Quantifying black crime: the commission’s deception
Dec 30th, 2011 by Unamused
Do blacks commit a disproportionate number of crimes? If so, does this account for their overrepresentation in the prison population better than racial discrimination?
These are the questions with which we are confronted.
Task Force on Race and the Criminal Justice System
Recall that in 2010, the Urban League of Metropolitan Seattle called for “a special commission on racial disproportions in the criminal-justice system and to make recommendations on how to modernize the system and ensure equal justice before the law.” That commission came to be known as the Task Force on Race and the Criminal Justice System (“the commission”), co-chaired by Judge Steven C. González, Chair of the Washington State Access to Justice Board, and Professor Robert S. Chang, Director of the Fred T. Korematsu Center for Law and Equality.
The commission’s first meeting was attended by “representatives from the Washington State Bar Association, the Washington State Access to Justice Board, the commissions on Minority and Justice and Gender and Justice, and all three Washington law schools, as well as leaders from nearly all of the state’s specialty bar associations, and other leaders from the community and the bar.”
By the time the commission published its 2011 “Preliminary Report on Race and Washington’s Criminal Justice System” (“the report”), no fewer than 27 organizations and institutes had participated, including the Administrative Office of the Courts, the American Civil Liberties Union of Washington, the Central Washington University Department of Law and Justice, the Defender Association/Racial Disparity Project, the Gonzaga University School of Law, the Mother Attorneys Mentoring Association of Seattle, the Seattle University School of Law, the Seattle City Attorney’s Office, the University of Washington College of Arts and Sciences and School of Law, the Washington State Access to Justice Board, the Washington State Bar Association, the Washington State Criminal Justice Training Commission, the Washington State Gender and Justice Commission, and the Washington State Minority and Justice Commission.
It is therefore somewhat surprising that the illustrious commission’s report, albeit preliminary, is so full of shit.
To be precise: the commission claims to have shown, among other things, that
- “the assertion that Black disproportionality in incarceration is due solely to differential crime commission rates is inaccurate,” and
- “racial and ethnic bias distorts decision-making at various stages in the criminal justice system, thus contributing to disproportionalities in the criminal justice system.”
However, the report’s methodology appears to be seriously flawed.
The available data
Crime commission rates are the subject of section III-A of the report. The commission begins by dismissing their own data as “incomplete” or “problematic,” then uses them anyway.
The commission’s source of criminal victimization data (i.e., what the victims of violent crime have to say) is the Bureau of Justice Statistics’ extensive, representative, annual National Crime Victimization Survey (NCVS). The report claims that “household crime victimization survey data presents an incomplete picture of crime commission rates by race” because “information about victim perceptions of perpetrators’ race is only available for a few violent offenses.”
In fact, the NCVS provides information on the following violent crimes: rape (completed and attempted), sexual assault, robbery (completed and attempted, with and without injury), aggravated assault (with and without injury, with and without a weapon), and simple assault (with and without minor injury). The offender’s (perceived) race was known and available for 74.0% of single-offender victimizations (unknown or unavailable for 26.0%), and the offenders’ racial makeup was known and available for 91.3% of multiple-offender victimizations: one race in 72.2% of victimizations, mixed races in 19.1%, and unknown or unavailable races in 8.7%.
The only prominent omission from the list of violent crimes covered by the NCVS is homicide (i.e., murder and non-negligent manslaughter), which cannot be included in a victimization survey for one obvious reason. But homicide is relatively rare, accounting for only 0.73% of combined arrests for homicide, forcible rape, robbery, and aggravated and other assaults in 2005. Therefore, it is not clear in what sense “information about victim perceptions of perpetrators’ race is only available for a few violent offenses,” other than the Census Bureau’s inability to survey every living American, every year.
The commission’s source of arrest data is the FBI’s Uniform Crime Reporting (UCR) Program. The commission claims that “[u]se of arrest data also seems problematic when clearance rates… are as low as they are.” However, the relevance of low clearance rates to the share of black arrestees is not clear.
