goal-directed behavior. This trait has been associated with educational performance and motivation (Barrick and Mount, 1991). Some personality psychologists argue that the core trait is dependability, because those with high conscientiousness scores are careful, thorough, responsible, and organized. Other psychologists posit a slightly different version, with the trait being an expression of volition or will to achieve.

In terms of predicting job performance, conscientiousness is considered the most important (Roberts et al., 2005). Occupational studies have established that conscientious people work harder and get more tasks accomplished, thus increasing their life satisfaction (Heller et al., 2002). Judge et al. (1999) determined that conscientiousness was a significant predictor of career success, as measured by salary, promotion or status. Beyond job performance, conscientiousness has been found to be related to other positive outcomes such as college student retention, positive social functioning, marital stability, and good health (Furnham et al., 2002; Roberts et al., 2005). Conversely, unreliable employees are four times more expensive than those with high conscientiousness scores (Hogan and Hogan, 1989). Barrick and Mount (1991) found that conscientiousness was related to job performance and training performance of police.

Several studies linking the Big Five to police performance have been meta-analyses (e.g. Barrick and Mount, 1991; Cortina et al., 1992; Mount and Barrick, 1998). While the Big Five, and especially conscientiousness, have been consistently related to measures of occupational performance of police, these studies have important limitations. Barrick and Mount (1991), for example, used only published studies from the 1950s to the 1980s, and only 13 percent of their sample studies were of police. The findings of meta-analyses are not sufficient to show strong predictive value because of study design and publication bias (Slavin, 1986). Additionally, few existing studies of the Big Five seem particularly applicable to the average police department. One well-cited study of the Big Five, (Black, 2000) was conducted in New Zealand and may not be relevant to American policing.

Hypotheses

Based on the literature reviewed, it is expected that each of the Big Five traits will effect individual police officer job performance. First, extroversion will be positively related to job performance. Second, neuroticism will be negatively related to job performance. Third, agreeableness will be positively related to job performance. Fourth, openness will be negatively related to job performance. Last, conscientiousness will be positively related to job performance. Further, it is expected that of the Big Five traits, conscientiousness will have the strongest effect on officer performance.

Methods

Sample

While much of the police selection research has examined large departments, most police departments are small. Overall, 90 percent of all local police agencies have fewer than 50 officers, and over half of departments employ less than ten officers (Weisheit et al., 1999). Less than 5 percent of agencies have more than 100 officers (Hickman and Reaves, 2003). Given that most police departments are relatively small, there is much to be gained from the study of police selection within these agencies.

The sample of officers was drawn from a single northern Kentucky county that borders Ohio. The county includes an urban area along with many smaller suburban cities, and a few rural areas. At the 2000 Census, the county had just over 151,000 residents, most (94 percent) are White. The median household income in the county is nearly $44,000, with 9 percent of residents below the poverty line in 1999. In total, 14 municipal police departments and one county police departments were invited to participate in the survey – eight departments agreed to include their officers in the survey; seven of those were municipal departments and one was the county police department. Those departments range from eight to 41 full-time sworn officers, and their municipality's populations range from 5,200 to 16,600 residents. The sample consists of 96 police officers. Each of the departments required all officers to participate.

The Big Five

When departments agreed to be included in the project, the chief was given a set of instructions that directed individual officers to an online version of the Big Five. The online test is a short version of the Big Five Inventory (BFI) and is available at: www.outofservice.com/bigfive. John developed a 44-question instrument to measure the Big Five traits using shorter phrases and fewer questions than previous instruments (John and Srivistava, 1999)[1]. The BFI is scored by summing the answers to a series of eight to ten questions and taking the average. For example, extroversion is measured with questions 1, 6, 11, 16, 21, 26, 31, and 36. The test provides each respondent with a score of 1-5 for each trait. Higher scores indicate more of the trait. For more information on scoring see John and Srivistava (1999).

Police performance

The dependent variable of police performance was measured in two ways. The first measure was each officer's total score from a supervisor performance evaluation. This instrument was the standard yearly evaluation form currently used by one of the departments in the sample (see Appendix). The instrument was selected after police executives in other departments expressed their confidence in it. The evaluation was made up of eight content areas. Supervisors scored each officer from one to five in each of the eight areas. For the smaller departments, the police chief (or assistant chief) completed the performance evaluations for the officers. However, in the departments with more than ten employees, evaluations were done by the officer's direct supervisor, typically a sergeant. The eight content areas were:

job knowledge;
quality of work;
cooperation;
responsibility;
initiative;
quantity of work;
dependability; and
interaction with public

