Jeff Sessions used our research to claim that sanctuary cities have more crime. He’s wrong.

The following article by Loren Collingwood and Benjamin Gonzalez-O’Brien was posted on the Washington Post website July 14, 2017:

Attorney General Jeff Sessions waits before speaking to federal, state and local law enforcement officials about sanctuary cities and efforts to combat violent crime on July 12 in Las Vegas. (John Locher/AP)

On Wednesday, Attorney General Jeff Sessions gave a speech in Las Vegas on sanctuary cities and local law enforcement. He announced, “According to a recent study from the University of California, Riverside, cities with these policies have more violent crime on average than those that don’t.” Almost certainly, the reference is to our study, which we first published here at the Monkey Cage in The Washington Post last October and later in the academic outlet Urban Affairs Review.

The attorney general’s summation of our study, however, is not true. In fact, our study suggests a different conclusion: Municipalities that chose to designate themselves as sanctuary cities for undocumented immigrants experience crime rates no higher than they otherwise would. We state this clearly throughout our study.

How could our findings have been so misrepresented? Although we do not know for sure, it is possible that Sessions’s speechwriters at the Department of Justice based this inaccurate characterization on an article published on Fox News or an article published on World Net Daily.

In the WND article, the author mischaracterized and repackaged our findings. She simply ignored the statistical uncertainty underlying our analyses — something loosely akin to ignoring the margin of error in polling. She then claimed that sanctuary cities had higher crime rates than non-sanctuary cities, ignoring the fact that these differences were not large enough to be statistically significant. This erroneous conclusion contributed to an increasingly popular but inaccurate narrative that sanctuary cities — and their immigrant populations — are dangerous.

In this particular analysis — summarized in the graph below — we matched each sanctuary city to a non-sanctuary city that was otherwise similar in its demographics and other factors associated with crime rates. This was our attempt to identify pairs of cities that are as similar as possible with the exception of sanctuary policy.

The vertical lines — known as “confidence intervals” — in the graph capture the statistical uncertainty. These confidence intervals were deleted in the WND article. Because the confidence intervals are relatively large compared with the difference in crime rates, we cannot conclude that the crime rates in sanctuary and non-sanctuary cities are significantly different from each other. Further analysis in our paper, which accounted for even more differences between the two city types, revealed the same finding.

In a third analysis, we compared crime rates in sanctuary cities pre- and post-passage of sanctuary city policies. This analysis, summarized in the graph below, also showed that sanctuary policies had no consistent effect on crime: Some cities saw increases, some saw no change and some saw decreases after becoming sanctuary cities.

In short, we find no evidence that crime is higher in cities that become sanctuaries. We hope that the attorney general will accurately state that finding in the future.

View the post here.