The Maricopa County Sheriff's Office Annual Traffic Stop Analysis

Project Dates: 

In 2013, the Maricopa County Sheriff’s Office (MCSO) came under a federal court order regarding racially biased policing practices. As part of meeting the requirements of the court order, the MCSO contracted with the Center for Violence Prevention and Community Safety (CVPCS) to receive technical and analytical assistance to both increase the data and analytical infrastructure surrounding the MCSO’s traffic stop data analysis work group and enhance the MCSO’s capacity to collect, maintain, analyze, and disseminate traffic stop data.

The goals of the annual report are to evaluate the quality of the traffic stop data the MCSO gathers from deputies making self‐initiated stops as well as to understand the presence and extent of racially biased policing within the patrol function of the MCSO.

In regards to the traffic stop data, to date, the MCSO has made significant progress in increasing the quality of their data. Previous issues, such as missing data and duplicate stops, have either been eliminated or drastically reduced. We encourage the MCSO to continue to be responsive to the needs of improving its traffic stop data collection and management.

Image result for mcso car

To examine the relationship between racially disparate policing and traffic stops within the patrol function of the MCSO, our team attempts to answer the following research questions with theavailable traffic stop data:

1. Does descriptive, internal benchmarking identify any deputies who are engaging in policing behavior (i.e., arrest, search, seizures, and citations) towards race/ethnic minority drivers that is markedly different from their similarly situated peers?

2. In the fiscal year of 2016‐2017, are there racial/ethnic differences in post‐stop outcomes within the patrol function of the MCSO?

3. Are deputies’ assigned to traffic enforcement details associated with differences in post‐stop outcomes across driver race/ethnicity, and if they are, how do those affects work?

4. If there is evidence of racial or ethnic bias in the above analyses, is it due to systemic bias within the patrol section of the MCSO or are the differential effects across race/ethnicity due to a few deputies who show a pattern of problematic behavior?

5. Are deputies who have been identified as engaging in potentially problematic behavior in the previous reporting years of 2014‐2015 and 2015‐2016 responsible for the differential race/ethnicity effects for arrest, search, citation and length of stop in 2016‐2017, provided those differences exist?

6. Have the differential race/ethnicity effects changed over time? More specifically, do the differential race/ethnicity effects found in years 2014‐2015 and 2015‐2016 continue into 2016‐2017, and if they do, are the 2016‐2017 race/ethnicity effects different in size and direction than years previous?

We employ two types of internal benchmarking to examine the above questions. First, we use simple ratio analyses, which are meant to identify if deputies are engaging in potentially biased behavior. To construct the ratios, we compare deputies’ rates of a specific post‐stop outcome by race to the rates of other deputies conducting self‐initiated stops in the same district, to determine if deputies are engaging in behavior at a rate that is two or more times higher than their peers in the same district. When this occurs, deputies receive a “flag;” flags are indicators that a deputy may be engaging in racially biased policing and heir behavior may warrant investigation. With the ratio method, we find a number of deputies whose rates at least two times higher than what their peers are doing in the same district on the outcomes of citations, arrest, search, and seizure. While the ratios are a commonly used identifier of behavior that isoutside of what is typical in a district, their calculation is sensitive to low numbers of stops. As such, when using ratios to determine if deputies policing behavior should be examined more closely, we suggest the MCSO consider the number of stops that go into constructing the ratio. Moreover, the ratio analyses donot take into consideration other aspects of a stop that may account for disparate behavior by deputiesacross drivers’ race/ethnicity.   

Next, we use statistical modeling to determine if there is systemic bias within the patrol function of the MCSO, as well as identify deputies who may be engaging in potentially biased activity net of a number of driver, deputy, and traffic stop characteristics that are associated with post‐stop outcomes. Check out the full reports below!


Research Staff: Danielle WallaceDavid H. Tyler, Kelsey Kramer, Brooks Louton, Gary Sweeten, Matthew Gricius, Katie Brown, Jessie Huff, Jake Nelson, Shi Yan, Anthony Grubesic, and Charles Katz.