Each content area was defined for the supervisors as was each score. For example, job knowledge was defined on the instrument as “how well does employee understand the requirements of job to which assigned”. Possible scores were defined for raters with a score of 1 defined as “thoroughly understands all aspects of the job”, 2 was defined as “more than adequate knowledge of the job”, 3 was defined as “has sufficient knowledge to do job”, 4 was defined as “insufficient knowledge of some phases of the job”, and 5 was defined as “continually needs instruction”. The items were reverse coded. The eight items were then combined into a total performance scale (supervisor rating). Each officer's score had a possible range of five to 40. The alpha reliability score for the scale was 0.847. The second performance measure was obtained from each department's police chief or assistant chief. Police executives were asked to rank from 1 to n, all police officers in the department (where 1 is the best/most valuable officer).

Demographic information was collected from officers. Age was measured in years. Gender was measured as a nominal variable with 1=male and 0=female. Race was measured as a nominal variable with 1=White, 2=Black, and 3=Other. Years at current department was collected. Rank was measured on an ordinal scale with 1=Patrol, 2=Sergeant, 3=Lieutenant, 4=Assistant Chief, and 5=Chief. A question-measuring attitude was included. Work attitude was measured by asking supervisors if the person's typical work attitude was:

very upbeat and positive;
somewhat positive and upbeat;
somewhat negative or cynical; or
very negative or cynical.

Finally, measures of non-performance were also gathered. Supervisors were asked to report (for the last six months) each officer's number of absences, number of informal complaints filed, number of formal complaints, and number of sustained complaints.

Findings

The sample consists of police officers from one primarily suburban county. As Table I shows, the majority of the 96 officers are White and male. Most officers included in the sample were patrol officers with an average of seven years tenure at their current department and a mean age of 36.6. Work attitude was measured from 1 (very upbeat) to 4 (very negative or cynical). Work attitudes were generally positive with supervisors rating 50 percent of officers as 2=somewhat positive.

In terms of Big Five scores, each officer's score ranged from one to five for each of the traits. Conscientiousness scores were the highest with a mean of 4.1 and a range of 2.8 to 5.0 (Table II). The sample also scored high on both agreeableness and extroversion. Generally, the sample scored high on what psychologists would view as positive traits for both social and job functioning. Similarly, officers scored quite low on neuroticism. As a group, they were quite emotionally stable. Table III shows the dependent variables measuring performance and also several measures of poor performance such as absences and complaints.

As Table III illustrates, the mean score for each of the eight performance evaluation questions is fairly high. Of the eight performance areas, Quantity of Work had the lowest mean at 3.5, and Interaction with Public the highest at 4.1. Table III also shows the range for each variable, noting that for five of the eight performance areas (Quality of Work, Cooperation, Responsibility, Dependability, and Interaction with Public), no officer in the sample received the lowest score of 1. Total performance scores could have ranged from 5 to 40, while reported performance scores ranged from 19 to 40 with a mean of 30.7. Three officers were given a perfect score of 40 by their supervisor.

In terms of non-performance measures, there were very few absences or complaints against officers in the sample. While the mean for absences was 1.7, the median and mode were zero, indicating that most officers had zero absences in the past six months. Similarly, most officers had no complaints, whether departments categorized them as formal or informal, sustained or not sustained. In short, performance ratings were quite high and measures of non-performance exceedingly rare.

Multivariate regression

It was hypothesized that individual police officer's personality characteristics as measured by the Big Five would be related to job performance. Specifically, conscientiousness was expected to have the strongest predictive power, and to be able to distinguish excellent performers from those with lower scores. Table IV presents the OLS regression estimates of the effects of the Big Five on supervisor total performance scores, as well as chief ranking. Both models provide a reasonably good job of explaining variations in performance. The first model explains almost one-quarter of the variation in police performance, while the second model has an adjusted r ² of 0.19. Contrary to expectations, conscientiousness was not predictive of good police performance, as rated by either the supervisors or police chief. Also, agreeableness and extroversion were not significantly related to either performance measure. The Big Five then had minimal effect in predicting performance and was disappointing in terms of separating the outstanding performers from low performers or mid level performers.

There were, however, some significant effects. Age, while unrelated to supervisor total performance score, had a significant association with chief rank order. Recall that chiefs were asked to rank officers from 1 to n (where n=number of officers in department) with a rank order of 1 being the best. Thus, the negative relationship means that older officers are more likely than younger ones to be ranked highly, meaning that chiefs find older officers more valuable.

The findings for work attitude are significant for both supervisor's performance score and chief's rank order. Those with higher scores have a more negative or cynical work attitude, and received lower performance score totals from supervisors and also ranked near the bottom when chiefs were asked to rank officers from 1 to n. A cynical work attitude was then predictive of poor performance ratings from both supervisors and police chiefs.

