Bonus: Is crime really why Targets are closing?

Rebecca Crosby, Judd Legum, and Tesnim Zekeria have a great article out on Popular Info. They use available police data to evaluate the claim that Target made as to the reason for its store closings.

Most media was happy to just believe Target spokespeople when they explained why they were closing 9 Targets across the country: they were just losing too much money from theft in those stores to be viable.

Crosby, Legum, and Zekeria found a clever way to test that hypothesis. In New York, San Francisco, and Seattle, they were able to find crime data for the immediate area near the Targets that were closing, and also around nearby Targets that were staying open. If the closing Targets were chosen based on high crime, they reasoned, we should see it in the police data.

In fact, they found that crime in the immediate area around nearby stores was equivalent to or greater than the crime nearby the closing stores. This seems to pretty effectively rule out the hypothesis that the stores were closed for actual crime.

Data nuggets

The authors point out some immediate objects, and provide some pretty good answers. First, the police data doesn't give us the number of crimes perpetrated against Target -- only the number in the immediate vicinity. So technically those numbers are upper limits for crime against Target. But for this to change the result, there would have to be "crime centers" located right next to each of the other Targets -- crime centers that don't affect target somehow. Maybe there are Rolex stores next door, and Rolex criminals wouldn't be caught dead in a Target or something. 

Second, they point out that many crimes go unreported. This is undoubtedly true, but as they point out, the only way it affects the analysis is if crimes are reported differently at different Targets, or in the areas around different Targets. If not, then even if the absolute numbers aren't precise, the relative rates are still meaningful -- in which case, we still have a hard time with the claim that the closing Targets were chosen because of the crime there.

Data crimes

The authors call out a data crime we just talked about in the last post: The National Retail Federation reported that "shrink... was 1.57% of total retail sales, up from 1.44% in 2021." See what they did there? We've seen this crime before, haven't we? Already, dear reader, I know you are asking: "Sure, but what was shrink in 2020 and before? What's the context of this increase?"

While the NRF didn't include that information (because it didn't stoke fear and alarm), the Popular Info authors did:

Over the last seven years, shrink has oscillated between 1.3% and 1.6%. Experts note that “the share is largely in line with past years and is considered a normal and healthy level of shrink.”

The other data crime happens in the comments section. I know, never read the comments section, but I did. One representative commenter complained:

Reading this article felt like a summary with no conclusion. I kept waiting for the paragraphs that would provide illumination into the reasons why these stores actually closed, but that never came.

Plenty of others jumped in with comments along the lines of, "See, I knew those corporate guys were lying. The REAL cause of the closing was [fill-in-the-blank]."

But this is a strength of the article, not a weakness. They set out to evaluate the unsubstantiated claims of Target execs. They found that those claims are unfounded, period. Now we can form and test other hypotheses. The fact that this one failed doesn't tell us anything about the others.

That makes for a more boring, less clickbaity article. It doesn't drive fear, it doesn't provide neat, pat answers. But it does make a clear contribution, and puts the "crime made us close stores" claim into the fiction category. This is how we proceed to the truth, step by careful step.

Go read the article. That's how it should be done.