Printer Friendly


Scholarship about racial disparities in jury selection is extensive, but the data about how parties examine potential jurors in actual trials is limited. This study of jury selection for 792 potential jurors across twelve randomly selected North Carolina capital cases uses conversation analysis to examine the process that produces decisions about who serves on juries. To examine how race influences conversations in voir dire, we adapted the Roter Interaction Analysis System, a widely used framework for understanding the dynamics of patient-clinician communication during clinical encounters, to the legal setting for the first time. This method allows us to document the conversational dynamics of actual questioning of potential jurors that precedes the decision to seat or strike a juror, or to excuse her for cause. Our preliminary analysis of this uniquely rich archival data suggests ways in which the discourse of jury selection varies by race, and provides the foundation for future work looking at the ways in which the evaluation of fitness for jury service itself is skewed and contributes to racial disparities in jury selection.

INTRODUCTION                                         689
 A. Discrimination in Jury Selection                 692
 B. Stereotypes and Racial Bias                      695
II. CONVERSATION ANALYSIS                            698
III. THE CURRENT STUDy                               701
 A. Sample Design                                    701
  1. Death Penalty Reservations Subset               702
 2. Prior Criminal Accusation                        703
 3. Hardship Arising from Jury Service               704
 4. Employment with Police or Prosecutors            704
 B. Conversation Coding                              705
  1. Adapting the RIAS Rubric to Legal Analysis      706
  2. Coding & Reliability                            706
  3. Global Affect Rating                            708
IV. A OVERVIEW OF THE DATA                           708
 A. The Study Sample                                 708
 B. Capacities of the Coding Scheme                  711
  1. Qualities of Speech                             712
  2. Moving Beyond Individual Variables to Look at
     Patterns or Systemic Variation                  713
  3. Working with Single Cases, In Depth             717
V. DISCUSSION & CONCLUSION                           724
APPENDIX 1. DIRECTORY OF TEXT CODES                  726


Table 1. Study Sample by Target Group and Case
Table 2. Study Sample by Jury Selection Outcome and Case
Table 3. Study Sample by Rate of Strike and Cause Removals by Race
Table 4. Study Sample by Average Number of Utterances and Amount of
Table 5. Presence of Open and Closed Question Codes by Voir Dire
Section across All Cases
Table 6. Prosecutor/Juror, Defense/Juror, and Judge/Juror Speech
Ratios by Case, Overall (col. B) and by Race (cols. C-F)
Figure 1. Conversation Profile by Race, the Case of John Badgett


While scholarship about jury selection is extensive, the data about how parties examine potential jurors in actual trials is limited. (1) This study goes behind the outcomes of peremptory strike decisions documented in our earlier work to examine the process that produces those decisions. (2) We use conversation analysis to document the questioning of potential jurors that precedes the decision to seat or strike a juror, or to excuse her for cause. Our preliminary analysis of uniquely rich archival data suggests ways in which the discourse of jury selection varies by race, and begins to suggest that the evaluation of fitness for jury service itself is skewed and contributes to racial disparities in jury selection. This article provides the foundation for future work in that promising vein.

Jury selection unfolds in stages. The judge and sometimes the attorneys first question potential jurors to establish their ability to be fair. Any juror who cannot be fair will be excused. Jurors next face peremptory challenges by both parties. Parties may dismiss potential jurors with a peremptory challenge for any reason or no reason at all, except race or gender. In this way, the parties feel confident that the system is fair. (3)

The process has, however, been plagued by racism. In Batson v. Kentucky, the Court issued a clear constitutional prohibition against consideration of race in strike decisions, holding that purposefully excluding people from jury service based on race was unconstitutional and undermined public confidence in the justice system. (4) The Batson Court built on a significant line of cases seeking to make juries more inclusive. (5) Nevertheless, strong evidence suggests that improper factors continue to play a role in jury selection. While the Supreme Court established an elaborate three-step process for challenging a strike as based on race, parties can readily defeat the challenge in the third step by proffering a plausible race-neutral reason for the strike decision. (7) Trial courts rarely reject these reasons as disingenuous or "pretextual." (8)

