The House Select Committee on Capital Punishment December 13, 2006 edited and slightly modified May 27, 2009 Statistical Evidence of racial discrimination in the death penalty Elliot M. Cramer University of North Carolina Cramer@unc.edu www.unc.edu/~cramer Background I'm Elliot Cramer. I'm a retired UNC professor. I taught and did research in quantitative methods and I've worked as an applied statistician for over 40 years in a variety of areas. I've published extensively in statistical and quantitative journals. I've done a good deal of consulting and legal testimony on statistics related to discrimination and I've been qualified as an expert in statistics in a number of court cases. I've testified for the State in one death penalty case. I am neither a proponent or opponent of the death penalty. Introduction There can be no doubt that there has been a history of racial discrimination in the death penalty in North Carolina. A concern of this committee is whether or not this discrimination is ongoing. In my handout you can see that from 1910 through 1961, many more Blacks were executed than Whites, beginning with 42 Blacks and 8 Whites in 1910-20 and continuing to 78 Blacks and 26 Whites in 1941-50. Executions plummeted with 13 Blacks and 2 Whites in 1951-61 but no further executions until 1984. Prisoners were executed for rape and robbery as well as for murder. This shows and obvious pattern of racial discrimination, From 1984 through 2009 the situation changed dramatically with 13 Blacks and 28 Whites executed, a complete reversal from previous years. No longer could critics of the Death penalty claim that many more Blacks than Whites were being executed in North Carolina, Instead it is claimed that there is a new form of discrimination revealed in the pattern of executions. Looking at intra-racial executions, one finds White kill Black 1 Black kill White 6 This is readily explained by the fact that murders involving Whites killing Blacks are relatively rare; According to a study by Isaac Unah, there are about five times as many murders involving Blacks killing Whites as Whites killing Blacks. If one considers murders with aggravating circumstances that are eligible for the death penalty, there are eight times as many involving Blacks killing Whites. This does not appear to be evidence of discrimination and studies in North Carolina have not claimed racial discrimination in inter-race murders. A more interesting statistic is the comparison for intra-racial murders where they appears to be discrimination against Whites since there are approximately equal numbers of homicides involving White victims as Black victims. White kill White 27 Black kill Black 7 Death penalty opponents have turned things around by claiming that Black life is undervalued so that murderers of Blacks are less likely to get the death penalty than murderers of Whites. Since about 94% of the murderers of Blacks are Black, executing more murderers of Black victims would inevitably mean executing more Blacks. My work on the death penalty I first became involved in death penalty issues in 1985 when I was asked by the attorney general's office to be an expert witness in the case of Wake County vs Rook. I was unfamiliar with death penalty issues and I assumed that if there were discrimination in application of the death penalty, it would involve a tendency for African-Americans to be unfairly sentenced to death. I was somewhat surprised to learn that this was not the claim at all; rather the claim was that murderers of African-Americans were LESS likely to receive the death penalty than murderers of Whites. This conclusion was not based on looking at actual Court decisions but rather on complex statistical analyses that attempted to take into account the many factors that might be relevant to these decisions. I believe that this claim was first put forward by Baldus in Georgia in a study that has received an undeservedly good press. His study was severely criticized in District Court and Baldus's statistical arguments were justifiably rejected by the Court. The Court of Appeals court and Supreme Court never had to consider the validity of the Baldus study because they rejected generalized statistical evidence as being relevant. The theory expressed by Baldus was that Black life is undervalued so that less attention is paid to homicides with Black victims. In the case I was involved with, John Rook, a young White male, raped a young White woman, ran over her with his car, and left her to bleed to death in a field. It was argued that the death penalty was discriminatory because if he had raped and murdered a Black woman, he would have been less likely to receive the death penalty. I was presented with the first of three quantitative death penalty studies that have been performed in North Carolina and I had access to the actual data. It turned out that of about 1700 homicides, over half the victims and suspects were African-American although African-Americans comprised less than 20% of the population. Most involved Whites killing Whites and Blacks killing Blacks. Cross racial homicides were very much in the minority. If one were to sentence more of these murderers of African-Americans to death, a majority of them would probably be African-American. I do not believe that this is what opponents of the death penalty have in mind. There is another explanation; perhaps homicides