The Jury Sunshine Project

Jury Selection Data As A Political Issue

In this Article, the authors look at jury selection from the viewpoint of citizens and voters, standing outside the limited boundaries of constitutional challenges. They argue that the composition of juries in criminal cases deserves political debate outside the courtroom. Voters should use the jury selection habits of judges and prosecutors to assess the overall health of local criminal justice: local conditions are unhealthy when the full-time courtroom professionals build juries that exclude parts of the local community, particularly when they exclude members of traditionally marginalized groups such as racial minorities. Every sector of society should participate in the administration of criminal justice.

This political problem starts as a public records problem. Poor access to records is the single largest reason why jury selection cannot break out of the litigator’s framework to become a normal topic for political debate. As described in Part III, the authors worked with dozens of students, librarians, and court personnel to collect jury selection documents from individual case files and assembled them into a single database, which we call “The Jury Sunshine Project.” The database encompasses more than 1,300 felony trials and almost 30,000 prospective jurors.

Part IV presents some initial findings from the Jury Sunshine Project to illustrate how public data might generate political debate beyond the courtroom. Part V explores the possible explanations for the racial patterns observed in jury selection. Some accounts of this data point to benign nonracial factors as the real explanation for the patterns observed. Other interpretations of the data treat these patterns as a new type of proof of discriminatory intent: evidence that cuts across many cases might shed new light on the likely intent of prosecutors, defense attorneys, or judges in a single case. A third perspective emphasizes the community effects of exclusion from jury service. Finally, Part VI generalizes from the data about the race of jurors to ask more generally how accessible public records could transform criminal justice. Sunshine will open up serious community debates about what is possible and desirable in local criminal justice systems.

I. Introduction

Lawyers treat jury selection—no surprise here—as an issue to litigate. They file motions, objecting to mistakes by the clerk of the court when she calls a group of potential jurors to the courthouse for jury duty. After those potential jurors arrive in the courtroom, lawyers file further motions, testing the reasons that judges give for removing a prospective juror. The lawyers also watch for signs that their opponents might rely on improper reasons, such as race or gender, to remove potential jurors from the case. Again, there is a motion for that. The law of jury selection has plenty of enforcers who stand ready to litigate.

In this Article, we stand outside the litigator’s role and look at jury selection from the viewpoint of citizens and voters. As citizens, we believe that the composition of juries deserves political debate outside the courtroom. Voters should consider the jury selection habits of judges and prosecutors when deciding whether to re-elect the incumbents to those offices. More generally, jury selection offers a stress test for the overall health of local criminal justice. Conditions are unhealthy when full-time courtroom professionals build juries that exclude parts of the local community, particularly when they exclude traditionally marginalized groups such as racial minorities. Every sector of society should play a part in the administration of criminal justice.

This political problem starts as a public records problem. As we discuss in Part II of this Article, the legal doctrines related to jury selection focus too much on single cases, and limited public access to court data makes that myopia worse. Poor access to courtroom records is the single largest reason why jury selection cannot break out of the litigator’s framework to become a normal topic for political debate.1

The paperwork in the typical case file, found in the office of the clerk of the court, does record a few details about which residents the clerk called to the courthouse, which panel members the judge and the attorneys excluded from service, and which people ultimately served on the jury. But many details about jury selection go unrecorded. And even more important, it is practically impossible to see any patterns across the case files in many different cases. The clerk normally does not hold the data in aggregate form or in electronically searchable form. Thus, there is no place to go if a citizen (or a news reporter or candidate for public office) wants to learn about the actual jury selection practices of the local judges or prosecutors. There is no vantage point from which to see the whole of jury selection, rather than the selection of a single jury.2

Until now. As we describe in Part III, we worked with dozens of students, librarians, and court personnel to collect jury selection documents from individual case files. Then we assembled them into a single database, which we call “The Jury Sunshine Project.” The paper records, housed in 100 different courthouses, depict the work of lawyers and judges in more than 1,300 felony trials, as they decided whether to remove almost 30,000 prospective jurors. The assembled data offer a panorama of jury selection practices in a state court system during an entire year.

In Part IV, we present some initial findings from the Jury Sunshine Project to illustrate how public data might generate political debate beyond the courtroom. Our analysis shows that prosecutors in North Carolina—a state with demographics and legal institutions similar to those in many other states—excluded nonwhite jurors about twice as often as they excluded white jurors. Defense attorneys leaned in the opposite direction: they excluded white jurors a little more than twice as often as nonwhite jurors. Trial judges, meanwhile, removed nonwhite jurors for “cause” about 30% more often than they removed white jurors. The net effect was for nonwhite jurors (especially black males) to remain on juries less often than their white counterparts.

The data from the Jury Sunshine Project also show differences among regions and major cities in the state. Prosecutors in three major cities—Greensboro, Raleigh, and Fayetteville—accepted a higher percentage of nonwhite jurors than prosecutors in three other cities—Charlotte, Winston-Salem, and Durham. While there may be reasons why prosecutors choose different jurors than judges or defense attorneys do, why would prosecutors in some cities produce such different results from their prosecutor colleagues in other cities?

Part V explores possible explanations for the racial patterns that we observed in jury selection. Some accounts of these data point to benign nonracial factors as the real explanation for the patterns we observed. Other interpretations of the data treat these patterns as a new type of proof of discriminatory intent: evidence that cuts across many cases might shed new light on the likely intent of prosecutors, defense attorneys, or judges in a single case.

A third perspective emphasizes the effects of exclusion from jury service. This system-wide perspective does not concentrate on what a single attorney or judge was thinking at the moment of removing a juror. Instead, what matters is how the work of all the attorneys, judges, clerks, and ordinary citizens in the courthouse forms a pattern over time. When courtroom actors exclude a portion of the community from jury duty in a persistent and predictable way, that outcome—regardless of the intent of the actors—undercuts the legitimacy of local criminal justice.

Finally, in Part VI, we generalize from our data about the race of jurors to ask more generally how accessible public records could transform criminal justice. We believe that sunshine will open up serious community debates about what is possible and desirable in the local criminal justice system. By widening the frame of vision from a litigant’s arguments about a single case, the quality of justice becomes a comparative question. For instance, voters and residents who learn about jury selection patterns will naturally ask, “How do the jury selection practices of my local court compare to practices elsewhere?” Researchers and reporters can answer those questions with standardized public data, comparing prosecutors and judges with their counterparts in different districts.

Data-based comparisons such as these make it possible to hold prosecutors and judges directly accountable to the public, in a world where voters generally have too little information about how these public servants perform their work. When challengers raise the issue during the re-election campaign of the chief prosecutor or the judge, and reporters write stories about the latest jury selection report, it could shape the selection of jurors across many cases.

With the help of public records—assembled to make it easy to compare places, offices, times, and crimes—the selection of juries could become something more than an insider’s litigation game of dueling motions. The patterns, visible in those public records, could prompt a public debate about what the voters expect from their judges and prosecutors. It takes a democratic movement, not just a constitutional doctrine, to bring the full community into the jury box.

II. Case-Level Data and Doctrines

Every defendant has a legally enforceable right to an impartial and representative jury, so lawyers and judges raise constitutional and statutory claims during criminal and collateral proceedings to protect that right. The litigators’ concerns about jury selection, however, keep the focus narrow. In this Part, we briefly review some of the legal doctrines that litigators use to enforce the ideals of jury selection, noting the doctrinal emphasis on single cases.

We then show how current public records laws and the practices of jury clerks reinforce the single-case orientation of the constitutional doctrine. As a result, it is nigh impossible to view jury selection at the overall system level. The existing archival empirical studies of jury selection reflect this difficulty: they deal with specialized crimes or targeted locations, making it difficult to draw general lessons about juries and the overall health of criminal justice systems.

A. Judge Removes Jurors for Cause

Before the start of a jury trial, lawyers for the prosecution and the defense may challenge jurors for cause. The judge, responding to these objections from the attorneys, must confirm that each potential juror meets the general requirements for service, such as residency and literacy requirements.3 At that point, the judge also evaluates possible sources of juror bias against the defendant or against the government.

The “cause” for removal might be a prospective juror’s relationship with one of the parties or lawyers.4 The judge also inquires into the prior experiences of the jurors; for instance, the judge might ask if any of the jurors was ever a victim of a crime. A juror who brings prior knowledge about the events surrounding the alleged crime receives special scrutiny. There is no limit to the number of jurors a judge might exclude on these grounds.5

The statutes and judicial opinions dealing with for-cause removals share two important features. First, the standards defer to trial judges. Appellate courts apply an “abuse of discretion” standard to these questions and rarely overturn the trial judge’s decision to grant or deny a party’s request to remove a juror for cause.6 Second, the law of for-cause removal of jurors looks to one trial at a time. Any challenge to the judge’s decision begins with a review of the court transcript for evidence of the individual juror’s alleged bias. A comparison to some other juror in the same case might be relevant, but the judge’s habits across many cases—or the actions of the local judiciary more generally during questions of removal—do not matter for litigators. Indeed, there are no aggregate data sources that could show how often trial judges remove jurors for cause. Litigators see this issue case by case, and appellate courts normally conclude that the trial judge acted within her discretion, whatever she chose.

B. Attorneys Remove Jurors with Peremptory Challenges

After the parties argue to the judge about removals for cause, lawyers for the prosecution and defense use peremptory challenges to strike a designated number of jurors.7 True to the name, peremptory strikes require no explanation. Perhaps one side wants to exclude jurors with certain political attitudes because the attorneys believe those jurors may not sympathize with their client’s side of the case. There are only a few ways that lawyers can take their peremptory strikes too far: they may not use peremptory challenges to exclude jurors based on race, gender, or other “suspect” categories for equal protection purposes. To do so would violate the Constitution.8

The method for litigants to prove racial discrimination in the use of peremptory challenges has changed over the years. Under the approach laid out in Swain v. Alabama,9 a party claiming discrimination had to present evidence reaching beyond the opponent’s behavior in the case at hand. The defendant would need to show that “in criminal cases prosecutors have consistently and systematically exercised their strikes to prevent any and all Negroes on petit jury venires from serving on the petit jury itself.”10

Two decades later, the Court in Batson v. Kentucky11 expanded the options for a party trying to prove intentional racial discrimination during jury selection. A litigant now may rely solely on the facts concerning jury selection in the individual case. Under this analysis, the attorneys try to reconstruct the state of mind of a single prosecutor (or a single defense attorney) who removed a prospective juror in a single trial. The relevant factual question is a familiar one in criminal court: what was the state of mind of a single actor at one moment in the past?

The Batson Court developed an oddly detailed constitutional test: a three-step analysis (plus one prerequisite) for examining invidious racial discrimination in the use of peremptory strikes during jury selection. As a prerequisite, the litigant must identify jurors belonging to a constitutionally relevant group, such as a group based on race, ethnicity, or gender.12 At that point, the moving party takes the first step by showing facts (such as disproportionate use of peremptory challenges against jurors of one race, or the nature of the questions posed on voir dire) to create a prima facie inference that the other attorney excluded jurors based on race.13

Second, the burden shifts to the nonmoving party to give neutral explanations for its challenges. The explaining party cannot simply deny a discriminatory intent or assert good faith. The attorney must point to some reason other than the assumption that jurors of a particular race would be less sympathetic to the party’s claims at trial.14 Finally, in the third step, the moving party offers reasons to believe that the other party’s supposedly neutral reasons for the removal of jurors were actually pretextual. On the basis of these arguments, the court decides if the nonmoving party’s explanation was authentic or pretextual.

Critics immediately spotted the potential weakness of the Batson framework and argued that it is too easy for attorneys to fabricate race-neutral reasons, after the fact, to exclude minority jurors.15 Appellate courts affirm convictions even when prosecutors invoke “nonracial” reasons that correlate with race-specific behaviors or stereotypes,16 and they sometimes affirm when prosecutors rely on the race-neutral reason only for nonwhite jurors.17 Some courts also uphold the use of peremptories where the attorney had mixed motives for the removal and at least one of the motives was nonracial.18 Several studies of published opinions confirm that appellate courts rarely reverse convictions based on Batson claims.19

Judges stress the fact-specific nature of their rulings on Batson claims.20 The Court’s latest case involving race and juror selection, Foster v. Chatman,21 reinforced this aspect of the doctrine: to use a bit of an understatement, the case did not involve subtle discrimination. Documents related to the jury selection in that case showed that the prosecutors made notations about the race of several potential jurors, writing the letter “b” alongside their names, highlighting their names in green, and placing these jurors in a category labeled, “definite NO’s.” It is hard to imagine many Batson claims with evidence this strong, certainly not for cases litigated after attorneys became more sophisticated in preparing for possible Batson claims.22

Since the Court decided Batson, critics have proposed improvements to the test.23 Chief among them, scholars persistently call for the abolition of peremptory strikes.24 At the end of the day, however, the Batson test has endured, more or less in its original form. Batson marks the boundaries of constitutional enforcement and these boundaries do not seem likely to move any time soon.25

C. Venire Selection

Litigants also sometimes object to the composition of the jury venire—the local residents whom the clerk of the court summons to the courthouse on any given day for potential jury service. Constitutional doctrine plays only a limited backstop role here, as it does with peremptory challenges.

