The report covers recent jail and prison construction, highlighting how fiscal and monetary policy context has shaped funding approaches and the $62 billion in local jail construction that has expanded jail capacity nearly 40 percent since early 2000s.

Explore the data here: trends.vera.org

New report here: www.vera.org/publications...

Hi Jeff! I think it’s worth getting something like that out, this thread is just a first step in the process.

I first discovered the problem in April 2024, and contacted the journal on April 12, and the authors on April 16. In order to ensure transparency for replication and reanalysis, I placed all the original code as well as my edits into a Github repository. github.com/jkangbrown/w...

Scholars have long shown community structural characteristics like disadvantage are related to police-recorded stats on crime. To find “no impact at all” should have been a red flag to a careful reader.

A clue to this merge problem might have been noticed earlier. The disadvantage variable in the authors models had very small coefficients that were not significantly different from zero.

In order to replicate the article’s results, I looked at the authors’ published materials, downloaded their data and code for replication from the Open Science Framework, and made necessary customizations to run the analysis.

If you fix the data merges and then use the rest of the authors’ analysis code, one arrives at opposite results. Reduced police pedestrian and traffic stops have no impact on violent or property crime. And reduced drug and public order arrests also have no impact.

While Lincoln Park’s actual poverty rate is 38 percent with a median household income of $37,166, it is matched to Lowry Field’s 8 percent poverty rate and household income of $102,313.

As one example, the Denver neighborhood Lincoln Park has correct police stop, arrest, and traffic accident information, but it is matched to Lowry Field’s population, housing, and demographic data.

The findings in the published paper, however, are completely untrue, and can only be replicated using an improperly merged dataset. The findings are an artifact of failed data management.

In their article, the authors find reduced police stops of pedestrians and vehicles are associated with more violent crime, and reduced drug-related arrests are associated with more property crime.

A recent article in Criminology indicates that it found evidence that “de-policing” increased violent and property crime in Denver neighborhoods. However, as I will explain, the so-called evidence for this claim is nonexistent. onlinelibrary.wiley.com/doi/10.1111/...

Some academics tried to make more sophisticated arguments about Black Lives Matter protests and used the term “de-policing” to refer to the reductions in aggressive policing tactics in the wake of calls for accountability.

During 2020, some carceral advocates blamed increased violence on Black Lives Matter demonstrators. These arguments were all over the media landscape.