Just one week after giving us the series “Price of Prosperity,” the DMN offers another Big Important Story, this time about race and jury selections called “Striking Differences.” Yesterday’s article focused on prosecutors, today’s deals with the defense. The stories are long, and the issues are complicated. I only know enough about statistics to know that numbers can be looked at in many different ways. And I think I know lawyers enough to know they can argue two sides of anything. So when the DMN reporters consult experts and quote defense attorneys, I have a hard time committing wholesale to their conclusions. For instance, District Attorney Bill Hill refutes the claim that Dallas prosecutors use race to strike potential jurors. The article acknowledges his denial, but Hill wants to make sure his side of the story is known. FB obtained the following internal memo Hill sent out to assistant district attorneys that includes another look at the data and a letter from Eric Fristch, a criminal justice expert at UNT, who writes:
The bottom line is this. I believe that the reporters are going to write a story that can basically be interpreted as an accusation that your attorneys are racially biased during jury selection. They may have interviews with a judge and an attorney who used to work in your office to make their point but they do not have the data to support such a conclusion based on their analysis of prior cases.
That memo and that letter…after the jump.
DATE: 8/22/05
TO: ASSISTANT DISTRICT ATTORNEYS
FROM: BILL HILL
SUBJECT: UPCOMING NEWS ARTICLE
________________________________________________________________________________________________As some of you know, reporters for The Dallas Morning News have been working for two years on an in-depth story about jury selection practices in Dallas County. Their stated intention in doing the story is to provide readers with an update to an article published in 1986 — the same year Thomas Miller-El was convicted and sentenced to death by a Dallas County jury.
The reporters began their research by requesting volumes of jury selection records from 2002. We fought their efforts to obtain juror information cards, which contain personal information about jurors that the law prohibits from becoming public. Judges eventually ruled in the reporters’ favor and ordered us to turn over the information.
The reporters did not originally plan to look at voir dire transcripts. Instead, they intended to use data from the juror cards to deduce the racial composition of our juries compared to the panels from which we choose them. They examined juror information cards for 108 of the 652 felony jury trials in 2002. Upon completing this portion of their research, they found that both our juries and our jury panels are approximately 70 percent Caucasian, 20 percent African American, and 10 percent Hispanic.
After coming to this conclusion, the reporters decided to seek out as many voir dire transcripts as were available from their 108-case sample. Ultimately, they obtained transcripts for 59 cases. They extracted information from those transcripts and constructed a model by which they conducted a sophisticated statistical analysis of the data.
Before presenting their findings to us, they interviewed several of our felony court chiefs and several other representatives of our office, in addition to prosecution experts across the country. Last Monday, they provided us with their four key findings and asked us to provide comments for the record. These were their findings:
– Prosecutors used peremptory strikes on eligible blacks (i.e., those not excused for cause) at more than twice the rate they struck whites;
– Whether a prospective juror was black was the strongest demographic factor in prosecution strike patterns;
– Even when blacks and whites responded the same way to questions about their contact with the criminal justice system and whether rehabilitation is the primary purpose of punishment, blacks were struck at higher rates; and
– The overwhelming majority of blacks excluded by peremptory challenges were struck by prosecutors.
The reporters informed me during an interview that they also found that defense attorneys strike whites at three times the rate they strike blacks.
Last week, I met with editors at the Morning News, along with other administrators and consultants from our office, and explained to them the myriad problems we have with their findings.
First and foremost, in our view, their key findings neglect to mention what we consider their most significant conclusion — that our juries are nearly identical in ethnic composition to the panels from which we choose them. Certainly, this stands in stark contrast to the realities of the mid-1980s.
Second, they have misrepresented the reliability of their findings. When our statistical experts from Magellan started asking questions, they learned that three of the four key findings are not derived from the sophisticated statistical model they constructed; they are simple percentage comparisons. This type of statistic would not be accepted for publication in a peer-reviewed journal because it is misleading by definition in that it identifies disparities but is not sophisticated enough to pinpoint the reasons for those disparities.