involving African-American victims are different from those involving White victims. Indeed inspection of the data showed that for cases involving White victims, there were two or more associated aggravating circumstances over three times as often as for Black victims. This pattern of less aggravating circumstances in African-American homicides has been shown to generally be the case in North Carolina and other states. This is the most plausible explanation for the disparities that have been observed. Professor Gross, the study author, claimed that he was "aware of no plausible alternative hypothesis that might explain away the observed pattern of racial discrimination in capital sentencing in North Carolina". This shows a lack of imagination and I believe that his conclusions were grossly incorrect. A second study by Barry Nakell has not received a great deal of attention. Methodologically it is probably the best death penalty study that has ever been done. His staff started out with the files of the Chief Medical Examiner and then did extensive interviews with prosecutors and defense attorneys as well as looking at court records and police reports. As a result he had a great deal of information that could not be obtained from records alone. I spent a great deal of time consulting on the data analysis and this has influenced my thinking about the utility of statistics in this area. It is extraordinarily difficult to take into account the many factors that are involved in death penalty cases. The data in this study are almost 30 years old and are for only one year. While I cannot support the statistical analyses that were published, I do agree with one conclusion. In the Nakell study there is clear evidence that district attorneys differ in their willingness to indict for first degree murder. Because of the small numbers in judicial districts, it is impossible to adequately account for this factor. This alone introduces a level of unpredictability which makes it impossible to adequately take into account other relevant factors, even if it were possible to quantify them. (See page two of handout) The Boger-Unah study The third study which you may be familiar with was introduced on April 16, 2001 with a press release and news conference. -- Press Release: LANDMARK NORTH CAROLINA DEATH PENALTY STUDY FINDS DRAMATIC RACIAL BIAS Raleigh, N.C. -- The most in-depth and detailed study of race and the death penalty in North Carolina's history shows clearly that race infects and undermines the capital punishment system in the state. The study, released today by the Common Sense Foundation and the North Carolina Council of Churches, finds that defendants whose victims are White are 3.5 times more likely to be sentenced to death than defendants whose victims are non-White. "Sadly, this study shows that skin color still plays a major role in deciding who lives and who dies in our criminal justice system," said Jack Boger, Professor at the University of North Carolina School of Law and principal director of the study. "Despite a generation of legal and cultural efforts to eliminate discrimination, these results show that racial bias still dramatically affects the most final of judgments - who gets the death penalty." Dr. Isaac Unah of UNC-Chapel Hill was the principal investigator in the study and came to one inescapable conclusion. "No matter how the data was analyzed, the race of the victim always emerged as an important factor in who received the death penalty," Unah said. The study's findings will be presented to the North Carolina General Assembly, where bills calling for a moratorium on executions have been introduced in the House and Senate. "This study is clear. Our death penalty system values Black lives less than White lives," said Chris Fitzsimon, Executive Director of the Common Sense Foundation. "Our policymakers should halt all executions until we can be sure we have addressed these deeply troubling findings." --- Jack Boger, our new dean, is a fine lawyer and a fine person but he just doesn't appreciate the limitations of the statistical art. His associate who is the primary author of the study is not a statistician, has not done research in statistics, and has never published an article on statistics. His study design is seriously flawed and one cannot draw any valid conclusions from this study. The "3.5 times" statement has been conceded to be incorrect by Dean Boger but just yesterday the North Carolina Coalition for a Moratorium cited it again. In the raw data the murderer of a White is 1.5 times as likely to get the death penalty and I believe that this is due to differences in aggravation circumstances. The data show that only 6% of the murderers of non-Whites were White. If the theory is correct that Black life is undervalued, executing more murderers of African-Americans would result in executing more African-Americans. It has been suggested (though it is not a conclusion in this study) that Blacks who murder Whites are substantially more likely to get the death penalty than Whites who murder Blacks. In this study, only 3% of the homicides involved Whites murdering non-Whites. In general cross-race homicides tend to be particularly aggravated. The study doesn't deal with executions; 64 of 99 sentenced to death are still on death row. Most of the others have probably been removed from death row. It was stated at the news conference that