Arrests as a proxy for crime commission
On the basis of NCVS and UCR data, the report claims that
there is strong evidence that the share of arrestees who are Black is significantly greater than the share of perpetrators identified as Black by crime victims. For example, in the 2005 crime victim survey [the NCVS], victims of non-fatal violent crimes (e.g., rape, robbery, assault) identified their assailants as Black 24.7% of the time. By contrast, 40% of those arrested for non-fatal violent crimes in 2005 were Black.
This led the commission to conclude that “arrests are a poor proxy for crime commission.” However, their “24.7%” statistic is an egregious misrepresentation of black criminality, for three reasons.
In order to arrive at a figure remotely close to 24.7% using data from the 2005 NCVS, the commission must have made a calculation along the following lines. According to the survey (Tables 40 and 46), the offender was black in 21.0% of single-offender victimizations (SOVs), which account for 79.0% of victimizations (to be precise: of the 95.7% where the number of offenders was known and available). All offenders were black in 33.9% of multiple-offender victimizations (MOVs), which account for 21.0% of victimizations. The weighted mean of 21.0% and 33.9%, according to the relative numbers of single- and multiple-offender victimizations, is therefore 23.7%.
Does anyone see yet why this calculation is utterly wrong?
- Because it considers “the share of perpetrators identified as Black” as a percentage of all victimizations, instead of as a percentage of victimizations where the offender’s race was known and available. As we have seen, the offender’s race was unknown or unavailable for 26.0% of SOVs and 8.7% of MOVs. Thus, although the victims certainly “identified their assailants as Black” in 21.0% of SOVs and 33.9% of MOVs, they “identified their assailants as Black” in 27.6% of SOVs and 37.1% of MOVs where the offender’s race was known and available.
To pretend that the first set of numbers (21.0% and 33.9%) accurately reflects the share of black perpetrators would be breathtakingly dishonest. This alone clearly invalidates the 24.7% statistic.
To put it simply, whenever the victim could not identify his attacker’s race, the commission assumed that the attacker could not have been black, so that any black person arrested for such a crime constitutes “strong evidence” that “arrests are a poor proxy for crime commission.”
- Because it considers the share of multiple-offender victimizations for which the victims identified all their assailants as black as a percentage of all victimizations, instead of as a percentage of victimizations where the offenders were not only of known and available race (point 1), but of a single, known race. As we have seen, 72.2% of MOVs involved like-race offenders; 19.1%, mixed race; and 8.7%, unknown or unavailable races.
To put it simply, whenever the victim’s attackers were of mixed races, the commission assumed that none of the attackers could have been black, so that any black person arrested for such a crime constitutes further “strong evidence” that “arrests are a poor proxy for crime commission.”
- Because it assigns equal weight to the “assailants” in single- and multiple-offender victimizations, despite the fact that multiple-offender victimizations, as the name suggests, involve multiple offenders. As we will see, the average MOV involves almost 3 offenders. Furthermore, black overrepresentation among violent offenders is consistently higher for MOVs than SOVs of most types. It follows that the report’s methodology significantly underestimates overall black overrepresentation.
This alone also clearly invalidates the 24.7% statistic.
To put it simply, the commission assigned equal weight to an assault, robbery or rape by one white as it did to an assault, robbery or rape by fourteen blacks. More than one black person arrested for the latter sort of crime constitutes yet more “strong evidence” that “arrests are a poor proxy for crime commission.”
Also note that an attack by thirteen blacks and one white would be counted as an attack by one “non-black” (point 2), as would an attack by fourteen people of unknown or unavailable race (point 1). Neither would contribute to “the share of perpetrators identified as Black by crime victims.”
Later, we will demonstrate that after accounting for these three factors and without accounting for any others (e.g., the severity of crimes within a given category, such as robbery with injury vs. without and multiple- vs. single-offender victimizations), the share of black arrestees is similar to and generally smaller than the share of black offenders. This invalidates the report’s conclusions.