The Supreme Court does read the Equal Protection Clause to prevent states from excluding racial groups from the jury venire by statute.26 The Court has also established that defendants may challenge the process of creating the venire, a right that stems from the Sixth Amendment’s promise of an impartial jury.27 A defendant who challenges the venire must show that a distinctive group (such as a racial group) is underrepresented in the pool, meaning that its jury venire numbers are “not reasonable in relation to” the number of such persons in the community.28 After showing a gap between the general population and the composition of the venire, the defendant must identify some aspect of the jury selection process that causes a “systematic” exclusion of the group.29

Statistics matter in proving the defendant’s claim. State courts and lower federal courts use several different techniques to measure the gap between the presence of a distinctive group in the population and on the jury venire.30 In that sense, the litigation related to jury venires places more weight on the pattern of outcomes and less on the intent of particular actors in a single trial.31 Nevertheless, litigators in this arena still look to a small set of trials—a single venire, typically a single day’s worth of trials—for the relevant evidence.32 Moreover, a judicial finding for defendants who challenge the composition of the venire is rare.33 Like the legal doctrines related to judicial removals for cause and litigant removals through peremptory challenges, the litigation surrounding the jury venire leaves most jury selection choices undisturbed—including some troubling outcomes.34

D. Public Records and Past Jury Selection Studies

As we have seen, when entire segments of the community remain under-represented in jury service, constitutional doctrines provide a remedy only in the most extreme individual cases. They do so without checking the broader context of courtroom practices. Unfortunately, record-keeping about jury selection compounds the doctrinal problem of single-case myopia.

State courts maintain records (typically in a nonelectronic format) about the construction of individual juries: which prospective jurors sat in the box, which jurors the judge removed for cause, and which jurors the two attorneys removed through peremptories.35 But aggregate data is another thing entirely: clerks do not traditionally compile data on the rate at which parties or judges exclude minority jurors over long periods of time.36 Even if state courts were to compile and publish their records to show jury selection practices across many cases, the case files are not fully comparable from place to place. The lack of data not only makes it difficult for litigants to ferret out racial discrimination in particular cases, but it also makes it difficult to identify patterns of behavior that supervisors might address through better training and accountability.37

Because of the fragmented nature of public records dealing with jury selection, researchers have not created many databases on this topic, and the limited data they have managed to collect focus on specialized crimes or on trials in a handful of locations. Comparisons across many locations, time periods, or types of crimes have not been possible.

For instance, most of the efforts of scholars and litigants to collect records about jury selection at the trial court level have related to capital murder trials. Researchers have tallied jury statistics in capital cases in Pennsylvania,38 North Carolina,39 South Carolina,40 and elsewhere.41

Other studies have ventured beyond capital murder trials but remained limited to a small number of county courthouses.42 The most comprehensive of these efforts includes a study of criminal trial juries based on records from two counties in Florida.43 Several studies focused on the creation of the jury venire, prior to any removals by judges and attorneys.44 Litigators—perhaps frustrated by silence from the academy—have also assembled some statistics regarding prosecutor exclusions from juries in single counties.45 Journalists have also assembled a few localized studies.46

Finally, a few studies have analyzed jury selection in the trial court through the lens of published opinions. Some studies used these opinions as a way to understand typical practices in trial courts, despite the selection bias problems involved.47 Other studies based on published appellate opinions restricted their analyses to the role of appellate judges in this litigation.48

What is missing from the archival research on jury selection is the power to look across all criminal trials, comparing different jurisdictions and different types of trials. Without that systemic view, judges and lawyers in one county can only speculate about whether the findings of specialized studies are generalizable to their home jurisdiction.

III. The Jury Sunshine Project

Public data, collected routinely in the criminal courts, could expand the frame of reference. If jury selection records were published in comparable form across jurisdictions, available without physical travel between courthouses, it would become feasible to compare one prosecutor’s or public defender’s office to another, and to compare one jurisdiction to another. Such comparisons might be valuable to supervising prosecutors, judges with administrative duties, researchers, voters, or even litigants.

To demonstrate how this data collection might operate, we set a goal to learn about jury selection for all felony trials in a single year, for an entire state. We chose felony trials in 2011 in North Carolina.49 Our main contribution to the existing public records was to connect the dots, pulling into one location the insights about public servants and public actions that are currently dispersed among paper files, voter records, and office websites. Although each data point comes from a public record, linking them is no easy job. In our case, it became a run through an elaborate obstacle course.

A. Traveling to the Courthouses

The first obstacle on the course was to identify trial files, separating them from the much more common cases that did not produce a trial. The North Carolina Administrative Office of the Courts (“NCAOC”) reports the number of charges tried each year, but they do not specify which cases are resolved through trial and which end with guilty pleas, dismissals, or other outcomes.50 NCAOC declined our request to generate a list of file numbers for all cases that were resolved through jury trials in 2011, citing resource limitations.51 We needed, therefore, a path around this obstacle.

Putting aside a few customized situations,52 our most useful strategy relied on public data from NCAOC to specify the trial cases. NCAOC posts raw data of court dispositions in a format not easily accessible by the public. After persistent and creative efforts by the information technology staff at our law school, we were able to download this data and format it for our purposes.53 On the basis of this NCAOC data, we generated a list of cases that led to a jury trial in each county.

In all likelihood, our lists from these various sources were incomplete. Some felony jury trials probably occurred in 2011 that never came to our attention. But based on comparisons between the number of trials we located and the number of trials that NCAOC listed in their annual reports,54 we are confident that we obtained a strong majority of the trials for that year. There is no reason to believe that our collected trials differ from the remaining trials for any relevant characteristic.55

The typical file for a felony trial, stored in the county clerk’s office, contains a jury selection form. The one-page form includes space for twelve separate jury boxes. In each box, an assistant clerk records the names of the jurors seated in that box.56 Other documents in the file indicate the judge, defense attorney, and prosecutor assigned to the case; the charges filed; the jury’s verdict for each charge in the case; and the sentence that the judge imposed.

In the fall of 2012, we conducted a pilot project in one county to test the viability of our collection plans, gathering the available file information for a few dozen trials. From that point forward, we relied on law students, law librarians, and undergraduate students to travel to most of the clerks’ offices for the 100 counties in North Carolina, between early 2013 and the summer of 2015.57 Remarkably, the clerks in 10 of the 100 counties reported that no jury trials at all occurred in their counties between 2011 and 2013.58

B. Completing the Picture for Jurors, Judges, and Attorneys

The clerk in each county summons prospective jurors who reside in that county,59 so we knew the name and county of residence of each prospective juror. Based on the research of Grosso and O’Brien in the capital trial context,60 we also knew that North Carolina maintains open public records about jurors who are also registered voters, so we assigned a cohort of student researchers to pursue the biographical background for each juror.61 Some prospective jurors were not present in the voter database because they were summoned for jury duty based on their driver’s license,62 but we did obtain the background information for a strong majority of the prospective jurors based on the voter database.63

The file for each trial indicated the judge, prosecutor(s), and defense attorney(s) assigned to the case. For most of these full-time courtroom actors, research assistants were able to identify race, gender, date of admission to the state bar (a proxy for the actor’s level of experience), and the judge’s date of appointment to the bench.64

In addition to the case-specific information about each trial and its participants, we also obtained information about each county, judicial district, and prosecutorial district.65 These data points included census information about the population and racial breakdown of each county and case-processing statistics about each prosecutorial district.

After all of the data road trips and Internet searches were done, we held records for 1,306 trials.66 This phase of the Jury Sunshine Project contains information about 29,624 removed or sitting jurors, 1,327 defendants, 694 defense attorneys, 466 prosecutors, and 129 superior court judges. We connected all of those bits of information into a single relational database.67

IV. Illustrative Comparisons of Jury Selection Practices

These data open up a new universe of questions about jury selection and performance. They shed light on simple descriptive issues about the relative contributions of judges, prosecutors, and defense attorneys in building a jury. They also allow us to compare jury practices in more serious felonies to those in the trials of lesser crimes. Because the data include the jury’s verdict on each charge,68 we can compare outcomes for a defendant with a single charge to outcomes in trials with multiple defendants and charges. It is possible to track case outcomes from juries of different compositions, based on juror age, gender, or race. Any of these questions might prove interesting to taxpayers and voters who want to understand their criminal courts.

But you have to start somewhere. In this Part, we present evidence related to racial disparities in jury service. We treat this as a demonstration project, to imagine in concrete terms the sort of public debate that might spring up when jury data become available in accessible form, allowing comparisons among jurisdictions.

Our first observations relate to the flow of prospective jurors through the courtroom. Table 1 indicates the contributions of each of the three courtroom actors.

Table 1: Total Jurors Removed and Retained

As Table 1 indicates, 57% of the jurors who sat in the jury box ultimately served on that jury. Defense attorneys were the most active courtroom figures, removing 14% of the total with peremptory challenges; judges removed 11% of the jurors for cause; and prosecutors exercised their peremptory challenges against 10% of the prospective jurors called into the box. Records did not indicate the source of the removal for 8% of the jurors.69

We know something about the order of removal because state statute creates a uniform framework for some aspects of the selection process.70 At the outset, the clerk of the court randomly selects prospective jurors from the venire to seat in the jury box. The judge instructs the jury about the general nature of the upcoming trial71 and then may ask jurors about their “general fitness and competency.”72 The parties “may personally question prospective jurors individually.”73

The judge removes jurors for cause before the parties make their peremptory challenges, basing this decision in part on motions from the attorneys. The judge rules first on the prosecutor’s motions, and the clerk replaces any jurors removed. After that, the prosecutor exercises challenges to the twelve jurors in the box. Again, the clerk refills any empty seats before the judge and prosecutor repeat the process. The defense attorney takes the next shift, asking the judge to remove jurors for cause and striking any jurors from the group of twelve that the prosecutor and judge left in the box.74 The judge and prosecutor again take the first turn on any replacement jurors who arrive in the box after the defense attorney is done with the first set of challenges.75

A. Demographic Differences Among Removed Jurors

Table 2 indicates the racial breakdown of jurors who were retained and removed. We identified 60% of our jurors as white, 16% as black, and 2% as some other race (including Hispanic ethnicity).76 The race was not indicated in our data for 22% of the jurors.77

The data indicate that black jurors and other nonwhite jurors serve on juries at a slightly lower rate than white jurors. The retention rate for white jurors was 58%, while the rate for black jurors was 56% and for jurors of other races was 50%.

Table 2: Juror Disposition, by Race of Juror

When it comes to the race of the jurors, a remarkable pattern appears in Table 2. The data show that judges removed nonwhite jurors at a higher rate than they did for white jurors.78 Then prosecutors removed nonwhite jurors at about twice the rate that they did white jurors. But in the end, defense attorneys nearly rebalanced the levels of jury service among races by removing more jurors than the judges or the prosecutors did and by using their peremptory challenges more often against white jurors than they did against black and other nonwhite jurors.

To bring these racial effects into focus, we express the differences in the form of a “race removal ratio.” In Table 3, we express the ratio of removal rates for black jurors to removal rates for white jurors: a ratio of exactly 1.0 would mean that the judges or attorneys removed black jurors and white jurors in exactly the same percentages.79 A ratio above 1.0 means that the actors removed black jurors at a higher rate than they removed white jurors. Conversely, a ratio below 1.0 means that actors removed white jurors more often. We adjusted the calculations for each courtroom actor to reflect the pool of jurors available at the time of that actor’s removal decision.80

Table 3: Removal Ratios, by Race, for Courtroom Actors

Table 3 indicates that prosecutors excluded black jurors at more than twice the rate that they excluded white jurors (for a 2.1 ratio, or 20.6% to 9.7%); similarly, they used peremptory challenges against other nonwhite jurors at twice their rate of exclusion for white jurors (producing a 2.0 ratio, or 19.5% to 9.7%). Defense attorneys, by contrast, excluded black jurors less than half as often as they excluded white jurors (with a 0.4 ratio, or 9.9% to 22.2%). Interestingly, the judges excluded black jurors for cause a bit more often (a 1.3 ratio, or 13.5% to 10.5%) but they excluded other nonwhite prospective jurors at a much higher rate (with a 2.1 ratio, or 21.7% to 10.5%).