During our discussions with The Dallas Morning News, our consultants gave an example of a study in which they participated regarding a police department’s racial profiling data. Percentage comparisons revealed that the number of African-Americans motorists being stopped, searched and arrested was disproportionately high compared to other groups. Upon conducting a more sophisticated analysis of several factors contributing to stops, searches, and arrests, the researchers found that the disparities were almost entirely explained by the presence of outstanding arrest warrants on African-American motorists.
We have no doubt that, if the reporters were to more thoroughly explore the reasons behind prosecution strikes, they would find significant non-racial factors that explain the disparities. However, it seems clear that they are unwilling to do this.
Our consultants also found, upon questioning the reporters, that the one key finding that was derived from their advanced statistical model was skewed in the way it was presented. That finding is:
– Whether a prospective juror was black was the strongest demographic factor in prosecution strike patterns.
Upon reading this finding, a reasonable person could easily walk away with the impression that race is the strongest factor in prosecution strike patterns. But that is not the case. The reporters actually found that two other factors were found to be stronger than race in determining prosecution strike patterns. Those factors regard prospective jurors’ answers to two questions commonly asked during jury selection. But instead of presenting the most significant factors, the reporters inserted the word “demographic” to make the finding sound more racially charged.
We are disturbed at the way this information was presented, and concerned that the reporters may have mischaracterized this same data to numerous expert sources across the country and asked them to offer comments based on incomplete and erroneous information.
Also of note is the fact that, of the 59 cases analyzed by the reporters, there were 15 Batson hearings, none of which was sustained.
In our conversations with Morning News editors, we also learned that the expert with whom the reporters consulted throughout the course of their research, Dr. David Baldus, is an anti-death penalty activist who has conducted several studies surrounding issues of race in jury selection. He has published research showing that race is the strongest factor in determining whether a defendant receives the death penalty and that blacks in Philadelphia are four times as likely as whites to be sentenced to death. His work was well-publicized when it was first published, stirring up a lot of anti-death penalty rhetoric. But when other respected anti-death penalty researchers conducted their own studies, they refuted his claims and found that African-Americans were actually underrepresented on death row compared to Caucasians.
In short, it is clear to us that The Dallas Morning News is more interested in stirring up controversy, printing scandalous headlines and promoting their own anti-death penalty agenda than they are in providing their readers with a fair, accurate, and balanced story.
I made it clear to Morning News reporters and editors that they have vastly oversimplified a complex process, not taking into account a multitude of factors that occur during jury selection.
Beyond that, I told them that I know each and every one of you, and that I am confident in your character, integrity, and professionalism.
Regardless of what the local newspaper ultimately prints regarding this, I want all of you to know that I am proud of you. I know how hard you work. I know how passionate you are for the cause of justice. And I know how much pride you take in doing the right thing. I am behind you 100 percent. I’m not going to allow The Dallas Morning News to discourage me or distract me from my vision for this office.
I am so privileged to be here, working alongside all of you. Keep up the good work!
And here’s the letter:
June 30, 2005
Bill Hill
District Attorney
Dallas County District Attorney’s Office
Frank Crowley Courts Building
113 N. Industrial Boulevard, L.B. 19
Dallas, TX 75207-4399Dear Mr. Hill:
I am writing this letter in response to the meeting we had yesterday with reporters from The Dallas Morning News. I am concerned about the way the findings and statistics are being presented by the reporters and wanted to point out my specific concerns. Based on a fax received by your office, four key findings were presented. Three of the four findings are based on percentage comparisons and include:
– Prosecutors used peremptory strikes on eligible blacks (i.e., those not excused for cause) at more than twice the rate they struck whites;
– Even when blacks and whites responded the same way to questions about their contact with the criminal justice system and whether rehabilitation is the primary purpose of punishment, blacks were struck at higher rates;
– The overwhelming majority of blacks excluded by peremptory challenges were struck by prosecutors.
The main problem with presenting percentage comparisons is that the analysis does not provide any explanatory reasons for the disparity discovered. For example, the first finding states that prosecutors used peremptory strikes on eligible blacks at more than twice the rate they struck whites. This statement does not address the question “Why?”. There are several factors that can explain this disparity but the reporters have failed to address any of these issues. Instead, it appears that they will present the percentage comparisons and then leave it up to the readers to provide their own interpretation of what the comparisons mean and why the disparities exist.