The Batson regime suffers from a major design flaw as it was intended to counter intentional discrimination. (9) Accordingly, attorneys strongly deny any suggestion that they engage in intentional discrimination. (10) This may or may not be true. Substantial social-psychological evidence supports the possibility that people--including prosecutors, defense counsel, and judges--harbor stereotypes about race that bear on people's attractiveness as jurors. (11) Evidence also suggests that both prosecutors and defense counsel use race as a proxy for bias, despite the constitutional prohibition. (12) In these instances, voir dire may serve as a tool to develop race-neutral justifications for the anticipated race-based strikes.

Significant psychological research suggests that racial bias also operates below the level of conscious awareness to affect people's perceptions and behaviors. (13) In these instances, stereotypes about which demographic groups are more or less likely to convict and ultimately sentence a defendant to death operate as an implicit starting hypothesis that informs how they collect information during the voir dire process. (14) The information collected then reinforces the stereotype and increases the likelihood that racial stereotypes influence strike decisions. (15)

Either way, the voir dire process might contribute to the improper influence of race. This research project seeks--in time--to document that contribution as part of a larger effort to limit the influence of race on jury selection. We coded the process of jury selection for 792 potential jurors across twelve randomly selected North Carolina cases in which the defendant received the death penalty. (16) We coded these conversations according to a rigorously tested method of conversation analysis in which every complete thought expressed receives a discrete pre-defined code. (17)

In this article, we explain how we adapted this methodology to jury selection, and then present an overview of the data. In Part I, we review the evidence that despite the Supreme Court's clear prohibition, race continues to matter in jury selection and explain why this may be so in light of the psychological processes at work. Part II reviews the history of conversation analysis. Part III explains the theory and details of the purposive sample frame. It then explains how we adapted the conversation coding methodology to this project and provides coding details. In Part IV, we present our sample and offer a closer look at ways in which the coding scheme can be used in analysis. In Part V, we discuss the implications of these patterns and future avenues of research.


Despite the Supreme Court's efforts in Batson to curb racial bias in the use of peremptory strikes, stark racial disparities persist. (18) While racism continues to influence jury selection, (19) the disparities also arise from other more-subtle psychological processes. (20)


Jury selection involves two distinct reviews of potential jurors. (21) First, a trial judge should remove any potential juror for "cause" if there is evidence that the juror cannot be impartial and follow the judge's instructions. (22) There is no limit to the number of jurors who may be removed for cause, but the basis for doing so must be explicit and fall within specific categories relating to the juror's fitness to serve (e.g., pre-existing opinions about the case or a relationship with one of the parties). (23) Second, each party may peremptorily remove, or "strike," a limited number of potential jurors for any reason other than race or gender, and typically without explanation. (24) While each review serves a well-established purpose of ensuring a fair and unbiased jury, the discretion afforded parties in exercising their peremptory strikes heightens the risk that improper factors such as race will influence decision making. (25)

Although consideration of race in strike decisions is constitutionally prohibited, research in both law reviews and social science journals indicates that race continues to play a role. (26) The difficulty of uncovering racial bias--whether deliberate or unconscious--has led many to conclude that the Batson regime cannot counter discrimination in jury selection. (27) Many scholars and several judges have called for the wholesale abolition of peremptory strikes. (28) Others have suggested changes to the Batson regime, such as reducing the number of peremptory strikes available to each side, so as to limit the opportunity for discrimination. (29)

While scholarship about peremptory challenges is extensive, data about how parties exercise these challenges in trials is limited. A significant body of experimental work has examined the role of race in mock jury selection. (30) Anecdotal evidence of discrimination also exists. In a 1986 training video, for example, Philadelphia prosecutor Jack McMahon emphasized the importance of striking certain kinds of jurors, such as "blacks from low-income areas" and blacks who are "real educated." (31)

Only a handful of published studies, however, have examined how parties strike jurors in actual trials. Every study of which we are aware found substantial racial disparities in both prosecutorial and defense use of peremptory challenges. (32) In a previous article, we presented evidence of racial disparities in jury selection in 173 North Carolina cases in which the defendant was sentenced to death. (33) We found that prosecutors in North Carolina capital cases between 1990 and 2010 exercised peremptory challenges against black potential jurors at twice the rate as jurors of other races, even after controlling for alternative grounds for removal.