The gender of prospective jurors complicates the selection patterns. On the whole, women and men served on juries at much the same rate. Judges, prosecutors, and defense attorneys did not differ much in their choices based on gender, at least when we look at all felony trials together.81 When race and gender intersected, however, the courtroom actors each pursued a different strategy.

Table 4: Total Removals, by Race and Gender

Black male jurors were scarce from the outset. They made up only 6.4% of the total pool of summoned jurors (compared to 9.3% for black females). Once the selection process began, judges and prosecutors removed black males at a higher rate than other jurors. Table 5 summarizes the removal rates for each of the courtroom actors.82

Table 5: Rates of Removal of Available Jurors

Defense attorneys did not remove male and female jurors of the same race at meaningfully different rates. Prosecutors, however, used their challenges proportionally more often against black male jurors (striking 23.6% of those available in the pool at that point in the process) than they did against black female jurors (18.5% of those available). A similar, but less pronounced, gap appeared in judicial removals for cause: judges removed 14.9% of the black male jurors and 12.6% of the black female jurors. All told, black males started the process underrepresented in the pool and ended up comprising only 6% of the jurors who served.83

B. Geographical Differences in Juror Removal Practices

Judges, prosecutors, and defense attorneys have different objectives at a trial and value different characteristics in jurors. It does not surprise us, therefore, to find that these courtroom actors produce different demographic patterns when they choose jurors.

Comparisons within these groups, however, are another matter. What might explain two different prosecutor’s offices that behave quite differently in their selection of juries? We explored this question through a comparison of the six largest cities in the state, all with populations larger than 200,000. Table 6 lists the removal ratios for the courtroom actors in the counties where those cities are located.

Table 6: Removal Ratios in Urban Counties

The prosecutor’s offices appear to fall into two groups. Greensboro, Raleigh, and Fayetteville all produced a removal ratio of 1.7 for black jurors; Greensboro and Durham also showed relatively low removal ratios for other nonwhite jurors. On the other hand, the prosecutor’s offices in Durham, Charlotte, and Winston-Salem excluded black jurors at a higher rate than elsewhere in the state. In the most extreme case, the prosecutors in Forsyth County removed black jurors from the box three times more often than they removed white jurors: that is, among the 151 black jurors reporting for duty in felony trials, the prosecutors exercised their peremptory challenges to remove 27.5% of the jurors available to them after the judges removed some jurors for cause. Out of 541 total white jurors, the prosecutors in Forsyth County removed 9.3% of the available candidates.

One more geographical comparison deserves our attention: the differences between urban and rural counties.84 Despite the differences in jury selection among the six largest cities in the state, urban counties as a group shared some features that distinguished them from rural counties. Table 7 summarizes the results.

Table 7: Removal Ratios, Urban and Rural Counties

For the judges and the prosecutors, it appears that the racial disparities in removal rates are most pronounced in urban counties. Defense attorneys, on the other hand, produced more racially imbalanced results in rural areas; their ratio of black-to-white removal rates became even smaller in rural counties.85

V. Preview of a Political Debate

The data from the Jury Sunshine Project speak only to outcomes in the jury selection process. The numbers show what judges and attorneys did when they picked jurors, but they do not show why. The competing—and complementary—explanations for these racial disparities in the jury selection process are a fitting topic for political debate.

In this Part, we preview the sorts of arguments that prosecutors, judges, defense attorneys, and interested community members are likely to advance during this debate. Some of these explanations for racial disparity emphasize the intent of the judges and attorneys when they exclude jurors. Others put intent to the side and ask instead about the effects of systematic exclusion on defendants and the community.

A. Intent-Based Interpretations

What might explain the patterns in jury selection that we observed in Part IV? Starting with the defense attorneys, who used their removal powers at the highest rate, perhaps the simplest explanation is best: they used all the available voir dire clues (including the race of the prospective jurors) to seat jurors who were more sympathetic to human frailty, or those who were more skeptical of local police. Perhaps the use of the jurors’ race was the explicit basis for the defense attorney’s choice, or maybe the race correlated with other clues, such as expressions of general respect for authority. Put simply, defense attorneys may have used race as one factor to pick a jury to win a trial.

As a matter of trial strategy, such choices are rational. Flanagan used our jury data to calculate the performance differences among juries of different racial compositions. He found that juries composed of more black men were more likely to acquit any defendant.86 Conversely, juries with more white men were more likely to convict, particularly when the defendant was a black man.87 Thus, it is easy to see why defense attorneys might want to save more of their peremptory challenges for white male jurors.88

As for the judges, it is more difficult to reconstruct the reasons why they removed a higher percentage of black jurors from the venire. The 30% increase in the rate of removal among black jurors, when compared to white jurors, might reflect greater economic stresses among black jurors, such as transportation difficulties or pronounced hardship from missing days away from a job.89 The higher rate of judicial removals for cause for nonwhite jurors might also reveal how judges align themselves with prosecutors, and respond more favorably to their requested removals for cause.

And then there are the prosecutors. One potential explanation for the race removal ratios higher than 1.0 would be intentional strategic decisions that incorporate race.90 Perhaps line prosecutors relied on race as a clue about the general receptiveness of jurors to a law enforcement perspective. Like the defense attorneys, the prosecutors may have relied in part on race to pick a winning jury.

It is also possible that prosecutors removed jurors based on a factor correlated with race—most prominently, jurors with a felony conviction, a prior arrest, or close family members who had negative experiences in the criminal justice system.91 Prosecutors might have been fully aware of the disparate racial impact of these choices and regretted that unintentional side effect of their removal strategy.

Again, our data suggest that such choices by prosecutors are strategically rational. Flanagan found that for every peremptory challenge that the prosecutor used, the conviction rate for black male defendants increased by 2–4%.92

None of these intent-based accounts, for any of the courtroom actors, can explain jury selection choices in individual cases. Racial disparities in aggregate jury selection outcomes speak only about averages. They reveal incentives that shape the larger patterns of removal. These arguments, therefore, might not win the day in the courtroom under current constitutional doctrine. But the reasons why prosecutors and judges exclude black jurors (especially males) at a high rate could be relevant to voters and community groups outside the courtroom as they discuss local criminal justice conditions.

B. The Effects of Juror Exclusion

A political debate about the exclusion of jurors might extend beyond the possible intent of courtroom actors. The discussion, based on data-driven comparisons of different places and actors, might also include the effects of juror exclusion.

Having a diverse jury can have life-changing implications for criminal defendants. White jurors are more likely to convict and are more likely to inflict harsh punishments on black defendants accused of killing white victims.93

The exclusion of minority jurors from service also affects the jurors themselves and the community where the trial occurs. Jury service creates a forum for popular participation in criminal justice.94 When major segments of the community remain outside the courtroom, with other more “favored” people issuing the verdicts, the legitimacy of the system suffers. Statewide statistics reveal in more systematic and detailed ways how different parts of the community find it easier or harder to serve on juries.

1. Impact on Excluded Jurors

In addition to the harm to criminal defendants, courts have long recognized that individuals who are excluded because of racial discrimination also experience a cognizable harm. For example, in Carter v. Jury Commission of Greene County, the Court noted, “People excluded from juries because of their race are as much aggrieved as those indicted and tried by juries chosen under a system of racial exclusion.”95

Even when courts have declined to hold that serving on a jury is an enforceable right, they have still agreed that jury service is a “‘badge of citizenship’ worn proudly by all those who have the opportunity to do so and that it would, indeed, be desirable for all citizens to have that opportunity.”96 Many courts have noted that exclusion of qualified groups not only violates the Constitution but also undermines “our basic concepts of a democratic society and representative government.”97 When state actors participate in this exclusion, it deepens the harm. As one court noted long ago, “When Negroes are excluded from jury service because of their color, the action of the state ‘is practically a brand upon them, affixed by the law, an assertion of their inferiority.’”98

2. Impact of Juror Exclusion on the Community

The exclusion of minority jurors also has a detrimental impact on the community. It is a basic notion of democracy that a jury should reflect the community. A jury that is “made up of representatives of all segments and groups of the community” is “more likely to fit contemporary notions of neutrality” and a combined “commonsense judgment of a group of laymen.”99

The Supreme Court has long recognized the importance of the role of jury participation in our society and has explicitly examined the impact that such exclusion has on the broader community. For example, in Taylor v. Louisiana, the Supreme Court recognized the importance in selecting a fair representation of jury members because of the potential impact on a community.100 The Court explained that the fair representation requirement was essential in (1) guarding against “the exercise of arbitrary power” and invoking the “commonsense judgment of the community as a hedge against the overzealous or mistaken prosecutor,” (2) upholding “public confidence in the fairness of the criminal justice system,” and (3) sharing the administration of justice as “a phase of civic responsibility.”101

Systemic exclusion harms the community because jury service creates a forum for popular participation in criminal justice.102 When major segments of the community remain outside the courtroom, with other people issuing the verdicts, the legitimacy of the system suffers. In Georgia v. McCollum, the Court explained that improper exclusion of jurors on the basis of race not only affects the juror, but that the harm also extends beyond the rejected juror “to touch the entire community”103 because discriminatory proceedings “undermine public confidence in the fairness of our system of justice.”104

The problems related to the systemic exclusion of racial minorities on juries are particularly acute when the subject matter of the case involves racial violence. The Court has long recognized the danger that such cases might create distrust within minority communities. For example, in McCollum, Justice Blackmun discussed cases involving racial violence in which peremptory challenges had resulted in the striking of all black jurors:

In such cases, emotions in the affected community will inevitably be heated and volatile. Public confidence in the integrity of the criminal justice system is essential for preserving community peace in trials involving race-related crimes. Be it at the hands of the State or the defense, if a court allows jurors to be excluded because of group bias, it is a willing participant in a scheme that could only undermine the very foundation of our system of justice—our citizens’ confidence in it.105

A homogenous jury, on the surface, does not look like a fair jury. The appearance of prejudice in the jury selection process leads to continuing pessimism and distrust concerning the operation of the criminal justice system among the omitted groups.106 The excluded community perceives that it is “shut out.” The court’s participation in discrimination and racism undermines its moral authority as the enforcer of antidiscrimination policies.107

The public at large also shares an interest in “demonstrably fair trials that produce accurate verdicts.”108 Diversity itself enhances the deliberations of juries. In Peters v. Kiff,109 Justice Marshall identified this contribution of a representative jury:

When any large and identifiable segment of the community is excluded from jury service, the effect is to remove from the jury room qualities of human nature and varieties of human experience . . . . [E]xclusion deprives the jury of a perspective on human events that may have unsuspected importance in any case that may be presented.110

In sum, excluding minorities from jury selection has negative implications beyond the harms that a criminal defendant might raise in the courtroom. Like other systemic issues in the criminal justice system, visible and systematic barriers to jury service can erode community trust and decrease legitimacy.111

The accountability of judges and prosecutors to the community is also compromised when particular races, neighborhoods, ages, or other social groups cannot contribute their fair share to the jury system. In particular, prosecutors who can exclude parts of the community from jury service effectively shield themselves from full accountability to the public.112 They can choose for themselves which segments of the population will set their priorities in the charging and resolution of cases.

Whether such disparities are the result of purposeful discrimination is difficult to prove, but even the perception that discrimination is occurring has important implications for the criminal justice system.113 These practices deserve scrutiny outside the courtroom, beyond the confines of constitutional doctrine.

VI. Access to Data and Criminal Justice Reform

In Part IV we highlighted data, for illustrative purposes, to address the question of exclusion from juries on the basis of race. But racial equity is only one possible objective for those who might use open jury data. In this Part, we explain how file data, made available in a searchable form that is comparable across district boundaries, could create an informed and engaged role for the public in positive criminal justice reform.

