Examining the measurement, disparities, and public health consequences of law enforcement traffic stops: a Q&A

Law enforcement is a common entryway to the US justice system. In a recent article in Injury Epidemiology, researchers examined whether re-prioritizing traffic stops could reduce motor vehicle crash outcomes and racial disparities. In this Q&A, Editor-in-Chief Guohua Li speaks to author Mike Dolan Fliss.

Editor in Chief, Guohua Li: Congratulations on your study of the Fayetteville Intervention!  Would you please briefly introduce yourself, your academic background and the context of your study?    

Mike Fliss: I’m a public health research scientist at the UNC Injury Prevention Research Center and frequent collaborator with the NC Division of Public Health. I recently finished my PhD dissertation in Epidemiology at UNC Chapel Hill (2019), where I focused on social, injury, and environmental epidemiology.  The study published in Injury Epidemiology is a chapter of my PhD dissertation on the measurement, disparities, and public health consequences of law enforcement traffic stops. My dissertation grew naturally out of my volunteerism with my local NAACP branch’s bias-free policing task force. There I met lawyers and community activists looking to promote accountability of law enforcement police and sheriff departments to their local communities.

Mike Dolan Fliss

GL: For most Americans traffic stops are their first interactions with law enforcement personnel and the justice system.  What are the different types of traffic stops and how serious is the problem of selective enforcement?       

MF: We group traffic stops into three categories. “Safety stops,” including moving violations like speeding or running a light, may make up less than half of all traffic stops in many agencies. “Economic stops” (e.g. broken taillights, driving without insurance or registration, etc.), effectively criminalize poverty. Lastly, “investigatory stops” may ostensibly be to police non-traffic crimes. We include seatbelt stops in this last, most subjective, “investigatory” category. While seat belts are a major public health victory, saving many lives a year, previous research has shown investigatory and seatbelt stops have similar racial profiles in North Carolina.

Selective traffic stop enforcement has serious consequences. The US Department of Justice, in their report on Ferguson, Missouri, acknowledge how targeted traffic stops programs extracted wealth from low-income communities of color, an aspect of racial capitalism. In the most extreme, selective traffic stop enforcement puts neighborhoods more at risk of escalating minor traffic stops to violence – the stories of Sandra Bland, Walter Scott, and Philando Castile, all killed by or died in law enforcement custody, involve non-violent traffic stops.

GL: Would you please explain the principles of the Public Health Critical Race Praxis (PHCRP)? How did you apply the PHCRS to understand the social dynamics in traffic stops?   

MF: Though I benefitted from the PHRCP (Ford & Airhihenbuwa, 2010), I am no expert. The Public Health Critical Race Praxis (PHCRP) is a public health framework based on the principles of critical rate theory and anti-racism. It includes four focus areas and ten principles that can guide researchers in study design and evaluation. I believe it should be required reading of any aspiring anti-racist public health researcher or practitioner.  I found PHCRP useful to critique the intervention design, the conventional public health framework around traffic stops, and my role as a researcher.

GL: What does the Fayetteville Intervention entail? 

MF: Faced with issues of motor vehicle crashes and eroded community trust, Chief Harold Medlock voluntarily requested a review of his department practices and policies by the US Department of Justice. Fayetteville police began collecting GPS data on all traffic stops, publicly selected ten high crash intersections each week for enforcement, and prioritized safety stops to prevent traffic crash fatalities and reduce racial disparities (by relatively deprioritizing other stop types). While interviews suggested these changes represented a significant culture shift at the department, we did not quantify that aspect to the intervention.

GL: What are the outcome measures used in the evaluation of the Fayetteville Intervention? 

MF: We assessed thirteen measures in four domain areas to assess the intervention’s impact. These four domains were (A) traffic stop prioritization measures to provided evidence the intervention was implemented; (B) traffic stop disparity measures assessed questions of improved equity; (C) motor vehicle crash measures assessed crashes averted and lives saved; and (D) crime measures assessed the possibility of a Ferguson Effect (i.e. that a de-prioritization of investigatory and economic stops was associated with an increase in crime).

GL: You used the synthetic control technique in the study.  What are the tradeoffs between synthetic control and traditional comparators (e.g., using Greensboro as the comparison group)? 

MF: Synthetic control has many benefits when compared to other techniques.  The counter-factual synthetic control unit (in this case, a Fayetteville that did not enact the intervention) is created by a weighted linear combination of other control units best matched on the pre-intervention period and, if known, selected time-invariant or time-varying covariates.  Those control units in the pool more similar to Fayetteville are upweighted, and those less similar are downweighed. Because of this matching on the pre-intervention outcomes, the synthetic control weights provide some adjustment for the processes, known and unknown, that create pre-intervention variation between control and intervention groups.

However, the synthetic control technique is admittedly more complicated than a conventional DiD model and should demand more explicit consideration of known and unknown confounding. It requires careful research selection of controls.  If researchers don’t select a diversity of controls and examine the weight matrices carefully the technique may regress to a simpler, single-control DiD (as happened with one outcome in this study), weighting one unit at 100% and dropping the others.