We are not aware of any studies that have examined racial disparities and the process of voir dire. In fact, very few studies have focused on voir dire. (34) Most studies in this area focus on the structure of voir dire, such as the relative benefit of judge versus attorney questioning or of individual versus group questioning of jurors. (35) Cathy Johnson and Craig Haney surveyed potential jurors and observed the process of voir dire in four felony trials. (36) The researchers then coded voir dire transcripts for personal-biographical content and various kinds of instructional communications about the jury system and the role of jurors to evaluate the effectiveness of jury selection by either the prosecution or defense in identifying ideal jurors. (37) Their focus, however, was on the structure of voir dire and its outcomes in light of frequent criticisms that the process is unduly long and misused by attorneys who seek to indoctrinate potential jurors. At least one other study has evaluated how the process of voir dire socializes jurors, (38) but we are unaware of any that examine nuanced differences in questioning pertaining to observed racial disparities.


A body of psychological literature indicates that decision makers tend to seek and interpret information in ways that will support existing hypotheses. (39) In the context of stereotyping, this phenomenon is known as stereotype maintenance. (40) Psychologists have demonstrated the impact of confirmation biases and information seeking behavior in experimental settings. (41) We are not, however, aware of any research that has replicated these findings in either experimental or applied research on jury selection. If such a process is at work when attorneys select jurors, a pre-existing belief about whether a potential juror is likely to favor or disfavor the prosecution or the defense will influence the discourse--namely the line and tone of examination--that takes place during voir dire.

Confirmation bias is the tendency to support a hypothesis by seeking consistent evidence while minimizing inconsistent evidence. (42) This kind of bias can lead to testing a hypothesis in a way that is likely to support it, (43) or to searching for new information in a biased or biasing manner. (44) Confirmation bias may lead police investigating a crime to focus on information consistent with the guilt of a suspect, and to minimize evidence pointing them in a different direction. (45) Confirmation bias is not deliberate, (46) and can be present even when there is no motivation to prefer a particular hypothesis. (47)

Psychologists have also demonstrated that confirmation bias can serve to reinforce stereotypes. Stereotypes are categorizations that allow people to both process information more efficiently and to generate hypotheses about how members of a particular group are likely to think and behave. (48) This efficiency, however, comes at a cost in that it can lead perceivers to relegate their targets to caricatures, and also because these perceptions are often resistant to disconfirming information. (49)

There are several mechanisms at work in maintaining stereotypes. (50) One stereotype maintenance process particularly relevant in the voir dire context involves the solicitation and interpretation of information. (51) People often seek information that confirms rather than disconfirms what they expect based on the stereotype they hold. (52) Moreover, to the extent perceivers receive stereotype-inconsistent information, they often seek to reconcile that information with what they expect to see or otherwise minimize the stereotype-inconsistent information's value. (53) In other words, the stereotypes people hold influence the information people seek and how they interpret new information. (54)

If attorneys start with the hypothesis that black jurors are more likely to have experiences and attitudes that would undermine their willingness to convict and impose the death penalty, and if attorneys act consistently with research in other domains that shows that people often search for evidence that confirms rather than disconfirms an initial hypothesis, then we should observe disparities not only in whom attorneys choose to strike or pass, but also in how they question jurors. This tendency would undermine not only the fairness of the process by contributing to racial disparities, but also its accuracy by skewing the evaluation of fitness for jury service.

