Why Is Murder Spiking in Certain Cities?

Editor’s Note: This article previously appeared in a different format as part of The Atlantic’s Notes section, retired in 2021.

As part of our related coverage of mass incarceration, Tressie McMillan Cottom, a sociology professor at Virginia Commonwealth University, wrote a piece for us last week that contends, “Every conversation about resources in the United States is also about race and racism.” Tyler Lane, an Atlantic reader in Melbourne, Australia, offers his stats expertise to address Cottom’s comments on the rise of violent crime in several cities over the past year:

A bit on my background: I hold an DPhil (PhD) from the Oxford, for which I spent a year and a half in South Africa researching the association between child labour/responsibility and familial illness, primarily HIV/AIDS. I’m not a criminologist, but I worked as an Assistant Statistician in the Ministry of Justice in London between completing my doctorate and moving to Australia. My expertise is quantitative social science and I am currently employed as a research data analyst.

My main issue with Dr Cottom’s piece is that her arguments—that racism drives everything, especially conversations about crime—strive to keep their distance from evidence of criminality in black communities. When she does engage with the data, it seems she does her best to obscure it. For instance, this quote from her on the “Ferguson Effect”:

Waves in the bathtub aren’t even that simple to explain, much less crime waves. No one with any serious training in data, statistics, and crime attributes isolated crimes to a national trend armed with only nine months of data.

Here’s the issue with that:

it isn’t just nine months of data, nor are they isolated crimes. It’s a nine-month trend in decades of data. Not only that, the data for 2015 include the summer, the period of the year that generally sees the most crime. Across the nation’s 50 most populous cities, there has been a 16% increase in murder rate (murder is generally viewed as the most reliable crime figure since it is almost always reported and done so accurately) compared to a similar period in 2014. This reverses nearly a decade of falls in crime, murder included.

On top of that, in the next paragraph, Dr Cottom tries to have it both ways with the stats, arguing, “Were the issue actually crime [i.e., not race], statistics would tell you that crime continues its longitudinal trend downward nationally.”

Several media outlets have been accused of cherry-picking cities with high rates of crime in support of the “Ferguson Effect” hypothesis, though I think it’s also equally short-sighted to look at national murder numbers. The theory is that communities in which there has been social unrest as a result of the “Ferguson Effect” would have elevated rates of crime due either to police scale back, opportunistic criminals, enraged communities, or some combination of the three. It would not be equally distributed across populations.

For instance, including numbers from Portland, Maine in such an analysis would be rather pointless, since there is a very small minority population and (to my knowledge) no social unrest/ethnic tension or issues with police brutality. Therefore, the national murder rate is a diluted version of the data you really want and can suppress the magnitude of the effect.

Where you really need to look to test your hypothesis—that there is a rise in crime due to the “Ferguson Effect”—are cities where there has been serious unrest related to Ferguson and police brutality, such as St Louis, Baltimore, Milwaukee and New York (I haven’t included Chicago because, while it is infamous for violence and murder and has had a substantial increase in murder, I can’t find much related to Ferguson or police brutality). Not to be grim, but you’d also have to make sure each city had a sufficiently high number of murders pre-Ferguson to have some confidence that the effect was not due to statistical fluctuations.

Referring again to the FiveThirtyEight article I cited above, we see year-on-year murder increases of 60% in St Louis (85 to 136 murders), 56% in Baltimore (138 to 215), 76% in Milwaukee (59-104), and 9% in New York (190-208)—partial years up to mid-August 2014 and 2015. Bear in mind these stats naturally amalgamate all communities within a city, failing to isolate areas with the greatest unrest. If you were able to isolate them, I suspect you’d see even sharper rises in murder rates.

Of course, we must test the “Ferguson Effect” hypothesis with humility and honesty. Don’t ask, “What evidence supports my position?” Ask, “What evidence would convince me,” and then look there.

My colleague Ta-Nehisi, pivoting off the work compiled by Brentin Mock, recently insisted that “there is no Ferguson effect.” My colleagues Conor and David explored the debate in more depth, especially as it applies to Baltimore. If you have something substantive to add, especially if you have a background in crime stats, please email hello@theatlantic.com.