That Study Saying Nearly Half of Black Males Get Arrested by Age 23? It’s Wrong

There are serious flaws in the method by which the researchers reached their conclusions, and that makes all their data suspect.


Here’s what a research study feels like to a black male teenager: “To be seen as a person with a name … then poof! A statistic, a memory and, to many, a shame,” said an 11th-grader named Asa Fludd.

I repeat this quote often to remind social advocates and cultural critics alike that our reckless recital of mind-numbing statistics is a burden to millions of young black men.

This week, countless media outlets bemoaned a study’s findings suggesting that 49 percent of black males and 38 percent of white males have been arrested by the time they turn 23. If you are the parent of a black male, these stats suggest that the hue of your son’s skin could make his likelihood of being arrested akin to a coin toss. In reality, parents of white males should also be worried that nearly 40 percent of their sons’ peers will be marred by a criminal record, but at least they can take solace in the fact that their sons fare better than black males. 

Robert Brame, Shawn D. Bushway, Ray Paternoster and Michael G. Turner published the study, “Demographic Patterns of Cumulative Arrest Prevalence by Ages 18 and 23,” in Crime & Delinquency. The study is so replete with errors that it should never have been released for public discourse. If you read the article, you will find that it is more of an experiment about handling missing waves of data from a longitudinal sample than a meaningful study about race, gender and crime. 

In the method section, the co-authors devote more than a page to their strategy for handling missing data. On how they measured arrests, they write, “Based on survey questions about arrest experiences, we compiled historical information about arrest experiences through ages 18 and 23 for each person” and nothing else. They don’t list the survey questions in an appendix or discuss the reliability or validity of the items. Without this basic information, the study is completely unreplicable, which disqualifies it from being considered good science.

However, the biggest flaw is that the study’s sample of non-Hispanic black males was only 537 who fully participated and 58 who were “missing at random.” Those who were “missing at random” participated in the first round of interviews in the longitudinal study but provided insufficient information in subsequent interviews. In the study methods, someone who had not been arrested at the time of the first interview could be assigned an arrest based on the patterns of their peers. Thus, it appears that black males could be guilty by association even in a data set.

The authors used self-reported data from the U.S. Department of Labor’s National Longitudinal Survey of Youth, 1997 cohort (NLSY97). The authors never explained why they used a data set that is designed to examine employment trends to look at arrests. The arrest data in the NLSY97 are meant to be supplemental.

It would be completely understandable if the study examined the long-term impact of arrests on employment. If Brame et al. were interested in examining race and arrest, they could use any number of surveys collected by the U.S. Department of Justice. The Office of Juvenile Justice and Delinquency Prevention has many data tools, including the National Juvenile Court Data Archive, that could give a more accurate estimate of arrests.

The NLSY97 youths were 12 to 16 years old as of Dec. 31, 1996, and were interviewed between 1997 and 2008. The respondents are 29 to 34 years old now. According to the OJJDP, the overall arrest rate for black juveniles peaked in 1995. Between their peak year and 2010, the juvenile arrest rates declined by 40 percent for black juveniles. Therefore, the study really examines a reflection, and the immediate aftermath of, the peak in juvenile arrests almost two decades later.