Moreover, even people who consciously endorse egalitarian views may exhibit subtle bias in interracial interactions. (55) Psychologists John Dovidio and Samuel Gaertner call this bias "aversive racism," and explain that even people who harbor little or no conscious racial animus still have unconscious biases that manifest in subtle, unintentional differences in how they interact with people of other races, particularly "negative decisions in complex situations in which bias could be attributed to factors other than race." (56)

A complex situation in which a number of factors besides racial bias can justify a decision describes jury selection perfectly. A longstanding criticism of the Batson regime is that the second prong, which requires the striking party to proffer a race-neutral reason for the strike, is too easy to satisfy. (57) If psychological processes like stereotype maintenance and aversive racism are in fact driving the racial disparities in how prosecutors decide to use their peremptory strikes, then we should expect to see subtle differences in the interactions that precede those decisions. We assess those potential differences through conversation analysis.


Conversation analysis describes a group of methodologies that document interpersonal interactions at a fine level of detail. (58) These methodologies focus on how all parties to an interaction speak and participate in an exchange. (59)

Although conversation analysis was initially used to understand informal interactions, researchers soon recognized its value to analyze communication in institutional settings. (60) In particular, researchers interested in the dynamics of doctor-patient interactions sought to adapt the method to that setting. (61) One frequently used method of conversation analysis in clinical settings is the Roter Interaction Analysis System ("RIAS"). (62) Debra Roter developed this system through a meta-analysis of research on patient-doctor conversations. (63) RIAS provides a way to characterize in a highly detailed way the verbal interactions that occur in a conversation. (64) Coding categories are determined in advance and tailored to the specific type of exchange under analysis. The exchange is treated as a process that takes place through specific acts--i.e., phrases, questions, and answers--that are amenable to categorization. (65)

Researchers have used RIAS to analyze racial disparities in patient-doctor communications. For instance, Mary Catherine Beach and colleagues (including Roter) reviewed recorded visits between doctors and patients, and coded the conversations for the presence or absence of various factors (e.g., verbal dominance, socio-emotional communication, question asking, information giving) to analyze differences in interactions by patient race. (66) Beach and colleagues used RIAS to find that healthcare providers were more verbally dominant when interacting with black HIV patients compared to white HIV patients. (67) Using the same method, Beach and colleagues found that healthcare providers engaged in less psychosocial talk with Latino patients treated for HIV relative to their non-Latino counterparts. (68)

The adaptation of this system to a legal setting provides unique opportunities to compare observed racial disparities in analyses of medical conversations with observed disparities in the courtroom. Like doctor-patient interactions, the jury selection process involves a clear hierarchy in which one of the actors (i.e., the doctor, or the judges and attorneys) plays the role of an expert seeking information from the other partner to the exchange (i.e., the patient, or the juror) to make a diagnosis of sorts. (69) While the subject matter and some of the motivations differ, in both instances the party with greater expertise about the context attempts to assess the other through a series of questions and follow-up. Neither the series of questions nor the follow-up is a casual conversation, as both the medical interview and voir dire have a traditional structure and content. (70) In essence, both doctors and legal professionals are testing hypotheses about the patients or jurors to determine the best course of action in light of their goals (how to treat an illness, or whether to strike or pass a juror). In addition, the conversation serves to educate the patients and potential jurors (providing information about a health condition and a treatment plan, or explaining the law they will have to apply as jurors). (71)

Researchers involved in the Arizona Jury Study Project used a similar type of analysis to code the conversations of jurors during their deliberations. (72) Diamond and colleagues coded the conversations that occurred among jurors as they decided verdicts using video recordings of their deliberations in actual trials to test the effect of proposed reforms. Through detailed coding of the content and process of the deliberations, they were able to examine factors such as relative frequency and length of each juror's participation in the conversation, as well as the specific topics discussed.

