Police Recruitment and Officer Quality

Participants will first complete a series of multiple questions that will measure their writing, mathematical, spatial, and logic reasoning ability. Next, they will complete a simple online game that requires them to make decisions about use of force under time pressure. Next, they will report their opinion about different types of advertisements for entry-level jobs as police officers. Finally, they will answer some background questions about themselves and their interests.

2026-08-01

2026-12-31

1) The statistical relationship between four measures of racially disparate use of force (skill dimension a) and cognitive ability (skill dimension b)
2) Does the focus of recruitment ads for law enforcement positions impact the average skill set of applicants?

Outcome 1: What is the relationship between measures of racially disparate use of force and cognitive ability?

Following Correll et al. (2002), we will focus on four measures of racially disparate decision-making from the Police Officer’s Dilemma game (Dimension A): (1) the linear difference (in milliseconds) in the time it takes someone to shoot an armed Black versus White man, (2) the linear difference in time it takes to clear an unarmed Black versus White man, (3) the percentage point difference in accuracy rates for Black versus White armed images, and (4) the percentage point difference in accuracy rates for Black versus White unarmed images. We will then calculate the correlation between each of these four measures and the total cognitive score (Dimension B), using all data from completed surveys.

Outcome 2: How does the focus of recruitment ads for law enforcement positions impact the average skill set of applicants?

We will then calculate the average of these five scores (four from Dimension A and 1 from Dimension B) for participants who preferred the crime-fighting job and those who preferred the inclusive job. The difference in these averages is the second major result of this project.

How does the offered salary and focus of recruitment ads for law enforcement positions impact the distribution of applicant skills?

Outcome 3: How does the offered salary and focus of recruitment ads for law enforcement positions impact the distribution of applicant skills?

We will examine how self-reported willingness-to-accept a crime-focused versus inclusion-focused job varies with skill type and level. In essence, we will estimate labor supply curves for inclusion-focused and crime-focused jobs for participants based on four measures of skill. The first measure will be cognition (Dimension B). The remaining three measures will be calculated based on their performance in the Police Officer’s Dilemma game (Dimension A). One will give equal weight to each of our four measures from Question 1, one will place twice the weight on disparate actions when faced with an unarmed image, and one will place twice the weight on disparate actions when faced with an armed image. We will use these alternative aggregations of Dimension A to remain agnostic about the relative importance of disparities in actions towards potential criminals and potential non-criminals. This result will be presented as eight graphs (two job types, four skill measures).

This final result will show how, holding an initial “bargained salary” constant (as it would be when starting officers’ salaries are subject to a collective bargaining agreement), how many potential workers are likely to fall in different quartiles of Dimensions A and B. In addition, dividing our data by skill and desired salary in this way can reveal how changes in salary may influence how the skill level of the applicant pool.

We will conduct an incentivized online experiment to assess the attractiveness of different job opportunities to police recruits with varying skill sets, which will be assessed using a combination of randomized and non-randomized tools used in previous social science research and by law enforcement agencies. We will then assess participants interest in different types of job advertisements using standard designs aimed at eliciting true preferences.

Not available

Within the embedded video game, the race and object held by the depicted person will be randomized by the Unity video game platform. The initial job offer, and offered salary, will be randomized within the Qualtrics survey platform.

Individual survey participant.

No

We anticipate recruiting approximately 1200 participants

1200 participants

All participants will be randomly presented with both Black, White, Armed and Unarmed images. 600 participants will receive an initial offer from firm A, and 600 participants will receive an initial offer from firm B.

To calculate a minimum sample size, we focus on estimates from two papers, Correll et al. (2002) and McLean et al. (2023). McLean et al. (2023) vary the messaging and imagery in police officer job ads. The crime-fighting elements consist of an image of SWAT officers and wording to highlight the thrilling challenges of being an officer. The service and community elements consist of an image of officers shaking hands with civilians and the words “community” and “service.” Table 1 summarizes the findings most relevant to Stage 2, along with the minimum sample sizes, from a power analysis based on those reports. Table 1: Summary of previous findings, with implied sample size required for 80% power, α=0.05 Correll et al. (2002) (n=40) Outcome; Black - White Difference ; Minimum Sample Size Milliseconds to React to Armed Target;1.80% ; 500 Milliseconds to React to Unarmed Target ; 1.76%;450 Armed Targets cleared per 20 trials ; -42.8%;350 Unarmed Targets shot per 20 trials ; 17.9%;1460 McLean et al (2023) (n=241) Outcome; Community/Service - SWAT/Challenge; Minimum Sample Size Interest in Application– Imagery ; 4.9%;30 Interest in Application– Language ; 13.2%;6