the data were available to interested parties. This report is still on the internet and it now says "After our analysis is accepted for publication in a refereed journal, we will make the data available to interested parties." It is almost six years since the press conference and I am still waiting. I expect that the data show a substantial difference in the aggravating conditions for African-American victims. Over five years later no article based on these data has been accepted for publication by any journal. At the time I noted many inconsistencies in the data. I have since seen three different papers, all using the same data, using different methods of analysis, and drawing somewhat different conclusions. None have been published. For example April 20, 2001 New York Times Professor Boger said the new study had found that the discrimination in death penalty cases was by prosecutors rather than juries. In fact, he said, what seems to have occurred is not that prosecutors sought the death penalty more often for Black defendants but that they were more willing to let defendants plead guilty in exchange for a lesser sentence if the victim was Black. In a 2002 paper based on the same data Unah and Boger wrote Due to political empowerment of nonWhites, local prosecutors in the South who once made race-conscious decisions concerning whom to prosecute for the death penalty now appear race neutral. Our findings point to unelected jurors rather than to elected prosecutors as the actors most culpable for racial bias in capital sentencing. The basic statistical problem In no study that has ever been done has it been possible to accurately predict who will be sentenced to death and possibly executed and who will not. This alone places a serious limit of one to accurately assert discrimination in the death penalty. There are many considerations that might lead a jury to sentence a criminal to death and they may not readily be quantifiable. One might think that the more variables one measures the better and the unsophisticated investigator will try to obtain measurements of everything he can. What is true is that the more relevant variables the better, but it usually is not possible to know a priori which are relevant. The experienced statistician knows that the more irrelevant variables included in an analysis, the more uncertain your statistical conclusions. It has been common in death penalty analyses to use hundreds of variables Given the large number of possible relevant variables and the small numbers of homicides in each judicial district, I do not see the possibility of any statistical study being able to adequately draw conclusions about racial disparities in administration of the death penalty. Conclusion Jack Boger has been an articulate and committed opponent of the death penalty and I'm sure that he can offer a number of good arguments as to why the death penalty should be abolished. I am agnostic on the issue. I am committed to the proper use of statistical methodology and I do not believe there is any statistical support for the view that the Death Penalty discriminates against African-Americans or anyone else. I do not think that the science of statistics has anything to contribute to the debate and death penalty opponents would do better to focus on the morality of the death penalty and the remote possibility of executing an innocent person. I agree with Mark Twain who wrote There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact. ________________________________________________________________ The House Select Committee on Capital Punishment December 13, 2006 Statistical Evidence of racial discrimination in the death penalty Elliot M. Cramer University of North Carolina Cramer@unc.edu www.unc.edu/~cramer ---- Executions in North Carolina Data from http://www.doc.state.nc.us/dop/deathpenalty 1910-20 42 Blacks executed 8 Whites executed 1921-30 51 Blacks executed 7 Whites executed 1931-40 98 Blacks executed 32 Whites executed 1941-50 78 Blacks executed 26 Whites executed 1951-61 13 Blacks executed 2 Whites executed 1962-83 No executions 1977 New death penalty statute enacted 1984-2009 13 Blacks executed 28 Whites executed 2000-2009 11 Blacks executed 15 Whites executed 1 Indian executed 1 Oriental executed 1984-2009 85 Whites removed from death row. 90 Blacks removed from death row White kill White 27 White kill Black 1 Black kill Black 7 Black kill White 6 Blacks kill 2-WF 2-WM 5-BF 1-BM 1-WF+WM 1-WF+WF 1-BF+BF+BM Whites kill 9-WF 11-WM 1-BF 4-WM+WF 2-WF+WF 1-WM+WM+WM Indian kill Indian 1 Oriental kill Oriental Blacks 10/13 involve women Whites 15/28 involve women ---------------------------- On death row April 13, 2009 Male Female Total White 60 2 64 Black 87 1 88 Indian 8 1 9 Other 4 0 4 ----------------------- Total 159 4 163 ________________________________________________________________ From Nakell and Hardy The arbitrariness of the death penalty (1987) Victim deaths due to homicide June 1, 1977 - May 31, 1978 Ordered by indictment rate from 10.5% to 100% Data in handout Statistics from Unah paper 94% of non-White victims have non-White murderers only 3% of the cases involved White kill nonWhite murders Def Vic W W 1291 36% NW NW 1670 46% NW W 515 14% 5-1 ratio cross-race murders W NW 116 3% Tot 3592 death sentence Def Vic W W 33 2.6% NW NW 29 1.7% NW W 33 6.4% W NW 4 3.5% 64 of 99 were still on death row on December 13, 2006