The jury selection process is likewise amenable to the kind of rigor and thoroughness the researchers involved in the Arizona Jury Study brought to coding the process of jury deliberations. In the next part, we explain how we adapted the RIAS method to examine how prosecutors talk to potential jurors in selecting juries for North Carolina capital cases.


This study analyzes the process of voir dire in twelve capital murder cases in North Carolina. Section III.A explains the purposive sampling scheme and the theoretical reasons for this approach, and provides details of execution. Section III.B documents the coding process.


We selected cases for inclusion in several stages. First, we identified thirty-four cases from the 173 North Carolina capital cases included in our prior study where the court had granted a motion for individual voir dire. (73) In these thirty-four cases, the attorneys and judge questioned each juror separately rather than in a group. Focusing on these cases allowed us to identify clearly the part of the transcript that is relevant to each targeted juror. Second, we identified the county of prosecution for each case. The thirty-four cases with individual voir dire came from twenty-five different counties.

Our next step was to draw a stratified sample of twelve cases from this list that achieved balance in defendant/victim racial combinations, and diversity of both place and trial participants. To achieve this we sorted the cases into the following defendant-victim categories: white defendant/white victim, black defendant/white victim, and black defendant/black victim. (74) We then used SPSS to randomly select four cases from each category. (75) We further stratified so that each case was from a different county, and that none of the attorneys or judges was involved in more than one case.

In prior research, we examined attorneys' strike decisions. (76) In these twelve cases, prosecutors struck 50.7% of black jurors versus 21.2% of all other jurors, a ratio of 2.4. (77) In contrast, defense attorneys struck 27.6% of black jurors versus 56.8% of all other jurors, a ratio of 0.49. (78)

Here, we shift our attention from the ultimate strike decisions to the process of making that decision by examining the conversation leading up to the decision to strike or pass a potential juror. That is, we are interested in the actual conversation between each targeted juror and the judge, prosecutor, and defense counsel trying the case. We therefore selected from the twelve cases in our sample only those potential jurors with particular characteristics most likely to elicit questioning. Specifically, we targeted jurors who had (1) expressed reservations about the death penalty, (2) previously been accused of a crime, (3) indicated that they or a close relative worked for law enforcement or a prosecutor's office, or (4) expressed concern that jury service would impose undue hardship. We targeted these jurors because these four characteristics strongly predicted prosecutorial peremptory strikes in our previous research, and they relate to common stereotypes about how racial groups differ as jurors. (79)

1. Death Penalty Reservations Subset

Prosecutors frequently argue that race disparities in the exercise of peremptory strikes arise because black citizens are more likely to oppose the death penalty. (80) While there is empirical support for that generalization, (81) controlling for that factor did not substantially mitigate the race disparities in the RJA Study. (82)

We found that while the expression of death penalty reservations greatly increased the odds that a prosecutor would strike potential jurors of any race (odds ratio = 11.44), black jurors with that characteristic were still significantly more likely to be struck than a non-black counterpart expressing similar views. (83) Among people who expressed hesitation about the death penalty, the prosecution passed (that is, declined to strike) disproportionately more white potential jurors (26.4% non-black jurors with death penalty reservations passed) than black potential jurors with such reservations (9.7%). (84)

Potential jurors in the cases we analyzed expressed reservations about imposing the death penalty often enough to allow us to examine the dynamics surrounding questioning about this factor in relation to race and outcome. Eleven percent of the strike-eligible venire members in the statewide sample we analyzed expressed some reservation about imposing the death penalty; about one-third of those jurors were black.

2. Prior Criminal Accusation

Another explanation frequently offered for racial disparities is that black potential jurors are more likely to have been accused of a crime. (85) Those with such experience might be less trusting of police and prosecutors, and thus be more skeptical of the state's evidence. (86) Indeed, in the statewide sample, people who had been accused of crimes were more likely to be struck by the prosecution (odds ratio = 2.1), but even after controlling for this factor, a pronounced racial disparity persisted (being black increased the odds of being struck by 2.5). (87) Prosecutors were less inclined to strike non-black jurors with this characteristic (63.8% passed) compared to their black counterparts (36.7% passed). (88)

As with death penalty reservations, this characteristic appears frequently enough in the data to meaningfully examine the dynamics surrounding questioning about this factor in relation to race and outcome. Ten percent of strike-eligible jurors in the statewide sample had this characteristic; just over one quarter of those jurors were black.

