by Courtney Hayes
On May 25, 2020, a Minneapolis police officer killed George Floyd by kneeling on his neck for almost 10 minutes. Floyd’s death sparked the largest mass protest movement in U.S. history and a national reckoning on race and police reform.
In the fall of 2020, the pressure to act was heating up. Not only were protestors calling for radical reform, but crime rates were spiking. Between 2019 and 2020, the U.S. murder rate rose 30%—the largest single-year increase in more than a century. And, as the election season approached, former President Donald Trump began blaming Democratic mayors.
|
| Justin de Benedictis-Kessner |
The escalating cries for change and political finger pointing caught the attention of Justin de Benedictis-Kessner, a professor of public policy at the Harvard Kennedy School and a faculty affiliate at the Institute for Quantitative Social Science (IQSS). As de Benedictis-Kessner recalls in a recent interview, “a lot of the national political rhetoric around crime and policing was that Democrats wanted to defund the police and democratic cities were getting more dangerous. And so we were really interested in trying to figure out, is this actually the fault of these Democrats or is it just an observational equivalence problem?”
Previous research suggests the ability of mayors to unilaterally effect change is often limited. And de Benedictis-Kessner wondered, “are they actually going to be able to change anything?”
In January 2022, de Benedictis-Kessner and co-authors Matthew Harvey, Daniel Jones, and Christopher Warshaw set out to examine these questions. Their study looks at whether a mayor’s partisan affiliation might lead to differences in crime rates, arrest rates, and the racial composition of arrests. In other words, does politics matter?
It was an enormous undertaking. In their study, de Benedictis-Kessner and his co-authors collected three decades of election data and employed a statistical methodology called a regression discontinuity design to determine the causal effect of electing a Democratic rather than Republican mayor on policing and crime outcomes. They didn’t want to look just at the correlation between partisanship and crime, he said. “We wanted to get around this in a causal way and say, what's the actual effect of electing different people?”
From the outset, their research team faced many hurdles. At the national level, researchers have access to large data sets. But data on elections, crime, and policing at the local level is more scarce. As de Benedictis-Kessner explains, “it's really difficult to get accurate data on what's actually happening in terms of crime or the characteristics of the police or who's leading police departments.”
This is where support from IQSS proved invaluable. de Benedictis-Kessner hired IQSS undergraduate research scholars who went to work collecting new data on the people who lead local police departments.
Over the course of a year, their research confirmed many of their initial expectations—mayoral partisanship does not affect overall crime rates, arrests, or police employment and expenditures.
However, there were some surprises. Their research reveals that partisanship can influence the racial distribution of arrests. In other words, who gets arrested? As de Benedictis-Kessner explains, “when we dug into who's actually getting arrested, here's where we start to see differences—electing a Democrat mayor leads to proportionally fewer Black arrests for drug-related and "other" uncategorized crimes.” But these differences are small—a two percentage point reduction in the Black share of drug arrests and a similar decline in the Black share of traffic stops and nuisance crimes where police have more discretion.
Mayors face many limiting constraints, but they do have the power to appoint police chiefs who oversee staffing and local police culture. Although the process is still ongoing, de Benedictis-Kessner and his team of IQSS researchers have collected data that suggests Democratic mayors may be more likely to hire nonwhite police chiefs. And as de Benedictis-Kessner explains, “there’s a lot of background research in political science suggesting that there are benefits to descriptive representation. So having police chiefs look like the people they are serving, they might be more able to represent the interests of those people.”
The impact of a diversified police force is hotly debated. And the January 2023 killing of Tyre Nichols—brutalized by Black police officers in a city led by a Democratic mayor and an African American police chief illustrates a vexing reality. Police reform is slow and the role of politics is often limited.
But de Benedictis-Kessner insists who we elect matters and even a small change in the share of Black arrests is significant. “It's just decreasing the probability by a little bit. But we think that's important because every single additional person who's arrested by the police adds the potential for use of force and the potential for terrible tragic incidents like Tyre Nichols' death,” he said.
In the fall of 2022, de Benedictis-Kessner and his co-authors began presenting their findings at universities across the country. And the feedback from fellow scholars was loud and clear. According to de Benedictis-Kessner: “It kept coming up—we need to know who the police chiefs are, who’s running this important arm of the government, what do they look like, where do they come from?”
These questions go to the heart of what motivates de Benedictis-Kessner. “The whole reason I work on local government and local elections is we should know these answers about this level of government, and we just don’t know who’s in these offices, what are their demographics, and how long have they been in office?”
The challenge of collecting those data seemed insurmountable at first. But as de Benedictis-Kessner points out, the incredible team of IQSS research assistants “made quick work of actually getting this police chiefs data. I just wouldn't have been able to collect 10,000 rows of data in the last two months without these IQSS undergrads.”
Collecting such detailed information on American police chiefs has never been done before. And as de Benedictis-Kessner explains, the significance is profound. “We couldn’t answer basic questions of, say, ‘do police chiefs look like the people they serve?’ It’s hugely important for assessing the quality of representation, especially along descriptive demographic lines.”
Their study shines a light on another issue with important implications for our democracy—the impact of distorted rhetoric and disinformation on journalists, voters, and policy makers. As de Benedictis-Kessner explains, “the rhetoric, especially in the media, but also coming directly from politicians, is really detached from the reality on crime. And I think it's really important as a social scientist to add some actual evidence to this debate.”
According to de Benedictis-Kessner, this is an area where IQSS is having a major impact. “IQSS plays a role here that's super important, which is translating social science evidence and complicated statistics into something that the average reporter and voter can understand, and really translating that in a way that can inform policy debates. And that's a really tough job,” he said.
For de Benedictis-Kessner and his co-authors, IQSS has truly expanded their idea of what’s possible in terms of research on urban policy. And the community made all the difference. “IQSS is a fantastic group of people that centralizes a lot of the quantitative social science work that's happening at Harvard,” he said. “It's so helpful to have people thinking about how they can best support the research mission of many of the professors here, but also speak to each other and create a community around that.”
