Sharing a Science family paper of topical relevance

Starting with a study of mayor partisanship on crime

I’m going to try (we’ll see how regularly I can do it) to start sharing papers from the Science family of journals on issues that come up in the political and media discourse. Rather than shouting my opinion on microblogs — which I did during the pandemic with no benefit — or on other social media sites where people I love disagree with me, I’m just going to put things here and let people read for themselves. I might add some of my interpretation at the end if I have one. But these will be papers in the Science family of journals where if there is something wrong with them, I’m ultimately responsible for getting them corrected, so I ultimately have responsibility for them even though other talented people who are qualified in the discipline made the decision that they will be published. I, of course, won’t put anything here that needs correction.

So, it looks like we are going to have spotlight on the city of Chicago next week, and we are likely to hear mention of “Democrat cities” in conjunction with this event. Here is a paper in Science Advances we just published called “The partisanship of mayors has no detectable effect on police spending, police employment, crime, or arrests.”

The authors are JUSTIN DE BENEDICTIS-KESSNER MATTHEW HARVEY DANIEL JONES, AND CHRISTOPHER WARSHAW.

Here’s the abstract:

In this paper, we examine whether mayors’ partisan affiliations lead to differences in crime and policing. We use a large new dataset on mayoral elections and three different modern causal inference research designs (a regression discontinuity design centered around close elections and two robust difference-in-differences methods) to determine the causal effect of mayoral partisanship on crime, arrests, and racial differences in arrest patterns in medium and large US cities. We find no evidence that mayoral partisanship affects police employment or expenditures, police force or leadership demographics, overall crime rates, or numbers of arrests. At the same time, we find some suggestive evidence that mayoral partisanship may modestly affect the racial composition of arrests. Overall, the results from our multimethod analyses indicate that local partisan politics has little causal impact on crime and policing.

The paper shows that there is no detectable effect of electing a Democrat to the office of mayor on the parameters listed above. The most important figure shows that three years after electing a Democrat as mayor, there is no statistically significant change in the amount of crime:

The paper discussed many possible reasons for this. My optimistic opinion would be that most people who become mayors have a job to do - get trash picked up, carry out pubic events, represent the city — and policing. Most of the decisions they make are not ideological; they’re just trying to get things done for their citizens. There are certainly plenty of other interpretations that will hopefully be tested in time.

Read the paper and decide for yourself. Comments open, and if you see any problems with the paper, you can send a note to science_data@aaas.org, and we’ll take a look.