“Disparity” Debunked

For many years on this blog I have denounced claims that a “disparity” between the percentage of a given racial or ethnic group subject to some criminal justice action and the percentage of the group in the general population supports an inference that bias in the system is the reason. I call this the Fallacy of the Irrelevant Denominator. The makeup of the general population is irrelevant because the general population is mostly law-abiding people, while serious criminal justice consequences are for people who have committed serious crimes.

Yesterday, the Bureau of Justice Statistics released a study that confirms what I have been saying as applied to violent crimes.

The Relevant Denominator

The first step in any investigation of a disparity of results for various groups is to compare the demographic characteristics of people receiving a given consequence with the demographics of people who commit the crimes for which that consequence is warranted. Of course, you have to find a source of perpetrator information that is not subject to the same bias you are trying to measure.

For the demographics of perpetrators of crimes other than homicide, we have such a source in the National Crime Victimization Survey (NCVS). This is a survey conducted by phone, much like public opinion surveys, where people are called at random and asked about crimes committed against them. (Obviously, that excludes homicide, since NCVS only surveys living victims.)

This method eliminates law enforcement bias from the survey part of the comparison, because no law enforcement officers are involved. The survey includes both crimes reported to the police and those that were not. The survey respondents are asked about the race of the perpetrators. Of course, they do not always know, and the way they identify a perpetrator’s race or ethnicity may differ from the way that person is identified upon arrest. Whether this factor affects the results may vary by group.

The new study compared perpetrator ethnic data from the NCVS with arrest data from the Uniform Crime Reports (UCR). Both crimes reported to the police and crimes overall were tallied from the NCVS. For the purpose of assessing law enforcement bias, crimes reported is the more relevant number. The police obviously can’t arrest anyone for a crime they don’t know about.

Study Results for Black Persons

Beginning with the group most easily identified by victims, black persons (as perceived by the victims) make up 34.9% of perpetrators of non-fatal violent crimes and 33.0% of persons arrested for such crime (as designated at arrest). See Table 2 on page 2. Far from being arrested out of proportion to crimes committed, the disparity goes slightly the other way, and it is well within the margin of error. See Appendix Table 2 on page 11.

These numbers should not surprise anyone familiar with crime statistics, but they powerfully refute the prevailing narrative that the “disparity” between percentages of arrestees (and subsequently convicts) who are black and the percentage of black people in the general population is entirely or primarily the result of bias against black people in law enforcement. The difference in crime commission rates accounts for the entire disparity.

That is not to say that bias does not exist. Of course it does. But it is not the primary problem. The predominant cause of the arrest/population disparity for black Americans is the crime commission disparity. If we want to fix the problem, that is where we should be focusing our attention. What are the cultural influences that cause a disproportionate number to choose the path of crime? What can we do about changing them?

The Race v. Ethnicity Problem

Moving to other demographic groups, things get more difficult. The UCR and the NCVS define “race” and “ethnicity” as independent variables. The instructions to law enforcement agencies for coding the race of the offender describe these categories:

White—A person having origins in any of the original peoples of Europe, the Middle East, or North Africa

Black or African American—A person having origins in any of the black racial groups of Africa

American Indian or Alaska Native—A person having origins in any of the original peoples of North and South America (including Central America) and who maintains tribal affiliation or community attachment

Asian—A person having origins in any of the original peoples of the Far East, Southeast Asia, the Indian subcontinent including, for example, Cambodia, China, India, Japan, Korea, Malaysia, Pakistan, the Philippine Islands, Thailand, and Vietnam

Native Hawaiian or Other Pacific Islander—A person having origins in any of the original peoples of Hawaii, Guam, Samoa, or other Pacific Islands, e.g., individuals who are Carolinian, Fijian, Kosraean, Melanesian, Micronesian, Northern Mariana Islander, Palauan, Papua New Guinean, Ponapean (Pohnpelan), Polynesian, Solomon Islander, Tahitian, Tarawa Islander, Tokelauan, Tongan, Trukese (Chuukese), and Yapese.

A separate data item then asks for coding as “Hispanic or Latino”: yes, no, or unknown. “The ethnic designation of Hispanic or Latino includes persons of Mexican, Puerto Rican, Cuban, Central or South American, or other Spanish culture or origin, regardless of race.”

The NCVS also asks for the perceived race and ethnicity of the perpetrator separately, without definitions.

In the tabulations of results, persons coded as Hispanic are listed in a separate line, and those cases are deleted from the table lines for the racial group coded for each person. Table 1, footnote b, of the BJS report notes, ” ‘white’ refers to non-Hispanic whites and ‘black’ refers to non-Hispanic blacks.” Of the perpetrators designated Hispanic in the UCR, 95.3% were designated “white” on the race question. See Appendix Table 1, page 10.

How does a victim of a crime perpetrated by a stranger who is a “Hispanic white” going to know that he is Hispanic? Maybe he speaks with an accent, if he speaks at all, or maybe he doesn’t. There are varying degrees of pre-Columbian American ancestry among Hispanic Americans, which might permit the victim to answer the question correctly, or might not. The report says, “victims were unable to determine if the offender was Hispanic in 9% of single-offender incidents and 12% of multiple-offender incidents, which may have resulted in some underestimates of Hispanic offenders’ involvement in violent crime (not shown in tables).”

White people, minus those coded as white and Hispanic, were 48.3% of perpetrators of non-fatal violent crimes reported to police in the victim survey, and 45.9% of those arrested for such crimes. The difference of 2.4% is small, and it is not “statistically significant,” meaning that given the sample size of the survey we cannot say with 95% confidence that it was not sampling error. Further, it seems likely that some perpetrators were perceived as white by the victims and coded “not Hispanic” or “don’t know” in the victim survey but Hispanic in the arrest record. In any case, the smallness of the difference is sufficient to dispel the notion that there is a “white privilege” in arrests, i.e., that white people get away with crimes that others would be arrested for, in any substantial degree.

Hispanics were 12.6% in the victim survey and 17.6% of arrestees, a difference that is statistically significant. See Table 2, page 2. However, given the perception problem discussed above, this study does not enable us to untangle the perception problem from the possibility of enforcement bias. Further research is needed on this point.

Results for Asian Persons

Asians have historically had the lowest crime rates of the major ethnic and racial groups. Tables 1 and 2 in this study indicates that Asians are 5.7% of the population, 1.8% of perpetrators of violent crimes reported to police in the victim survey, and 1.3% of persons arrested for such crimes in the UCR. The difference in the latter two numbers is not statistically significant. Differences between the survey and the arrest records were also not significant for the “American Indian/Alaska Native” and “Native Hawaiian/Other Pacific Islander” groups.

The Bottom Line

Overall, this study confirms a truth that many people have known for a long time but which goes against the current narrative. To have any chance of solving problems, we must diagnose their causes correctly. A wrong diagnosis that is adopted because it fits political agendas rather than the facts causes harm by discouraging remedies that will actually work and promoting socio-quackery that will not.