Conditional effects of personality on officer job performance

Although no direct effects were found for any of the personality characteristics of the Big Five on measures of job performance, it is possible that the effect of personality on performance is more complex. In the previous model, age was a significant predictor of chief rank, but not supervisor total performance score. It could be that age is a proxy for socialization into the police subculture, and therefore various personality characteristics might be seen as more or less appropriate depending on one's exposure to the agency's culture. For example, while extroversion might be seen as a positive characteristic for veteran officers, it might be less so for younger, less experienced officers. Five product terms were constructed (age X each of the Big Five factors) and added each sequentially to the base model presented in Table IV. The results of these analyses are presented in Table V.

Table V shows the conditional effects of age and openness, the only significant interaction effect, on both measures of job performance. Since the equation includes the components as well as the interaction effect, this produces multicollinearity. Allison (1976) demonstrates that while the betas become uninterpretable, the b for the interaction term is unaffected by the collinearity and therefore is meaningful. The parameter estimate for the joint effect of age and openness on chief's ranking is negative (b −0.318, t=−2.226), indicating that when age is high and openness is high, chief rankings are lower.

Non-linear effect of age on officer performance

Because age had conditional effects on personality, the next step in analyzing age was to determine if the relationship between age and performance was non-linear. One could expect from Niederhoffer's (1967) research on officer cynicism that the relationship between age and job performance rather than being a positive linear relationship could be curvilinear. It is likely that younger (shorter tenured) officers are seen as less valuable performers because they lack experience and job knowledge. However, their enthusiasm and energy may be perceived quite positively by supervisors and chiefs. Conversely, while older officers are valued by the organization for their job knowledge and skill, they may be seen by supervisors and administrators as being burnt out, cynical, and less active than younger officers. Previous research has found younger officers to be more active (Ricksheim and Chermak, 1994; Sherman, 1980)[2].

The second analysis regarding the effect of age is shown in Table VI. This model explores a possible non-linear relationship between age and job performance. Table VI displays the form of the relationship between age and job performance and estimates the non-linear effects of age on performance. Consistent with expectations, age exhibits an inverted U shaped relationship with supervisor total performance score. However, there was no non-linear effect of age on chief ranking (though, as reported in Table IV, age was directly related to chief ranking). This model confirms what one might expect, that young officers are viewed by supervisors as less valuable to the agency, while middle-aged officers are seen as the best performers. Officers in the older range are also seen as lower performers by supervisors. This may be because supervisors have to work harder to train and supervise younger officers. It could be that experienced officers become less active and make less use of their skills and knowledge. It is also possible that an officer's best work is at some mid-range of age, and that after a certain age, officers begin to slow down or experience burnout.

Overall, the findings were somewhat surprising. The Big Five traits did not predict officer job performance, regardless of how performance (chief rank or supervisor rating) was measured (Table VI). Openness, agreeableness, extroversion, and neuroticism did not predict good performance. Conscientiousness, based on earlier occupational studies, was thought to have the greatest possible utility for selecting in good officers. However, conscientiousness provided little in the way of explaining variation in individual police officer performance scores.

The second unexpected finding was the importance, not of individual personality, but of officer age. While age and experience have been shown in previous research to be correlates of police performance, in this analysis, age mattered in complex ways. Both age and work attitude were much better predictors of police performance than were any of the Big Five personality traits. As one might expect, the age relationship to performance is non-linear. Younger officers are ranked lower, while officers in the middle age range are the highest ranked. Additionally, officers who were older were deemed lower performers by Chiefs.

Discussion

The purpose of this study was to evaluate the usefulness of personality traits, namely the Big Five, as a means of selecting in good police officers. Unexpectedly, personality characteristics had no direct bearing on individual officer performance. Other measures of non-performance such as absenteeism and citizen complaints also were of little use for distinguishing good and poor police performance.

Rather than abandoning the idea of trying to find a way to predict which recruits will become the best police officers, three things should be considered. First, the difficulty of measuring individual police officer performance continues to be a stumbling block in the police officer selection and promotion process. Second, the question remains of how much an individual officer's personality traits matter in the undertaking of policing. Too much focus on individual personality traits of officers might overshadow the importance of the job environment, department style, and occupational culture. Last, what is to be gained for departments through better hiring and selection. Is a continued effort in honing the selecting in process a worthwhile task?