A. The Analogy to Traffic Stop Data

Constitutional doctrines such as Batson have not opened the door to jury service for minority groups.114 But is there any better (or quicker) alternative than advocating for changes in the constitutional doctrine? The American experience with traffic stops and pedestrian stops by police over the last two decades suggest that there is, in fact, a better way. In that setting, a frustrating and limited constitutional doctrine does not tell the whole story. The increased availability of data about the patterns of police stops created a political debate that continues to shape police conduct. Through the political process, members of these communities are able to insist on changes in police department policies with the aim of reducing racial profiling.

Just as in the jury selection context under Batson, the Supreme Court’s approach to racial profiling under the Fourth Amendment allows law enforcement officials to cloak constitutionally impermissible conduct in race-neutral terms. Equal Protection jurisprudence insulates these practices from systemic reform.

The centerpiece of this evasion is Whren v. United States.115 The case involved two vice squad officers’ decision to stop a car. One possible ground for the stop was illegal driving (making a right turn without a signal); another plausible reason for the stop was the officers’ unsupported hunch that the driver and passenger were involved in drug distribution. Which was the true reason? The Court said that it didn’t matter. As long as the circumstances give officers reasonable suspicion to believe a driver violated a traffic law, courts treat the stop as reasonable under the Fourth Amendment.116 An officer can use race as a basis for suspicions about criminal behavior, stop suspects of only one race, and shroud those discriminatory stops in race-neutral language.117 David Harris summed up the impact of constitutional law on pretextual stops this way: a judicial finding of racial profiling is “the legal equivalent of lightning bolts hurled by Zeus.”118

As a result, constitutional litigation standing alone has not changed field practices very much. Numerous studies conducted over several decades have demonstrated that law enforcement officers disproportionately select racial minorities for traffic stops, disproportionately search them during these stops, and disproportionately subject minority drivers to “stop and frisk” practices.119

The greater impact of constitutional litigation was delayed and indirect. Some of the earliest statistical clues about racial profiling practices came to light during litigation over constitutional claims, which routinely ended in losses for plaintiffs who wanted to change these police practices.120 Eventually, advocates changed the venue for their arguments. They broadened their strategy and took their claims to legislatures. As a result, many states enacted legislation to address racial profiling, including some laws that require law enforcement to collect and report data about their stop practices.

As part of a strategy to prevent racial profiling, about eighteen states now require, by law, mandatory data collection for all stops and searches.121 Public agencies now make these data available to the public, sometimes through a centralized entity and at other times through individual law enforcement agencies.122

Private individuals and groups have stepped forward as intermediaries to monitor and interpret these data, making the information accessible and useful for the public and for policy entrepreneurs. Researchers employed in universities produced some studies,123 while policy advocacy organizations performed some of their own analyses.124

Journalists also found stories within these numbers. Some news outlets reported the results of academic and advocacy studies.125 In addition, teams of reporters created their own analyses, sorting and summarizing the overwhelming databases for their readers. For instance, the New York Times examined police traffic stop records between 2010 and 2015. In consent searches in Greensboro, North Carolina, “officers searched blacks more than twice as often but found contraband only 21 percent of the time, compared with 27 percent of the time with whites.”126

The collection, publication, and interpretation of traffic stop data fundamentally changed the conversation. Advocates claim that collecting data about race is the best way to gather tangible evidence of widespread unconscious bias toward minorities during police traffic stops.127 Compared to case studies or anecdotal evidence of an individual who was harmed due to police brutality or over-policing, statistical evidence might persuade a wider range of people.128

The public discussion of data also changes internal management for police departments. When the police know that data analysts and reporters are watching them work, they work more carefully.129 Where this transparency exists, reform advocates can target more precisely the local police practices that they suspect are most troubling. In some cases, the data will reveal no problems; in others, they might confirm for police leadership the factual basis for a complaint that once seemed amorphous or speculative.130

When the government collects and publishes data in a format that allows for comparisons between places, reports give the public and local police leaders a benchmark for police performance. One department that stands out from other law enforcement agencies across the state—either in a positive or negative way—can reflect on the reasons for those local differences. Similarly, data collected over time may identify trends, allowing police leaders to see in a concrete way whether a new policy is working.

In sum, the move from constitutional argument in the courtroom to political argument in the public arena loosened a stalemate on the question of police traffic stops.131 We believe that something similar can happen if government agencies collect and report jury selection data and if academics, advocates, and journalists step forward to interpret and publicize those data.132

B. The Effects of Sunshine Across Different Criminal Justice Areas

The transformative power of data, in our view, is not limited to traffic stops or jury selection. We place our proposal in the larger context of using transparency to change criminal justice practices for the better.

1. Use of Data to Regulate a Range of Actors

As Andrew Crespo has pointed out, the criminal courts already collect useful facts that remain hidden because they are scattered in single files or inaccessible formats.133 An effort to assemble these facts in aggregate form could improve the courts’ efforts to regulate the work of other criminal justice players, such as police and prosecutors.

Careful record-keeping and transparency regarding the collected data already contributes to accountability in diverse parts of the criminal justice system. In the context of correctional institutions, transparency of data has been instrumental in ensuring fair treatment of prisoners, as Alabama and other states’ courts have held that their state open-record acts apply to prisoners.134 While correctional institutions have been hesitant to comply, this requirement has shed light on prison deaths, suicides, beatings, and other prison conduct, hopefully holding these correctional institutions accountable and giving the legislature a chance to address misconduct.135

Similarly, experts have pushed for increased transparency in the context of officer-involved shootings, arguing that a lack of transparency surrounding these incidents has impeded reform.136 In a test of the reform power of data, President Obama signed the Death in Custody Reporting Act.137 This law requires states and local law enforcement agencies that receive federal money to make quarterly reports about the deaths of any persons who are detained, arrested, or incarcerated.138 The theory is that national data will help policy-makers “identify not only dangerous trends and determine whether police use force disproportionately against minorities, but best practices, and thus ultimately develop policies that prevent more deaths.”139 The next few years might reveal whether this government-mandated reporting regime can produce more comprehensive results than the more decentralized efforts of newspapers and others in the private sector to build databases of police-involved shootings.140

2. Internal Management Uses of Data

The practical impact of jury selection data depends, in part, on how prosecutors, judges, court clerks, and others use the data once the information becomes available. These criminal justice professionals have the capacity to collect for themselves the jury selection statistics and to generate reports on the topic.141 Managers in the prosecutor’s office, the chief judge’s chambers, or the clerk’s office might be more open to the use of jury selection data if they were to collect the data themselves.

On the other hand, data collection mandated by statute, statewide regulation, or rule of procedure could produce more uniform results in different localities and allow for the sort of place-to-place comparisons that make it easier to diagnose local problems. For example, the Florida legislature recently passed a pathbreaking law that requires key criminal justice actors to collect and post criminal justice data in a format that will allow comparisons across localities.142

A sense of professionalism among judges or prosecutors might motivate them to take data seriously when it shows a departure from the standard practices of their colleagues elsewhere in the state.143 After learning about patterns in jury selection across many cases, they might change practices on their own initiative. For instance, accessible data might convince supervisors to train prosecutors to avoid racial bias during jury selection.

3. External Public Uses of Data

Internal management use of routine criminal justice data is only half the story. In the end, we look to public accountability—through the ballot box or other forms of democratic input into criminal justice practices144—to convert jury selection data and other comparable datasets into drivers of change.

The information visible to the public about how prosecutors and judges perform, compared to their peers, is historically thin.145 That is starting to change. Private nonprofit organizations, such as Measures for Justice, are funding, collecting, and disseminating data that allow citizens to compare their local courts to others in the same state and elsewhere.146 Data such as this could make it possible to evaluate practices across time and across places. When news reporters, advocates, academics, and analysts interpret that data for the general public, the data could shift public priorities. It could create more informed accountability in a world where criminal court professionals get very little feedback from the communities they serve.

We do not claim to know how voters will ultimately react when these data about the criminal courts become accessible to them. It is possible that in some places, the most politically engaged members of the community will not care about jury selection; they might even resist the idea of expanding jury participation to include every population group. But local variety is built into the criminal justice systems in the United States.147 Voters and engaged community groups in most places, we hope, will value inclusive practices in their criminal courts and will expect their agents, operating in the sunshine, to deliver the results.


The fulcrum that could move jury practices sits in the office of the clerk of the court. Public employees in those offices already collect some basic background facts about prospective jurors and record the decisions by judges, prosecutors, and defense attorneys to remove jurors or to keep them. And if the clerk’s office is the fulcrum, the lever to shift the entire jury selection process in the direction of greater inclusion will be public records laws, embodied in state statutes, local court rules, and office policies.

It is startling that public courts, in an age when electronic information surrounds us on all sides, make it so difficult to track jury selection practices across different cases. It should not require hundreds of miles of driving between courthouses; access to the data should not depend on special requests for judicial approval.148 Information about the performance of public servants in the criminal courts, in aggregate form, would be easy to collect and to publish. Jury selection goes to the heart of public participation in criminal justice: this is precisely where the sun needs to shine first.

1. See infra Section II.D.

2. For a review of periodic efforts to assemble jury selection data related to specialized categories of cases (particularly in capital cases), see infra Section II.D.

3. See 42 Pa. Cons. Stat. § 4502 (2016) (declaring that citizens are not qualified to be jurors if they are “unable to read, write, speak and understand . . . English . . . ;” are not able to “render efficient jury service” due to mental infirmity; or have been “convicted of a crime punishable by imprisonment for more than one year . . . .”); Tex. Code Crim. Proc. Ann. art. 35.16 (West 2016) (allowing a challenge for cause for jurors with felony or misdemeanor convictions).

4. Judges encounter special problems during for-cause removals in death penalty cases. A juror who declares that he or she would always vote to impose the death penalty, or not to impose the death penalty, will be excluded for cause. See Witherspoon v. Illinois, 391 U.S. 510, 520–23 (1968).

5. See Mo. Rev. Stat. § 494.470 (2016) (“A prospective juror may be challenged for cause for any reason mentioned in this section and also for any causes authorized by the law.”); N.C. Gen. Stat. § 15A-1214(d)–(e) (2016).

6. See Oswalt v. State, 19 N.E.3d 241, 245 (Ind. 2014); State v. Lindell, 629 N.W.2d 223, 239–40.

7. See Ohio R. Crim. P. 24(D) (2009) (“[E]ach party peremptorily may challenge three prospective jurors in misdemeanor cases, four prospective jurors in felony cases other than capital cases . . . .”); Tenn. Code Ann. § 40-18-118 (2016) (providing eight strikes for each side in cases punishable by imprisonment for more than one year but not death, and three for each side if crime is punishable by less than one year).

8. See Miller-El v. Dretke, 545 U.S. 231, 237–39 (2005); Batson v. Kentucky, 476 U.S. 79, 89 (1986).

9. 380 U.S. 202, 222–23 (1965); Norris v. Alabama, 294 U.S. 587, 589 (1935).

10. Swain, 380 U.S. at 223.

11. 476 U.S. at 96–97.

12. See United States v. Mensah, 737 F.3d 789, 803 (1st Cir. 2013) (Asian Americans); United States v. Heron, 721 F.3d 896, 902 (7th Cir. 2013) (recognizing circuit split and state court split on religion-based challenges); United States v. Roan Eagle, 867 F.2d 436, 440–41 (8th Cir. 1989) (Native Americans); Commonwealth v. Carleton, 641 N.E.2d 1057, 1058–59 (Mass. 1994) (Irish Americans).

13. See People v. Bridgeforth, 769 N.E.2d 611, 616–17 (N.Y. 2016) (holding that removal of dark-skinned juror can satisfy step one); Hassan v. State, 369 S.W.2d 872 (Tex. Crim. App. 2012) (applying step one); City of Seattle v. Erickson, 398 P.3d 1124, 1131 (Wash. 2017) (holding that removal of only minority juror in pool can establish prima facie case).

14. See People v. Gutierrez, 395 P.3d 186, 198 (Cal. 2017) (rejecting adequacy of proffered race-neutral reasons); State v. Bender, 152 So. 3d 126, 130–31 (ruling that prosecutor not required to present arrest records in order to support race-neutral explanation for peremptory strike); People v. Knight, 701 N.W.2d 715, 730 (Mich. 2005) (finding prosecutor presented adequate race-neutral reasons for excusing prospective jurors).

