As noted here, a recently published study has reviewed the decisions to seek the death penalty by the U.S. Department of Justice for evidence of racial bias and other purported problems. Here are a few observations on this study.
One of the problems in research on the death penalty is the very fact that it is so controversial. One tends to suspect that a study is done to support a predetermined result rather than to actually find the truth. This latest study, funded by the National Institute of Justice, dealt with that problem by doing multiple studies in one. The same data were provided to three different teams of researchers. Three teams using three different methodologies and coming from different perspectives all found no firm evidence of racial bias.
One team was headed by Stephen Klein of RAND Corporation. Dr. Klein had previously done research for the California Attorney General at a time when California’s death penalty was under attack. His study showed no effect of the race of the defendant or the victim. See Klein & Rolph, Relationship of Offender and Victim Race to Death Penalty Sentences in California, 32 Jurimetrics J. 33, 44 (1991). Also on this team were David Freedman of UC Berkeley and Roger Bolus of the Research Solutions Group.
A second team was headed by Richard Berk of UCLA. He was one of the experts hired by the petitioner in the McCleskey litigation of the mid-1980s. See McCleskey v. Zant, 580 F. Supp. 338, 352 (ND Ga 1984). The co-author is Yan He of UCLA.
The third study was done by Matthias Schonlau, a RAND researcher. I did not find any indication that he had previously been involved in death penalty research.
The study had an expert consultant panel of mostly defense-leaning academics: David Baldus, Alfred Blumstein, Samuel Gross, Raymond Paternoster, and Franklin Zimring. However, James Q. Wilson was also on this panel. The technical advisory committee was Ilene Nagel, Daniel Nagin, and John Rolph.
Studies and Results
All three teams noted initially that just taking raw data, the death penalty is sought more often in cases where the victim (or at least one of the victims) is white. This statistic regarding the federal death penalty has been known since the Department of Justice’s well-publicized data dump in 2000. See U. S. Dept. of Justice, The Federal Death Penalty System: A Statistical Survey (1988-2000) (2000). These numbers were cited at the time as “pervasive disparity.” See, e.g., Bonner & Lacey, Pervasive Disparities Found in the Federal Death Penalty, N. Y. Times, Sept. 12, 2000, at A6. Given that “disparity” simply means differences in numbers, that is true but meaningless. Disparity in raw numbers was not and is not proof that capital sentencing or the decision to seek the death penalty is driven by racism, notwithstanding the chronic eagerness of some people to leap to that conclusion.
Klein, Freedman, & Bolus
The Klein, Freedman, and Bolus team built a model by identifying as aggravators the factors that appeared more often in the cases where the death penalty was sought and identifying as mitigators those factors found more often in cases where it was not. The factors are then summed into aggravating and mitigating scores. These legitimate factors are then used in logistic regression models with race to see if race is a significant predictor of the charging decision after controlling for legitimate differences in the cases. In some models, the federal district is also included to check for what is sometimes called “geographic disparity.” The race variables “did not even come close to ... statistical significance.” That is, the differences noted are well within the range expected in random variation. Including the district “indicate[s] that there is not much difference across districts....”
Berk & He
Berk and He use a method called “random forests.” This method is briefly explained on pages 92-93. Suffice it to say it is a very different method from the KF&B group’s. The method yields an index to predict whether the death penalty will be sought using the nonracial variables. Adding race to the mix produced very little change in the ability of the model to predict the outcome. Their conclusion is cautiously worded. “In summary, it is difficult to determine definitively whether there is any meaningful association between race or ethnicity and prosecutor recommendations and decisions to seek or not to seek the death penalty in federal capital cases. On balance, there seems to be no evidence in these data of systematic racial effects that apply to the full set of cases we studied.” They further note that if race plays “a more complicated or subtle role,” that role is unlikely to be found.
Schonlau uses a method called propensity scoring. This is a method of adjustment to compare cases with other similar cases. As with other methods, the death penalty is sought substantially more often in white victim cases in the unadjusted data, but the difference disappears into the realm of random variation when the adjustments are made.
The fact that different researchers using different methods who have previously been associated with different sides of the debate came to the same result is very significant. In assessing what these studies show and what they do not show, it is important to consider not just “statistical significance,” but what is important from a public policy standpoint. While knowledge in this area is always imperfect, how likely is it that we do not know something that would cause us to change our policies if we did?
As Berk and He note, these studies do not prove conclusively that race has no effect. Racial issues remain pervasive in our society, and no institution could realistically expect to purge all effects of race on discretionary decision making. The question to be answered, though, is the charge of the anti-death-penalty movement that the death penalty in America today is pervasively racist. That was the suspicion underlying the Supreme Court’s 1972 decision in Furman v. Georgia, which threw out the death penalty laws existing at the time.
These three studies provide substantial evidence that the answer to the important question is no. The kind of bias that would provide a strong argument for abolition of the death penalty is a bias that makes racial minority defendants much more likely to be sentenced to death than white defendants for similar crimes and criminal histories. That claim has been so thoroughly refuted that it barely warrants mention any more. These studies drive three more nails in its coffin.
The more common claim from the mid-1980s onward has been the “race-of-the-victim bias” claim. This is, in reality, a claim that the death penalty is not imposed often enough in minority-victim cases, yet it is still used to argue against the death penalty. These studies provide further evidence that this claim is false, adding to the evidence from various state studies discussed here.
The “complex” effects of race that Berk and He suggest might exist undetected would not cause us to change our policy if they did exist. Neither would weak effects existing below the level that studies of this kind could detect. Raw racism is simple and powerful. If it pervaded the system at the level that the opponents of capital punishment claim, studies such as this would very likely find it, despite their limitations.
Capital punishment serves important interests. For some crimes, it is the only adequate punishment, and anything less is a miscarriage of justice. Evidence continues to mount that the death penalty has a substantial deterrent effect and saves innocent lives when it is actually enforced. The discretionary system that the Supreme Court has required (at the behest of the opponents) will never be perfect, and it far too often lets murderers off with less than they deserve. If race enters into the picture in small, subtle, complex ways beyond our ability to detect, those effects would not warrant giving up the benefits of a genuinely enforced death penalty. We know what we need to know. The effects of race on capital punishment are not significant enough to warrant the inadequate justice and loss of innocent life that would come from abolishing it.