The problem is that many will jump to the conclusion that the disparity in peremptory strikes by race is due to racial bias on the part of prosecutors. This is not what the above finding means; the conclusion is false and cannot be supported by the analysis done by the reporters. Further analysis is required to determine the reasons behind the disparity. All that can be determined from the finding above is that a disparity exists, but the reasons for the disparity are unknown.
This is the same thing that occurs in police departments when they present racial profiling data to their governing bodies. Disparities in stops, arrests, and searches are found in the data and individuals leap to the conclusion that officers are making traffic stops based on race. I participated in a study that was able to determine the factors that explain the disparities in stops, arrests, and searches. For example, one police department was significantly more likely to arrest and search blacks in comparison to whites. By looking at the factors that could explain the disparity, it was found that the disparity was almost completely explained by the increased likelihood of the presence of outstanding arrest warrants on black motorists. In other words, the disparity in arrests and searches had nothing to do with racial bias but instead had everything to do with outstanding arrest warrants. Because we were able to do a detailed analysis, we were able to answer the question “Why are black motorists more likely to be arrested and searched?” The answer in this case was not racial bias but the presence of outstanding arrest warrants which led to stops, searches and arrests. Unfortunately, the percentage comparisons presented by the reporters are not able to determine the reasons for the disparity in juror strike patterns, which leaves the question “Why?” unanswered.
The fourth finding stated that:
-Whether a prospective juror was black was the strongest demographic factor in prosecution strike patterns.
This finding is based on a logistic regression rather than percentage comparisons. Although logistic regression is a sophisticated statistical technique, the information provided in the presentation and during the meeting yesterday was insufficient to determine the quality of the statistical model and thus the validity of the finding. I posed several questions to the reporters regarding the logistic regression and received incomplete and vague answers in almost all instances which leads me to question the model presented.
When I asked about how many variables were used in the model, the answer was “a lot”. When I asked about how many variables were significant in the model, I got an answer of “about 10-12”. When I asked what variables were significant, I got a nonresponse. I have requested the data, output tables, and list of variables used in the model so the veracity of the model can be determined. I don’t expect to receive any of this information but it is necessary to make a determination regarding the quality of the analysis. Several questions need to be answered before the model can be evaluated, such as:
– Were the assumptions of logistic regression met prior to running the model through the use of various diagnostic tests?
– What was the strength of the model? How good is the model at predicting which group (struck/not struck) an individual will be placed?
– What variables were used in the model?
– What variables were significant? What was the strength of their significance?
– What variables were left out of the model that should have been analyzed?
This is just a sampling of questions that need to be answered before the quality of the model can be determined, but they remain unanswered. The reality is that you can have a significant model, but the model overall can still be very weak at predicting in which group an individual will be placed.
I also take issue with the wording of the finding that whether a prospective juror was black was the strongest demographic factor in the prosecution strike patterns. The key word is demographic. The concern is that readers will read this to mean that race is the strongest factor overall when in fact race is the third strongest factor overall. The potential jurors’ responses to the questions about rehabilitation and contact with the justice system were stronger predictors than race. Even though race was significant, the information provided at the meeting cannot be used to determine the strength of its significance.
I understand I am coming at this from an academic perspective, but my conclusion is that the data analysis and presentation of findings are unacceptable. I also want to make you aware of the fact that I am the co-editor of Youth Violence and Juvenile Justice: An Interdisciplinary Journal published by Sage Publications. This is a journal with worldwide readership. As editor, I critique the methods and analyses performed by academics throughout the world who want me to publish their research articles. I say this just to point out that I do this every day and my statement of “unacceptable” comes from years of experience.
The bottom line is this. I believe that the reporters are going to write a story that can basically be interpreted as an accusation that your attorneys are racially biased during jury selection. They may have interviews with a judge and an attorney who used to work in your office to make their point but they do not have the data to support such a conclusion based on their analysis of prior cases.
Sincerely,
Eric J. Fritsch, Ph.D.
Associate Professor
University of North Texas
Department of Criminal Justice