15. See Wilkerson v. Texas, 493 U.S. 924, 928 (1989) (Marshall, J., dissenting) (“To excuse such prejudice when it does surface, on the ground that a prosecutor can also articulate nonracial factors for his challenges, would be absurd. . . . If such ‘smoking guns’ are ignored, we have little hope of combating the more subtle forms of racial discrimination.”); Michael J. Raphael & Edward J. Ungvarsky, Excuses, Excuses: Neutral Explanations Under Batson v. Kentucky, 27 U. Mich. J.L. Reform 229, 236 (1993) (“[I]n almost any situation a prosecutor can readily craft an acceptable neutral explanation to justify striking black jurors because of their race.”).

16. See United States v. Herrera-Rivera, 832 F.3d 1166, 1173 (9th Cir. 2016) (finding that government’s proffered reasons for striking potential juror were not pretextual and that strike was based on juror’s having criminal history and family members who used drugs); United States v. White, 552 F.3d 240, 251 (2d Cir. 2009) (accepting the explanation that a juror had “an angry look that she wasn’t happy to be here”); Lingo v. State, 437 S.E.2d 463, 471 (Ga. 1993) (prosecutor excluded black male juror who appeared “angry”); Clayton v. State, 797 S.E.2d 639, 643–45 (Ga. Ct. App. 2017) (State’s reliance on fact that prospective black juror had gold teeth was not race-neutral); State v. Clifton, 892 N.W.2d 112, 126–27 (Neb. 2017) (holding that trial court did not err in finding race-neutral the prosecutor’s rationale that juror had years of alcohol and crack addiction).

17. See Lewis v. Bennett, 435 F. Supp. 2d 184, 191–92 (W.D.N.Y. 2006) (striking unmarried juror); State v. Collins, No. M2015-01030-CCA-R3-CD, 2017 WL 2126704, at *14 (Tenn. Crim. App. May 16, 2017) (jurors had family members affected by drug abuse, prosecutor removed the only black juror).

18. See Cook v. LaMarque, 593 F.3d 810, 817 (9th Cir. 2010) (using comparative analysis of stricken versus nonstricken jurors rather than a mixed-motive test); Andrew Verstein, The Jurisprudence of Mixed Motives, 127 Yale L.J. 1106, 1116–17 (2018).

19. See Jeffrey Bellin & Junichi P. Semitsu, Widening Batsons Net to Ensnare More Than the Unapologetically Bigoted or Painfully Unimaginative Attorney, 96 Cornell L. Rev. 1075, 1092 (2011) (examining 269 Batson challenges in federal court from 2000–2009); James E. Coleman Jr. & David C. Weiss, The Role of Race in Jury Selection: A Review of North Carolina Appellate Decisions, N.C. St. B. J., Fall 2017, at 13–14 (comparing reversals in North Carolina to other southern states); Daniel R. Pollitt & Brittany P. Warren, Thirty Years of Disappointment: North Carolinas Remarkable Appellate Batson Record, 94 N.C. L. Rev. 1957, 1961 (2016).

20. See Gray v. Brady, 592 F.3d 296, 301 (1st Cir. 2010) (“[W]hether to draw an inference of discriminatory use of peremptories is an intensely case and fact-specific question . . . .”) (quoting Gray v. Brady, 588 F. Supp. 2d 140, 146 (D. Mass. 2008)). Despite the doctrinal emphasis on fact-specific judicial review of jury selection, the parties often present formulaic, prepackaged arguments to explain their removal of jurors. Litigation in this area has unearthed training materials from local prosecutor’s offices, listing ready-made “neutral” justifications that prosecutors might use to overcome a Batson challenge. See, e.g., Commonwealth v. Cook, 952 A.2d 594, 601 (Pa. 2008) (describing a training video for new prosecutors calling for prosecutors to strike black people and women from juries and explaining how to conceal discriminatory strikes). Lawyers litigating claims of racial bias in the North Carolina criminal justice system collected materials demonstrating such prosecutor training practices. See generally Catherine M. Grosso et al., A Stubborn Legacy: The Overwhelming Importance of Race in Jury Selection in PostBatson North Carolina Capital Trials, 97 Iowa L. Rev. 1531, 1535 (2012). In some instances, trainers specifically instructed prosecutors to exclude members of racial minority groups from juries. See, e.g., Miller-El v. Dretke, 545 U.S. 231, 265–66 (2005) (Dallas County); Robert P. Mosteller, Responding to McCleskey and Batson: The North Carolina Racial Justice Act Confronts Racial Peremptory Challenges in Death Cases, 10 Ohio St. J. Crim. L. 103, 104 (2012); Brian Rodgers, Local DA Encourages Blocking Blacks from Juries, Wharton County Prosecutor Says, Hous. Chron. (Mar. 22, 2016, 9:51 PM), http://www.

21. 136 S. Ct. 1737, 1743–45 (2016).

22. See, e.g., Ex parte Floyd, 227 So. 3d 1, 13 (Ala. 2016) (affirming conviction after remand to reconsider in light of Foster, despite prosecutor use of list designating jurors by race).

23. See Aliza Plener Cover, Hybrid Jury Strikes, 52 Harv. C.R.C.L. Rev. 357, 372 (2017); Scott Howe, Deselecting Biased Juries, 2015 Utah L. Rev. 289, 337 (2015); Nancy S. Marder, Foster v. Chatman: A Missed Opportunity for Batson and the Peremptory Challenge, 49 Conn. L. Rev. 1137, 1176 (2017) (proposing allowing defendants to obtain more information, such as prosecutor notes, or inferring discriminatory intent from discriminatory effect or practice); Caren Myers Morrison, Negotiating Peremptory Challenges, 104 J. Crim. L. & Criminology 1, 22 (2014); Anna Roberts, Asymmetry as Fairness: Reversing a Peremptory Trend, 92 Wash. U. L. Rev. 1503, 1541 (2015); cf. Andrew G. Ferguson, The Big Data Jury, 91 Notre Dame L. Rev. 935, 969 (2016).

24. See Rice v. Collins, 546 U.S. 333, 344 (2006) (Breyer, J., concurring) (“I continue to believe that we should reconsider Batson’s test and the peremptory challenge system as a whole.”); Bellin & Semitsu, supra note 19, at 1107; Charles J. Ogletree, Just Say No!: A Proposal to Eliminate Racially Discriminatory Uses of Peremptory Challenges, 31 Am. Crim. L. Rev. 1099, 1149 (1994); Antony Page, Batsons BlindSpot: Unconscious Stereotyping and the Peremptory Challenge, 85 B.U. L. Rev. 155, 179 (2005); Amy Wilson, The End of Peremptory Challenges: A Call for Change Through Comparative Analysis, 32 Hastings Intl & Comp. L. Rev. 363, 371 (2009); David Zonana, The Effect of Assumptions About Racial Bias on the Analysis of Batsons Three Harms and the Peremptory Challenge, 1994 Ann. Surv. Am. L. 203, 241.

25. See Leonard L. Cavise, The Batson Doctrine: The Supreme Courts Utter Failure to Meet the Challenge of Discrimination in Jury Selection, 1999 Wis. L. Rev. 501, 528 (decrying the doctrine’s “useless symbolism”); Camille A. Nelson, Batson, O.J., and Snyder: Lessons from an Intersecting Trilogy, 93 Iowa L. Rev. 1687, 1689 (2008) (“Batson’s promise of protection against racially discriminatory jury selection has not been realized.”); Bryan Stevenson, Illegal Racial Discrimination in Jury Selection: A Continuing Legacy, Hum. Rts. Mag. (Fall 2010),
vol37_2010/fall2010/illegal_racial_discrimination_in_jury_selection.html. Change might occur instead at the subconstitutional level. In April 2018, the Washington Supreme Court approved a new procedural rule that removed a showing of discriminatory intent as a basis for disallowing an improper peremptory challenge. See Wash. State Ct. Gen. R. 37.

26. In the first case to deal with the question, Strauder v. West Virginia, the Court sustained an equal protection challenge to a statute excluding black people from the jury venire. 100 U.S. 303, 309 (1880). In later cases, the Court did not require the defendant to show complete exclusion of a racial group from jury service: a substantial disparity between the racial mix of the county’s population and the racial mix of the venire, together with an explanation of how the jury selection process had created this outcome, would be enough to establish a prima facie case of discrimination. The government would then have to rebut the presumption of discrimination. See Castaneda v. Partida, 430 U.S. 482, 499 (1977) (underrepresentation of Mexican Americans); Turner v. Fouche, 396 U.S. 346, 359 (1970) (underrepresentation of black people).

27. In Taylor v. Louisiana, the Court held that a Louisiana law placing on the venire only those women who affirmatively requested jury duty violated the Sixth Amendment’s requirement that the jury represent a “fair cross section” of the community. 419 U.S. 522, 530 (1975).

28. Missouri v. Stewart, 714 S.W.2d 724, 727 (Mo. Ct. App. 1986).

29. See Duren v. Missouri, 439 U.S. 357 (1979). At that point, the burden of proof shifts to the government to show a “significant state interest” that justifies use of the method that systematically excludes a group.

30. The Court, in Berghuis v. Smith, described three different measures of the participation gap: the absolute disparity test, the comparative disparity test, and the standard deviation test. 559 U.S. 314, 316 (2010); see also State v. Plain, 898 N.W.2d 801, 826–27 (Iowa 2017) (challenges to jury pools can be based on multiple analytical models).

31. See Jessica Heyman, Introducing the Jury Exception: How Equal Protection Treats Juries Differently, 69 N.Y.U. Ann. Surv. Am. L. 185, 203 (2013).

32. Id.

33. See United States v. Fadiga, 858 F.3d 1061, 1063–64 (7th Cir. 2017) (holding that evidence that 20% of the population in the two counties that provided jurors for the district court were black and that no juror on defendant’s forty-eight person venire was black was insufficient to establish prima facie case of discrimination); United States v. Best, 214 F. Supp. 2d 897, 902–03 (N.D. Ind. 2002) (holding that jury venire did not violate Sixth Amendment fair cross-section requirement, even if percentage of black people in counties from which venire was drawn was 19.6% and percentage of black people on this venire was only 4.8%).

34. See David M. Coriell, Note, An (Un)Fair Cross Section: How the Application of Duren Undermines the Jury, 100 Cornell L. Rev. 463, 465 (2015).

35. Clerks in some states also maintain a record of the order of removal. Jurisdictions vary in how much information they collect and retain about individual jurors. See, e.g., Md. Code Ann., Cts. & Jud. Proc. § 8-314(a) (West 2016) (“A jury commissioner shall document each . . . decision with regard to disqualification, exemption, or excusal from, or rescheduling of, jury service.”); Minn. Gen. R. Practice R. 814 (2017) (“[N]ames of the qualified prospective jurors drawn and the contents of juror qualification questionnaires . . . must be made available to the public . . . .”); 42 Pa. Cons. Stat. § 4523(a) (2016) (“The jury selection commission shall create and maintain a list of names of all prospective jurors who have been disqualified and the reasons for their disqualification. The list shall be open for public inspection.”).

36. For an exception, see N.Y. Jud. Law § 528 (McKinney 2016).

The commissioner of jurors shall collect demographic data for jurors who present for jury service, including each juror’s race and/or ethnicity, age and sex, and the chief administrator of the courts shall submit the data in an annual report to the governor, the speaker of the assembly, the temporary president of the senate and the chief judge of the court of appeals.

Id. We are unaware of any state that requires the clerk of the court to collect information about the removal of jurors from the venire at the case level, in all jury trials, and to report that data routinely, both at the case level and in aggregate form. See S.B. 576, 2017 Leg. (Cal. 2017) (requiring jury commissioner to develop a form to collect specified demographic information about prospective jurors, prohibiting disclosure of the form, but also requiring jury commissioner to release biannual reports with aggregate data).

37. The best overview of these shortcomings in the public records appears in Catherine M. Grosso & Barbara O’Brien, A Call to Criminal Courts: Record Rules for Batson, 105 Ky. L.J. 651, 654 (2017); see also Russell D. Covey, The Unbearable Lightness of Batson: Mixed Motives and Discrimination in Jury Selection, 66 Md. L. Rev. 279, 322 (2007) (“[T]here is extremely little evidence available even in a full-blown Batson hearing to shed much light on the question of whether an explanation is credible.”); Peter A. Joy & Kevin C. McMunigal, Racial Discrimination and Jury Selection, 31 Crim. Just., Summer 2016, at 43, 45 (“[E]very jurisdiction needs to do a better job of collecting data both on the composition of the jury venires and on the use of peremptory challenges.”); Mary R. Rose & Jeffrey B. Abramson, Data, Race, and the Courts: Some Lessons on Empiricism from Jury Representation Cases, 2011 Mich. St. L. Rev. 911, 954–56 (noting poor quality of juror data that courts maintain and report).