GL: What are the main findings of your study and their implications for traffic safety, social justice and public health?

ML: Fayetteville traffic stop policing changed significantly: both the number and relatively proportion of safety stops increased, up to over 80% stops from a low of 30% in 2010 of tens of thousands of stops a year. The Fayetteville Intervention reduced traffic fatalities by 28%, injurious crashes by 23%, and total crashes by 13%. It also helped reduce disparity, including reducing Black percent of traffic stops by 7% and the Black-to-White traffic stop rate ratio by 21%. In contrast to the Ferguson Effect hypothesis, the relative de-prioritization of investigatory stops was not associated with an increase in non-traffic crime outcomes, which were reduced or unchanged, including index crimes (−10%) and violent crimes (−2%).

Our study suggests that traffic stop programs and priorities are highly malleable. Law enforcement agencies who care about public health, equity, and disparities should carefully design their traffic stop programs to save lives and reduce disparities, ideally guided by the consent and active co-design of their communities.

GL: It seems that you attribute the observed effects of the Fayetteville Intervention to re-prioritization of traffic stops.  Do you think part of the observed effects is due to the substantially increased intensity of enforcement post-intervention?  

ML: This is an important point! Certainly, it is reasonable to imagine both the number of stops as well as the proportion of those stops that were safety related would contribute to the intervention effect. How we measure these constructs matters for conclusions: for instance, though the relative percent of Black non-Hispanic stops decreased, given the increased police presence, the raw number of stops of all people increased. This challenges a simple view of what a disparity reduction might be.

GL: Your coauthors are from several disciplines.  Dr. Steve Marshall and Dr. Charles Poole are well-known and highly accomplished epidemiologists. Could you tell us a little bit about Dr. Frank Baumgartner, Dr. Paul Delamater and Dr. Whitney Robinson and what you have learned from them?  

MF: I have a great deal of appreciation for the mentorship Steve and Charlie gave and continue to give.  Dr. Steve Marshall mentored me through many injury epidemiology projects, provided great feedback as chair of my committee, and now provides support as the director of the CDC-funded Injury Prevention Research Center where I work. Dr. Charlie Poole spent many hours one-on-one with me during my PhD coursework, and was one of the first faculty to provide feedback on this project when it was a community public health project, still far from a dissertation idea.

Dr. Frank Baumgartner is the Richard J. Richardson Distinguished Professor of Political Science at UNC Chapel Hill. He is an accomplished authority on not only traffic stops, but many years of work on racial disparities in the death penalty.  Frank worked with me from the beginning of the project and has given me opportunities to support other work on the same topic.

Dr. Whitney Robinson is an accomplished social epidemiologist. Every paper she puts out I find instructive, challenging, inspiring, and actionable, pushing me to better methods and ethics around measuring disparities and a more honed understanding of race and racism in my models.

Dr. Paul Delamater is an Assistant Professor in the Geography department with a long focus on the spatial component of public health. His work includes disease modeling and health care utilization.  He helped me understand and assess the spatial dynamics in traffic stops.

I want to thank Steve, Charlie, and Whitney in our department in particular for providing advising, mentorship, and support when my first advisor, Dr. Steve Wing, passed away during my PhD. Steve’s community-lead work on environmental injustice and racism in North Carolina influenced not only my dissertation, but connected me to networks of community activists and researchers who enable my continued volunteerism around environmental justice and public health in our state.

GL: Any advice for doctoral students working on their dissertations? 

MF: Some doctoral students will have more straight-forwarded projects than mine: projects with more data, perhaps less political or sensitive, with cleaner methods, designed prospectively, with less community controversy and interest.  But I cannot recommend enough real-world projects that demand challenging, critical, ethical thinking! If you are taking on a challenging project that might seem more on public health’s periphery, I do have some advice. Peer-support is essential; find those who share your values who will also ask hard questions and keep you accountable. Some projects are harder to fund than others; you may need to balance those projects that fund you and your work with passion projects you most believe in. On a technical note, take care of your dissertation code; I recommend every modern epidemiologist learn about modern repositories and basic software engineering and efficient coding practices lest your dissertation code sprawl into spaghetti code.

Lastly, take care of yourself, and help take care of each other! As a student, though the dissertation defense will one day be over, dissertation work will continue through manuscript revisions and future projects. Be careful of thinking the next rest is just over the next hill. Our work in injury epidemiology involves topics like violent death, structural racism and sexism, and mental health distress. Though we injury epidemiologists may sometimes be abstracted from these lived realities by modeling and math, I think we must engage our humanity in these questions as well, heavy as they are. Taking care of ourselves and each other enables us to better join the analytic and human parts of ourselves in these questions, which makes for better public health for everyone.

GL: Thank you for sharing your work with us!  I look forward to reading your next paper in Injury Epidemiology. 

MF: Thank you for providing an opportunity to respond to these questions on your blog! I likewise look forward to my next submission to your journal and am thankful for your publishing support.

View the latest posts on the On Health homepage