You are currently viewing Crime measurement and the black-white crime gap

Crime measurement and the black-white crime gap

A lot of instant experts have appeared on the black-white crime question. Many of these are somewhat unfamiliar with the quite large literature on the topic. The people who mostly study crime are criminologists and sociologists. However, as we know, these are some of the most left-skewed academics out there:

Prior post: Do Republic professors hide their voting intentions? (2018)

Ignoring the possible issue of vote hiding, we see that sociology has about 44 Democrats per Republican. I don’t know of a survey of criminologists, but it is probably similar to sociologists.

My prior post on how to measure antisocial and criminal behavior (2019) covers many of the measurement issues, and they only get worse with the race question on top. Some pointers for non-woke reading materials:

Besides the above, one thing I want to highlight here is the problem of falsely reporting in surveys. Much is made about smaller race gaps in self-report data. Luckily, there is a way to compare. Sometimes they ask about drug use and also test people to see the truth. This produces measures of discrepancy, which can reflect lying on surveys or forgetting that one took drugs last week. In either case, the effect size by race is massive:

For the combined category (any drug), we see 99% concordance for whites, and only 81% for blacks, with Hispanics at 92% intermediate position.

  • Hughes, A., Heller, D., & Marsden, M. E. FEATURE PAPER: The Validity of Self-Reported Tobacco and Marijuana Use, by Race/Ethnicity, Gender, and Age. SESSION 1: EMERGENCY PREPAREDNESS AND SURVEILLANCE, 87.

They report data for both cannabis (marijuana) and normal tobacco, but the results are very similar. The key metrics to pay attention to are: 1) self-reported rate, and 2) positive urine rate, and 3) relative bias. The latter tells which way the bias goes. If we look at the cannabis results, we see that whites self-report slightly more use compared to blacks (8.2% to 8.1% ), but it’s not real, urine testing shows the rates are 10.8% to 15.9%. The relative bias is ~50% for blacks, showing that about half of people under-reported use compared to real rate! The results are materially the same for regular, legal tobacco, but less stark.

These kind of reporting issues are also congruent with other data:

In other words, every group has some lizardmen among them, but some groups have a lot more lizardmen than others.