38. See David C. Baldus et al., Racial Discrimination and the Death Penalty in the PostFurman Era: An Empirical and Legal Overview, with Recent Findings from Philadelphia, 83 Cornell L. Rev. 1638, 1662 (1998); David C. Baldus et al., Statistical Proof of Racial Discrimination in the Use of Peremptory Challenges: The Impact and Promise of the Miller-El Line of Cases as Reflected in the Experience of One Philadelphia Capital Case, 97 Iowa L. Rev. 1425, 1449 (2012).

39. See Grosso et al., supra note 20, at 1533; Barbara O’Brien & Catherine M. Grosso, Beyond Batsons Scrutiny: A Preliminary Look at Racial Disparities in Prosecutorial Preemptory Strikes Following the Passage of the North Carolina Racial Justice Act, 46 U.C. Davis L. Rev. 1623, 1627 (2013).

40. See Ann M. Eisenberg et al., If It Walks like Systematic Exclusion and Quacks like Systematic Exclusion: FollowUp on Removal of Women and AfricanAmericans in Jury Selection in South Carolina Capital Cases, 19972014, 68 S.C. L. Rev. 373, 373 (2017); Ann M. Eisenberg, Removal of Women and African Americans in Jury Selection in South Carolina Capital Cases, 19972012, 9 Ne. U. L. Rev. 299, 302 (2017).

41. See David C. Baldus et al., The Use of Peremptory Challenges in Capital Murder Trials: A Legal and Empirical Analysis, 3 U. Pa. J. Const. L. 3, 22–28 (2001); Aliza Plener Cover, The Eighth Amendments Lost Jurors: Death Qualification and Evolving Standards of Decency, 92 Ind. L.J. 113, 116 (2016) (qualitative study of Witherspoon strikes in eleven Louisiana trials resulting in death verdicts from 2009 to 2013); Brandon L. Garrett et al., Capital Jurors in an Era of Death Penalty Decline, 126 Yale L. J.F. 417, 419 (2017) (survey of persons reporting for jury duty in Orange County, California, asking questions about eligibility to serve on hypothetical death penalty case); Justin D. Levinson et al., Devaluing Death: An Empirical Study of Implicit Racial Bias on JuryEligible Citizens in Six Death Penalty States, 89 N.Y.U. L. Rev. 513, 520 (2014) (analyzing nonarchival study of 445 jury-eligible citizens in six death penalty states).

42. Two noncapital studies analyzed single parishes in Louisiana. See La. Crisis Assistance Ctr., Blackstrikes: A Study of the Racially Disparate Use of Peremptory Challenges by the Jefferson Parish District Attorneys Office 2 (2003),; Billy M. Turner et al., Race and Peremptory Challenges During Voir Dire: Do Prosecution and Defense Agree?, 14 J. Crim. Just. 61, 63 (1986) (examining data from 121 criminal trials in one Louisiana parish). Another working paper analyzed 351 jury trials from Los Angeles County, Maricopa County (Arizona), Bronx County, and Washington, D.C. See Jee-Yeon K. Lehmann & Jeremy Blair Smith, A Multidimensional Examination of Jury Composition, Trial Outcomes, and Attorney Preferences 9 (2013),

43. See Shamena Anwar et al., The Impact of Jury Race in Criminal Trials, 127 Q.J. Econ. 1017, 1026 (2012). Some of the single-jurisdiction studies collected data about juries for a remarkably small number of cases. See Mary R. Rose, The Peremptory Challenge Accused of Race or Gender Discrimination? Some Data from One County, 23 Law & Hum. Behav. 695, 697 (1999) (compiling data from thirteen noncapital felony criminal jury trials in North Carolina; black people were much more likely to be excluded by the prosecution and white people by the defense).

44. See Maureen M. Berner et al., A Process Evaluation and Demographic Analysis of Jury Pool Formation in North Carolinas Judicial District 15B, at 2 (2016),’s-judicial-district; Bob Cohen & Janet Rosales, Racial and Ethnic Disparity in Manhattan Jury Pools: Results of a Survey and Suggestions for Reform 1 (2007),
reports/Citizen-Action-Jury-Pool-Study.pdf; James Michael Binnall, A Field Study of the Presumptively Biased: Is There Empirical Support for Excluding Convicted Felons from Jury Service?, 36 Law & Poly 1, 3 (2014); Edward J. Bronson, On the Conviction Proneness and Representativeness of the DeathQualified Jury: An Empirical Study of Colorado Veniremen, 42 U. Colo. L. Rev. 1, 4 (1970); Ted M. Eades, Revisiting the Jury System in Texas: A Study of the Jury Pool in Dallas County, 54 SMU L. Rev. 1813, 1814 (2001). 

45. See Equal Justice Initiative, Illegal Racial Discrimination in Jury Selection: A Continuing Legacy 4 (2010), (summarizing statistics indicating racial disparities among prosecutors during jury selection for eight southern states: Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, South Carolina, and Tennessee); Grosso & O’Brien, supra note 37, at 657 (summarizing collection of jury selection data in capital litigation context).

46. See Steve McGonigle et al., Striking Differences, Dall. Morning News, Aug. 21–23, 2005 (finding that in felony trials in Dallas County, Texas, prosecutors tended to reject black jurors, while defense attorneys tended to retain them).

47. See Kenneth J. Melilli, Batson in Practice: What We Have Learned About Batson and Peremptory Challenges, 71 Notre Dame L. Rev. 447, 463 (1996) (inferring that criminal defendants make approximately 90% of Batson claims; only 17% of challenges with black people as the targeted group were successful, 13% for Hispanic people, and 53% for white people).

48. See Shaun L. Gabbidon et al., RaceBased Peremptory Challenges: An Empirical Analysis of Litigation from the U.S. Court of Appeals, 20022006, 33 Am. J. Crim. Just. 59, 62 (2008) (analyzing 184 race-based peremptory challenge cases, concluding that appellants rarely win such challenges); Pollitt & Warren, supra note 19, at 1962. In light of the challenges of assembling archival data, some researchers opt instead for experimental studies. See Samuel R. Sommers & Michael I. Norton, Race and Jury Selection: Psychological Perspectives on the Peremptory Challenge Debate, 63 Am. Psychologist 527, 533–34 (2008).

49. We began this effort in the fall of 2012, so we chose the most recent complete year of records. The state constitution at the time guaranteed that all felony trials in the state would be tried to a jury. N.C. Const. art. I, § 24. Only a few misdemeanor charges were decided by juries: those “appealed” from the district court to the superior court for a trial de novo. See N.C. Gen. Stat. § 7A-271(b) (2016) (providing for appeals from district court to superior court).

50. Annual case activity reports for felonies, misdemeanors, and infractions appear at Case Activity ReportsFiscal Year 20162017, N.C. Ct. Sys., (last visited May 18, 2018).

51. Our contact in NCAOC had cooperated with past data requests, with minimal burden on the office, but asserted that NCAOC leadership appointed by the governor who was elected in 2012 had instructed employees not to cooperate with this type of request. Recent litigation established that court records are housed in the clerks’ offices, not in a centralized file housed with the NCAOC. See LexisNexis Risk Data Mgmt., Inc. v. N.C. Admin. Office of the Courts, 775 S.E.2d 651, 656 (N.C. 2015).

52. A few counties (such as Guilford and Mecklenburg) maintained their own records about the cases that proceeded to trial. In those cases, we relied on the county clerk’s records to identify cases that proceeded to trial. In one case (New Hanover County), our researcher focused on “thick files” in the collection as a rough proxy for the cases that went to trial. In other cases, we asked the county clerk to request from the NCAOC a list of trials for that county. NCAOC treated requests from the county clerk of the superior court as a legal obligation, unlike statewide requests from scholars.

53. We are grateful to Trevor Hughes and Matt Nelkin for their work on this project.

54. NCAOC data track the number of criminal charges resolved through trials, while our database records the number of criminal trials, treating multi-charge or multi-defendant cases as a single trial. We collected jury selection data on 1,307 trials, while NCAOC listed 2,112 charges resolved by jury trial for fiscal year 2011–2012.

55. We also plan to keep this research project open for some years and will add further trials to the 2011 data as they come to our attention.

56. We were disappointed to find that some clerks recorded only the fact that a prospective juror was removed from the box without indicating which courtroom actor was responsible for the removal. We coded these jurors as “Removed.” The jury form also usually indicated the order of removals for any particular actor (that is, the form showed that a prospective juror was the third peremptory challenge by the defense or the fourth removal for cause by the judge) but not the overall order of removal of jurors in the voir dire process. One county (Guilford) adopted a notation that did capture this information about the overall order of removals.

57. Based on what we learned from the pilot study, we refined a data collection protocol for students, as recorded in a codebook and standard spreadsheet. The field researchers focused on trials in 2011, but in smaller counties with very few trials per year, they also collected information for trials in 2010 and 2012. We are grateful to Elizabeth Johnson, a reference librarian at the school of law, for coordinating this complex field operation. See Liz McCurry Johnson, Accessing Jury Selection Data in a PreDigital Environment, 41 Am. J. Trial Advoc., Summer 2017, at 45, 49.

58. The counties with no jury trials were Bertie, Camden, Chowan, Clay, Franklin, Madison, Mitchell, Montgomery, Pamlico, and Warren.

59. See N.C. Gen. Stat. § 9-4 (2016).

60. See Grosso et al., supra note 20, at 1533.

61. The board of elections provides online data including the name, home address, gender, race, age, and party affiliation of each voter. See Voter Search, N.C. St. Board Elections & Ethics Enforcement, (last visited May 18, 2018). A few counties (including Mecklenburg) adopted notation techniques that included a record of each juror’s race and gender within the clerk’s file. Students worked on matching juror profiles with voter records between spring 2013 and summer 2016.

62. See N.C. Gen. Stat. § 9-2(b) (“In preparing the master list [of prospective jurors], the jury commission shall use the list of registered voters and persons with driver’s license records supplied to the county by the Commissioner of Motor Vehicles . . . .”).

63. We gave researchers a protocol to follow when deciding whether a prospective juror from the clerk’s records matched a voter from the online board of elections records. The clerks in some offices provided us with the jury venire lists, which they maintained separately from the files for each trial; the venire lists provided home addresses for the jurors, increasing our confidence that the jurors listed in the clerk’s records matched the voters listed in the voter records for the county. After clerks learned that we were asking for access to file information about jurors, some superior court judges issued orders prohibiting the clerks from releasing the juror venire lists to anyone other than the parties to the case. The North Carolina General Assembly also amended the statute to restrict access to the addresses and birthdates recorded on the jury venire lists. See N.C. Gen. Stat. § 9-4(b); 2013 N.C. Sess. Laws 166; 2012 N.C. Sess. Laws 180.

64. In some cases, this information was available from the public data stored on the site of the North Carolina State Bar regarding licensed attorneys. See Search for a North Carolina Lawyer, N.C. St. B., (last visited May 18, 2018). We also learned which office defense attorneys worked in (private firm or public defender’s office). In North Carolina, the public defender service covers sixteen of the judicial districts in the state. The remaining districts operate with appointed counsel. See N.C. Gen. Stat. § 7A-498.7. Students followed a written protocol to search in standard locations and a prescribed order for the professional biographies of the courtroom actors.

65. North Carolina divides the state into forty-four different prosecutorial districts and thirty different superior court districts. See N.C. Gen. Stat. § 7A-41. The judicial districts break into eight different divisions; judges spend six months each year in their home district and six months traveling to other districts within the division.

66. The NCAOC data list a total of 2,112 charges that were resolved through trial for fiscal year 2011–2012. The breakdown of charges for individual counties suggests that we obtained the records for almost every felony trial that occurred in the state during calendar year 2011. The total number of defendants who faced trial in North Carolina in 2011 remains speculative because each prosecutor retains the discretion to file separate counts either as separate file numbers in the office of the clerk or as separate counts covered under a single file number.

67. We checked the quality of the field data during the process of loading county-specific spreadsheets into the central database. Another statewide version of the data exists in spreadsheet form, as assembled by Dr. Francis Flanagan of the Wake Forest University Department of Economics. See generally Francis X. Flanagan, Peremptory Challenges and Jury Selection, 58 J.L. & Econ. 385 (2015); Francis X. Flanagan, Race, Gender, and Juries: Evidence from North Carolina (2017) (unpublished article) (on file with the author) [hereinafter Flanagan, North Carolina Jury Evidence].