Difficulty measuring officer performance

The ineffectiveness of the Big Five at predicting job performance may be due in part to the lack of variation in supervisor performance ratings. Most officers were judged by their supervisors to be above average – the equivalent of a “B student”. This illustrates a basic limitation of supervisor ratings; they may be tapping other concepts besides officer behavior. Inflated scores could be due to halo effects, or supervisors' concerns that low evaluations reflect poorly on them (Daniels and King, 2002). It could also reflect a department policy where officer evaluations are tied to promotions and raises.

In this study, chief rankings provided a more viable method of separating the best performers from the lowest. Using multiple methods of performance evaluation has been recommended by some police scholars. Mastrofski (1981) has urged police departments to collect citizen evaluations, while Falkenberg et al. (1991) has encouraged the use of peer ratings.

Predicting performance should not only concern researchers and those involved in the hiring process, but any police administrator dealing with personnel functions. How do departments reward, retain, promote, or evaluate when the standards for individual officers remain unclear? Another key measurement issue is that indicators of good performance at one department (i.e. large number of arrests or tickets) might actually be a sign of poor performance at a smaller or non-urban department.

Though the Big Five provided little assistance in predicting officer performance with this sample, one must note that this sample of police officers not only had little variation in performance scores, but also had little variation in conscientiousness scores. Police officers in this sample had mean scores at the 79th percentile of conscientiousness as compared to other online BFI test takers. Roberts et al. (2005) found that the traits which make up the conscientiousness factor (industriousness, self-control, responsibility, traditionalism) have been shown to predict work dedication as well as social and health-related behavior. Thus, it is possible that conscientiousness does have predictive power for job performance, but that these agencies have already rejected undependable applicants. Perhaps those unsuitable for policing had already been weeded out through the selection process, or had been fired or quit. Again, psychological testing of officers has shown some success at identifying dysfunctional officers (Bartol, 1991; Hiatt and Hargrave, 1988a). However, identifying and selecting good officers still seems to be the more difficult task.

Individual officer personality or organizational culture

A second key question is the importance of the individual in policing. The joint effect of age and openness on chief ranking could be interpreted as a signal that individual personality matters very little in policing. Older officers who were more open (intellectually curious) were perceived as being poor performers by police chiefs, perhaps because chiefs expect that older officers should be fully socialized into the department and into the police culture. This supports Bartol's (1991) finding regarding the importance of maturity in police retention and Hogan's (1971) findings that conventional, traditional officers were more highly rated. Rather than individual characteristics indicating good performance, perhaps those who fit in are the most valued.

The relative importance of the police organization, and the lesser relevance of individual officer characteristics, has important consequences for police recruitment and selection policy. If individual personality does not matter, then agency socialization and training are key. Police agencies often blame the bad apples, those corrupt or inept police officers, who somehow make it through the selection process, rather than the rotten barrels which created them (Sherman, 1978).

Departmental gains through better hiring

The last important question is whether the value of improving the selection process is worth the effort, given the difficulty of the task. While the selection process weeds out the worst candidates, counterproductive behavior attracts less attention, but ultimately harms organizational effectiveness. The financial costs of bad personnel decisions are high. According to Schultz (1984), organizations could save millions of dollars using validated selection processes. Firing and re-hiring new officers can costs thousands of dollars, while lawsuits can cost departments millions of dollars. Also, the price in community trust and respect that comes with police officer misbehavior or serious abuse of authority is quite costly as well. A better selection process certainly could save departments money, and perhaps most effect smaller and non-urban departments which can ill-afford to absorb the loss of employees (Tate, 1999).

Conclusion

If individual officer performance were a continuum of bad behavior (brutality and corruption) versus outstanding performance, it is clear that very poor officers are easy to recognize and typically not difficult to fire (especially if they clearly use excessive force or engage in outright corruption). It is perhaps more difficult to recognize behavior that is merely not good (laziness, avoidance, absenteeism), and hardest of all to identify, measure, and ultimately select excellence. While administrators and citizen groups complain about the difficultly of terminating abusive or complaint-prone officers, it could be that the larger personnel problem facing administrators today is that of cumulative mediocrity. Finding the best individual officers and selecting those who will be the best is an enormous challenge and has not received enough attention from researchers or practitioners. What is good policing? If the current state of good policing is that police chiefs “know it when they see it,” then many questions remain.

Figure A1Evaluation instrument

Table IThe Big Five

Table IIPolice officer characteristics

Table IIIJob performance and non-performance dependent variables

Table IVOLS regression estimates for supervisor and chief performance evaluations

Table VOLS regression estimates for the joint effects on age and openness on supervisor and chief performance evaluations

Table VIOLS regression estimates for the non-linear effects of age on supervisor and chief performance evaluations

References

Allison, P. (1976), "Testing for interaction in multiple regression", The American Journal of Sociology, Vol. 83 No.1, pp.144-53.