68. Our field researchers entered separate codes for guilty as charged, guilty of lesser charge, mistrial, and acquittal.

69. These unexplained removals were based on incomplete records in a few counties. If we assume that the courtroom actors accounted for the “unknown” removals at the same rate that they did for the recorded cases, then defense attorneys removed a total of 15% of the pool, judges excluded 12% for cause, and prosecutors removed 11%.

70. See N.C. Gen. Stat. § 15A-1214.

71. See id. § 15A-1213.

72. See id. § 15A-1214(b).

73. The judge sometimes removes jurors for cause before the parties ask their questions, but the judge always remains free to remove additional jurors in light of their answers to attorney questions. Defense attorneys examine jurors only after prosecutors tender a complete set of twelve jurors. See id. § 15A-1214(c).

74. When jurors are replaced at any step along the way, the initiative passes again to the judge and the prosecutor, who may remove any new juror before the prosecutor “tenders” the newest set of retained jurors to the defense attorney. See id. § 15A-1214(d), (f). In capital cases, the process may advance one juror at a time. See id. § 15A-1214(j).

75. Local variations in this removal process and gaps in the file records leave us uncertain about the precise order of removals of jurors from any given trial. For instance, it is possible for the judge and the prosecutor to retain all twelve jurors initially placed in the box, for the defense attorney to exercise all six of the available peremptories, and then for the judge and prosecutor to remove some of the replacement jurors for those six boxes. In most counties, the clerk records the order of jurors removed by each particular actor (for instance, “D3” would indicate the third juror removed by defense counsel), but not the order of removals as between parties. Only one county (Guilford) tracked the order of removal overall.

76. The voter registration and juror records use the racial categories white, black, Asian, Hispanic, Native American, and other. Voters self-identify and do not have the option of choosing more than one race. Because of the small numbers recorded in four of those categories, we combine them into a single “other” category. Based on current census figures, we believe that these figures underestimate the number of Hispanic or Latino citizens called for jury service in felony trials today. White residents (excluding Hispanic or Latino ethnicity) comprised 65.3% of the 2010 population, while “Black or African American alone” residents made up 21.5%, and “Hispanic or Latino” residents made up 8.4% of the state population at that time. See Quick Facts: North Carolina, U.S. Census Bureau (July 1, 2017),

77. These jurors did not appear in the voter database or appeared in the voter database with race not indicated. Jurors not appearing in the voter database were placed into the juror pool in the county based on their appearance on the list of licensed drivers. The race of licensed drivers is not publicly available data in North Carolina. If the jurors whose race was unknown were assigned a racial identity in proportion to the rest of the pool, black jurors would constitute 20% of the pool. Under this scenario, white jurors would constitute 77% of the total pool, and other races would make up 3%.

78. The different removal rates for jurors of different races by each of the three courtroom actors are all statistically significant, using the chi-square test for significance.

79. We calculated this ratio after excluding the removals by unknown parties and the removal of jurors of unknown race. In every case, the rate of removal of jurors of unknown race sat in between the rate of removal for white jurors and for nonwhite jurors.

80. Judges have access to the entire pool. Prosecutors choose from the jurors remaining after the judge has chosen, while defense attorneys make their decisions regarding the jurors left after the prosecutors and judges have acted. There is some imprecision in this method because after one of the parties has exercised its full complement of peremptories, the clerk might place additional jurors into the box. While the attorneys may still challenge these additional jurors for cause, the removal depends on establishing the relevant legal basis for removal. The number of jurors that a party “retains” therefore includes some jurors that the party did not actively choose.

81. The retention rate for female jurors overall was 55%; for male jurors it was 55.4%. Judges removed 13% of females and 11.7% of males; prosecutors removed 12.1% of female and 13.8% of male jurors available to them; defense attorneys removed 21.5% of female and 20.6% of male jurors available to them. It is possible, on the basis of Jury Sunshine Project data, to compare the treatment of male and female prospective jurors in particular categories of cases, such as sexual assault or domestic violence charges. We reserve those questions for another time, concentrating here on the insights one can gain from exploring all felony trials as a group.

82. The percentages in Table 5 are based on the pool of jurors after excluding those with an unknown removal source. The percentages for prosecutors and defense attorneys also reflect the reduced pool of jurors available to those actors at the relevant point in the process. The differences in treatment between white and nonwhite jurors are statistically significant, using the chi-square test. For each group of actors, the p-value is < 0.00001.

83. Black males make up approximately 11% of the state population overall. We note for future research the potential relevance of the race and gender of the judges, prosecutors, and defense attorneys who select the jurors.

84. We designate the most rural counties as the thirty-three counties with the lowest population densities in the state. See North Carolina Population Density County Rank,,–population-density–county-rank.htm (last visited May 18, 2018). Among those thirty-three counties, eight conducted no jury trials at all and eleven recorded generic removals without attributing them to the judge or a party. Those counties made choices regarding 2,706 jurors (or 2,199 when excluding the jurors with an unknown removal source). For purposes of Table 7, we designated the most urban counties as the eleven counties with the highest population densities, covering all cities with populations more than 80,000. Those counties made choices about 13,037 jurors. The racial differences in rates of juror removal for each of the actors, as well as the urban-rural differences reflected in the removal ratios in Table 7, are statistically significant.

85. All three courtroom actors—judges, prosecutors, and defense attorneys—removed fewer available jurors in rural counties than they did in urban counties. Judges removed 15.7% of available jurors in urban counties, and only 8.1% in rural counties. The comparable figures for prosecutors were 14.3% and 8.4%; for defense attorneys, they were 22.3% and 12.3%.

86. See Flanagan, North Carolina Jury Evidence, supra note 67, at 14.

87. Id. at 13–15. Flanagan used instrumental variable regressions, using the demographic composition of the randomly selected jury pool as an instrument for the composition of the jury.

88. There is also another possible explanation for the exclusion pattern on the defense side: perhaps defense attorneys were aware that nonwhite jurors were underrepresented on the venire that the clerk called to the courthouse. Their removal of white jurors, then, might have revealed an effort to restore the jury to a racial balance that better reflected the community. See Berner et al., supra note 44, at 7.

89. The judges’ different treatment of white jurors and nonwhite jurors other than black jurors is equally puzzling. It might reflect a greater incidence of language barriers within this group, but that is speculation.

90. Cf. Michael Selmi, Statistical Inequality and Intentional (Not Implicit) Discrimination, 79 Law & Contemp. Probs. 199, 206 (2016).

91. See Binnall, supra note 44, at 3; Vida B. Johnson, Arresting Batson: How Striking Jurors Based on Arrest Records Violates Batson, 34 Yale L. & Poly Rev. 387, 389 (2016); Anna Roberts, Casual Ostracism: Jury Exclusion on the Basis of Criminal Convictions, 98 Minn. L. Rev. 592, 593 n.12 (2013).

92. See Flanagan, North Carolina Jury Evidence, supra note 67, at 14. Among the 1,327 defendants in our database, 666 (50%) are black males and 385 (29%) are white males. The race is unknown for 71 male defendants (5%). There are 74 (6%) black female defendants and 63 (5%) white female defendants.

93. See Bellin & Semitsu, supra note 19, at 1082–83.

94. See Akhil R. Amar, Americas Constitution: A Biography 15, 205 (2005); Stephanos Bibas, The Machinery of Criminal Justice 70 (2012).

95. 396 U.S. 320, 329 (1970).

96. See United States v. Conant, 116 F. Supp. 2d 1015, 1020–22 (E.D. Wis. 2000) (“While no court has yet recognized a constitutional right to serve on a jury, the possibility that such a right might exist is to be given the most careful scrutiny.”).

97. See Ciudadanos Unidos de San Juan v. Hidalgo Cty. Grand Jury Comm’rs, 622 F.2d 807, 825 (5th Cir. 1980) (quoting Smith v. Texas, 311 U.S. 128, 130 (1940)).

It is part of the established tradition in the use of juries as instruments of public justice that the jury be a body truly representative of the community. For racial discrimination to result in the exclusion from jury service of otherwise qualified groups not only violates our constitution and the laws enacted under it but is at war with our basic concepts of a democratic society and a representative government.

Id.; see also Cassell v. Texas, 339 U.S. 282, 303–04 (1950) (Jackson, J., dissenting).

Qualified Negroes excluded by discrimination have available, in addition, remedies in courts of equity. I suppose there is no doubt, and if there is this Court can dispel it, that a citizen or a class of citizens unlawfully excluded from jury service could maintain in a federal court an individual or a class action for an injunction or mandamus against the state officers responsible.

Cassell, 339 U.S. at 303–04.

98. White v. Crook, 251 F. Supp. 401, 406 (M.D. Ala. 1966) (quoting Strauder v. West Virginia, 100 U.S. 303, 308 (1879)); see also Nancy Leong, Civilizing Batson, 97 Iowa L. Rev. 1561, 1564 (2012) (proposing suits by prospective jurors to overcome informational obstacles to Batson challenges).

99. See Hiroshi Fukurai, Race, Social Class, and Jury Participation: New Dimensions for Evaluating Discrimination in Jury Service and Jury Selection, 24 J. Crim. Just., no. 1, 1996, at 71, 72 (quoting Apodaca v. Oregon, 406 U.S. 404, 410 (1972)).

100. See Taylor v. Louisiana, 419 U.S. 522, 526–27 (1975).

101. Id. at 530–31 (internal quotation marks omitted) (quoting Thiel v. S. Pac. Co., 328 U.S. 217, 227 (1946) (Frankfurter, J., dissenting)). Similarly, after the Court’s decision in Batson, the Court decided in Powers v. Ohio, 499 U.S. 400 (1991), to expand the right to complain against discriminatory use of peremptory challenges to defendants who were not members of the same race as the excluded jurors. The harm done to the community’s interest in jury service served as a key justification: “Jury service is an exercise of responsible citizenship by all members of the community, including those who otherwise might not have the opportunity to contribute to our civic life.” Powers, 499 U.S. at 402.

102. See Amar, supra note 94, at 15, 205; Vikram David Amar & Alan Brownstein, The Hybrid Nature of Political Rights, 50 Stan. L. Rev. 915, 981–94 (1998) (exploring historical basis for treating jury selection as a political right affecting the community).

103. 505 U.S. 42, 49 (1992) (quoting Batson v. Kentucky, 476 U.S. 79, 87 (1986)). The McCollum Court noted that “[t]he harm from discriminatory jury selection extends beyond that inflicted on the defendant and the excluded juror to touch the entire community.” Id. (internal quotation marks omitted) (quoting Batson, 476 U.S. at 87).

104. Batson, 476 U.S. at 87. This is a key insight from the “procedural justice” literature. See Richard R. Johnson, Citizen Expectations of Police Traffic Stop Behavior, 27 Policing: Intl J. Police Strategies & Mgmt. 487, 488 (2004) (noting that studies have shown that people are more likely to “defer to the law and refrain from illegal behavior” when police treat them fairly); Tom R. Tyler & Jeffery Fagan, Legitimacy and Cooperation: Why Do People Help the Police Fight Crime in Their Communities, 6 Ohio St. J. Crim. L. 231, 233 (2008).

105. See Tyler & Fagan, supra note 104, at 235–36. The 1980 Miami urban rebellion resulted in the death of eighteen people and $200 million in property damage and other losses. This rebellion followed an all-white jury acquitting four white police officers for the beating death of a black insurance executive after a change of venue from Miami to Tampa and after the defendants had used their peremptory challenges to exclude all black people on the jury venire. See Ihosvani Rodriguez, McDuffie Riots Shook Miami, Sun Sentinel (May 16, 2005), The Florida governor’s report of the disturbance specifically identified the practice of excluding black people from juries in racially sensitive cases as a cause of the riots and a reason for black people in Dade County to distrust the criminal justice system. Governor Bob Grahams Dade Cty. Comm., Report of Governors Dade County Citizens Committee 60–61 (Oct. 30, 1980),

106. Adam Benforado, Flawed Humans, Flawed Justice, N.Y. Times (June 13, 2015),

107. See M. Shanara Gilbert, An Ounce of Prevention: A Constitutional Prescription for Choice of Venue in Racially Sensitive Criminal Cases, 67 Tul. L. Rev. 1855, 1928 (1993).

108. Barbara D. Underwood, Ending Race Discrimination in Jury Selection: Whose Right Is It, Anyway?, 92 Colum. L. Rev. 725, 749 (1992).

109. 407 U.S. 493 (1972).

110. Id. at 503–04.

111. There is an ironic aspect to the Jury Sunshine Project: publication of data about uneven community access to jury service might exacerbate the problem by making it more visible. If the public debate never results in greater equality of jury service, that outcome is a sobering possibility.

112. This compounds the other weaknesses of the electoral check on the prosecutor’s performance in office. See Russell M. Gold, Promoting Democracy in Prosecution, 86 Wash. L. Rev. 69, 88–89 (2011); Ronald F. Wright, How Prosecutor Elections Fail Us, 6 Ohio St. J. Crim. L. 581, 582–83 (2009).

113. See Stephen Clarke, Arrested Oversight: A Comparative Analysis and Case Study of How Civilian Oversight of the Police Should Function and How It Fails, 43 Colum. J.L. & Soc. Probs. 1, 2 (2009); Kami Chavis Simmons, Beginning to End Racial Profiling: Definitive Solutions to an Elusive Problem, 18 Wash. & Lee J.C.R. & Soc. Just. 25, 30 (2011).

114. See supra Section II.B.

115. 517 U.S. 806 (1996); see also Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 265 (1977) (“Proof of racially discriminatory intent or purpose is required to show a violation of the Equal Protection Clause.”); Carlos Torres et al., Indiscriminate Power: Racial Profiling and Surveillance Since 9/11, 18 U. Pa. J.L. & Soc. Change 283, 285 (2015).

116. Whren, 517 U.S. at 819.

117. See Michael L. Birzer, Racial Profiling 72 (2013). A few examples confirm the limited power of equal protection doctrine to respond to racial profiling. In United States v. Avery, 137 F.3d 343 (6th Cir. 1997), the court turned aside the defendant’s equal protection claim and rejected statistics showing that police disproportionately targeted black people because the officers had a plausible, nonracial reason for detaining the defendant. Similarly, in Bingham v. City of Manhattan Beach, 329 F.3d 723, 736 (9th Cir. 2003), the Ninth Circuit affirmed summary judgment because the appellant failed to provide evidence to refute the officer’s race-neutral explanation for the traffic stop. See also Johnson v. Crooks, 326 F.3d 995, 999–1000 (8th Cir. 2003) (denying relief because plaintiff failed to provide evidence of discrimination to counter the officer’s race-neutral justification of the traffic stop).

118. David A. Harris, Racial Profiling Redux, 22 St. Louis U. Pub. L. Rev. 73, 75 (2003).

119. See, e.g., David Barstow & David Kocieniewski, Records Show New Jersey Police Withheld Data on Race Profiling, N.Y. Times (Oct. 12, 2000),; David A. Harris, Am. Civil Liberties Union, Driving While Black: Racial Profiling On Our Nations Highways ACLU (June 1999), (describing statistics from Maryland and Illinois). More recent data related to New York City’s “stop and frisk” policy tell a consistent story. Nearly nine out of every ten people that the New York Police Department stopped and frisked were completely innocent. Although black people and Hispanic people account for a little over half of the city’s population, 83% of the people stopped were black or Hispanic. See Racial Discrimination in StopandFrisk, N.Y. Times (Aug. 12, 2013),

120. See Harris, supra note 118, at 78.

121. See NAACP, Born Suspect: StopandFrisk Abuses & the Continued Fight to End Racial Profiling in America app.1 (Sept. 2014),; Patrick McGreevy, Brown Signs Legislation to Protect Minorities from Racial Profiling and Excessive Force, L.A. Times (Oct. 4, 2015, 3:00 AM), In 1999, North Carolina became the first state to mandate data collection regarding race for police who stop drivers. N.C. Gen. Stat. § 143B-902 (2016); R.I. Gen. Laws § 31-21.2-5(e) (2016).

122. Since 2002, all state highway patrol and police departments in North Carolina have collected the data and sent them to the North Carolina Department of Justice, which publishes the data through its website. See North Carolina Traffic Stop Statistics, N.C. Dept Pub. Safety, (last visited May 18, 2018).

123. One such academic study, by Frank Baumgartner, reported that black drivers were on average 73% more likely to be searched than white drivers in North Carolina. See Frank R. Baumgartner, NC Traffic Stops, U.N.C. Chapel Hill, (last updated Dec. 13, 2017) (concluding that Hispanic drivers were 96% more likely to be searched than white drivers and black male drivers were 97% more likely to be searched, yet black men were 10% less likely to have illegal substances than white men in probable cause searches; during consent searches, black men were 18% less likely to have illegal substances than their white counterparts).

In a separate study based on 4.5 million traffic stop records, Sharad Goel and other researchers at Stanford University found that 5.4% of black drivers were searched, compared to 3.1% of white drivers. See Camelia Simoiu et al., The Problem of InfraMarginality in Outcome Tests for Discrimination, 11 Annals Applied Stat. 1193, 1206 (2017), (revealing that, in nearly every department, black and Hispanic drivers were subject to a lower threshold of suspicion than their white and Asian counterparts; statewide, the thresholds for searching white people were 15%, for Asian people 13%, for black people 7%, and for Hispanic people 6%).

124. See Richard A. Oppel, Jr., Activists Wield Search Data to Challenge and Change Police Policy, N.Y. Times (Nov. 20, 2014), In 2015, the Southern Coalition for Social Justice published an interactive map on their website that allows a viewer to search the North Carolina stop data by police department. See Open Data Policing, S. Coalition Soc. Just., (last visited May 18, 2018).

125. See Tonya Maxwell, In Traffic Stops, Disparity in Black and White, Asheville CitizenTimes (Aug. 27, 2016, 2:34 PM), (describing Simoiu et al., supra note 123).

126. See Sharon LaFraniere & Andrew W. Lehren, The Disproportionate Risks of Driving While Black, N.Y. Times (Oct. 25, 2015), (city’s driving population is 39% black; 54% of those pulled over were black); see also Matthew Kauffman, Data: Minority Motorists Still Pulled Over, Ticketed at Higher Rates than Whites, Hartford Courant (Sept. 22, 2015, 7:02 PM),

127. Lorie Fridell et al., Racially Biased Policing: A Principled Response 116–17 (2001),; cf. Stephen Rushin, Using Data to Reduce Police Violence, 57 B.C. L. Rev. 117, 129–31 (2016).

128. Fridell et al., supra note 127, at 128. For a discussion of methodology issues in these studies, see Joyce McMahon et al., U.S. Dept Justice, Office of Cmty. Oriented Policing Services, How to Correctly Collect and Analyze Racial Profiling Data: Your Reputation Depends on It! 35 (2002), (last visited May 18, 2018). Critics argue that unless the record of the stop includes very specific data points, down to the cross streets where the stop occurred (which in many cases is not a required data point), there is no record of which areas of the jurisdiction are facing the most police presence. The specific location of the stop, according to this argument, is necessary to put the stop into context.

129. Martin Kaste, Police Are Learning to Accept Civilian Oversight, but Distrust Lingers, NPR (Feb. 21, 2015, 10:18 AM),

130. Sometimes, of course, police leaders offer benign interpretations of the data and deny any need for policy changes. See Joey Garrison, Nashville Police Chief Slams Racial Profiling Report as Morally Disingenuous, Tennessean (Mar. 7, 2017, 12:58 PM),

131. As a result of the New York Times investigation in 2015, the Greensboro police chief ordered officers to refrain from stopping drivers for minor infractions involving vehicle flaws, which are stops that are subject to individual officer discretion and stops for which black people and Hispanic people were more likely to be pulled over. See Sharon LaFraniere, Greensboro Puts Focus on Reducing Racial Bias, N.Y. Times (Nov. 11, 2015),; Oppel, supra note 124.

After having initially rejected protesters’ demands, the city [of Durham, North Carolina] . . . agreed to require the police . . . to obtain written consent to search vehicles in cases where they do not have probable cause. . . . “Without the data, nothing would have happened,” said Steve Schewel, a Durham City Council member . . . .

Oppel, supra note 124.

132. For an example of news coverage drawing on relevant, but limited, demographic information related to jury selection, see Pam Kelley & Gavin Off, Wes Kerrick Jury Wont Mirror Mecklenburgs Diversity, Charlotte Observer (July 27, 2015, 8:51 PM), (comparing jury pool in the criminal trial of a police officer who shot a suspect with overall county population demographics).

133. See Andrew Manuel Crespo, Systemic Facts: Toward Institutional Awareness in Criminal Courts, 129 Harv. L. Rev. 2049, 2109–10 (2016).

134. See Sarah Geraghty & Melanie Velez, Bringing Transparency and Accountability to Criminal Justice Institutions in the South, 22 Stan. L. & Poly Rev. 455, 460 (2011).

135. Id. at 458–63.

136. Mark Berman & Mark Guarino, Chicago Releases Unprecedented Evidence from Nearly 100 Investigations into Police Shootings, Use of Force, Wash. Post (June 3, 2016), https://www.washingtonpost.

137. Death in Custody Reporting Act of 2013, Pub. L. No. 113-242, 128 Stat. 2860 (2014).

138. Id. § 2(a).

139. See Kami Chavis Simmons, No Way to Tell Without a National Database, N.Y. Times: Room for Debate (July 13, 2016, 10:53 AM),

140. See Geoffrey P. Alpert, Toward a National Database of OfficerInvolved Shootings: A Long and Winding Road, 15 Criminology & Pub. Poly 237, 238–39 (2015); 2015 Washington Post Database of Police Shootings, Wash. Post, (last visited May 18, 2018) (displaying police shooting data drawn from “news reports, public records, Internet databases and original reporting”).

141. See Alafair S. Burke, Prosecutors and Peremptories, 97 Iowa L. Rev. 1467, 1485, 1485 n.97 (2012) (collecting proposals that would require prosecutors to maintain jury selection statistics); Jason Kreag, Disclosing Prosecutorial Misconduct, 72 Vand. L. Rev. (forthcoming 2019) (proposing the use of standardized letters to disclose prosecutor discovery violations to affected parties).

142. See Fla. Stat. Ann. § 900.05(3), (4) (2018); John Kennedy, Governor Signs Sweeping Court Data Collection, Sarasota HeraldTribune (Mar. 30, 2018),

143. See Sidney Shapiro & Ronald F. Wright, The Future of the Administrative Presidency: Turning Administrative Law InsideOut, 65 U. Miami L. Rev. 577, 587–90 (2011) (analyzing the restraining power of professional norms in bureaucracies such as prosecutor’s offices).

144. See Jocelyn Simonson, Democratizing Criminal Justice Through Contestation and Resistance, 111 Nw. U. L. Rev. 1609, 1621 (2017); Jocelyn Simonson, The Criminal Court Audience in a PostTrial World, 127 Harv. L. Rev. 2173, 2177 (2014).

145. See Russell M. Gold, Clientless Prosecutors, 51 Ga. L. Rev. 693, 701 (2017); Jason Kreag, Prosecutorial Analytics, 94 Wash. U. L. Rev. 771, 776–77 (2017); Ronald F. Wright, Beyond Prosecutor Elections, 67 SMU L. Rev. 593, 594 (2014). For a remarkable recent example of a prosecutor committing to regular release to the public of its own statistics about charging decisions, see Tanveer Ali, Cook County Felony Weapon Cases Up 43 Percent in 2017, Data Shows, Chi. SunTimes (Feb. 21, 2018, 3:24 PM),
news/felony-weapon-cases-up-43-percent-in-2017-county-data-shows (reporting change in office practices based on data set that Cook County prosecutor released voluntarily).

146. See Overview, Measures for Justice, (last visited May 18, 2018); Amy Ellis, MacArthur Foundation Awards FIU $1.7 Million to Study Prosecutor Behavior, FIU News (Mar. 9, 2018, 10:26 AM),

147. See Ronald F. Wright, The Wickersham Commission and Local Control of Criminal Prosecution, 96 Marq. L. Rev. 1199, 1200 (2013). But cf. William J. Stuntz, Unequal Justice, 121 Harv. L. Rev. 1969, 1973 (2008) (describing decline of local influence in last half of twentieth century).

148. Careful disclosure policies can protect the legitimate privacy interests of jurors without requiring case-by-case judicial approval of jury selection information. See Grosso & O’Brien, supra note 37, at 667–68; Nancy J. King, Nameless Justice: The Case for the Routine Use of Anonymous Juries in Criminal Trials, 49 Vand. L. Rev. 123, 152 (1996).

The full text of this Article is available to download as a PDF.