3. Hardship Arising from Jury Service

A potential juror who expressed concern that jury service would pose an undue burden was significantly more likely to be struck by the state (odds ratio = 3.0). (89) This characteristic is also proffered to account for racial disparities (that is, that due to pre-existing socio-economic inequalities, black potential jurors were more likely to face hardship through jury service). But again, race remained a significant predictor of prosecutorial strikes even when controlling for this factor. Within this category of people sharing this characteristic the prosecution passed disproportionally more non-black potential jurors (26.4%) than black potential jurors (9.7%). Of jurors in the statewide sample, 4.4% had this characteristic, of whom about 15.7% were black.

4. Employment with Police or Prosecutors

The first three characteristics discussed increased the odds of a state strike, but examining the questioning of jurors with characteristics that decrease those odds may also prove valuable. One such characteristic is employment with police or a prosecutor's office. Prosecutors were much less likely to strike a juror who personally had worked with police or prosecutors, or who had a close friend or family member who had done so (odds ratio = .63). In the statewide sample, 18% of jurors had this characteristic; 11% of those jurors were black.

All four characteristics discussed above--death penalty reservations, prior criminal accusation, hardship, and employment with police or prosecutors--are not only strong predictors of prosecutorial strike decisions, but also appear important to defense counsel. Although we have yet to construct a fully-controlled model of defense strikes, preliminary analysis suggests that many of the same factors are at work, but (not surprisingly) in the opposite direction. For instance, one preliminary model shows that defense counsel were significantly less likely to strike a juror who expressed ambivalence about the death penalty (odds ratio = .43). The same pattern emerged for jurors who have personally been accused of a crime, or who had a close friend or family member so accused (odds ratio = .62). In contrast, defense counsel are significantly more likely to strike a juror who had worked with the police or prosecutors (odds ratio = 4.37). Hardship is the only characteristic of the four that increases the odds of both defense and state strikes (odds ratio = 1.6 (defense), 3.0 (state)). Thus, given the importance of these factors to strike decisions, conversations with jurors who have these characteristics may provide a window into the decision-making processes of the parties.


This study focuses on the substance of voir dire itself in order to document disparities in the questioning process rather than solely in the particular content of questions and answers. Our unit of analysis is an utterance--the smallest part of expression that can be meaningfully coded. (90) Utterances are coded for both form and substance. For example, as illustrated below and in Appendix 1, we distinguish closed from open-ended questions. Coding categories are both exhaustive and mutually exclusive, which allow them to function either on their own or as components of other variables. Separate codes document personal and historical information, as well as length of voir dire, verbal dominance (who speaks most), and the extent of individual expression. (91) Codes are informed both by prior voir dire analysis and experimental work on stereotypes and information seeking, as well as literature on conversation analysis. (92)

1. Adapting the RIAS Rubric to Legal Analysis

The first task involved adapting the RIAS coding rubric to the legal context. The adaptation was often straightforward; in both the medical and legal contexts, the parties seek and give information and instructions, make small talk, and express concerns. For instance, a doctor explaining the next steps in the patient's treatment plan is analogous to a judge explaining how the trial will unfold.

There are, however, obvious differences between the two settings. The lawyers are working as adversaries and are bound by court rules that govern both what they say and how they say it. We therefore worked closely with Debra Roter and Susan Larson to determine which medical RIAS codes could be applied to voir dire conversations without modification, which were applicable with some modification, and which were simply too specific to the medical setting to be useful. (93) Along those lines, we created codes to capture aspects of conversation unique to the legal setting, such as arguments about Batson objections, as well as aspects of the conversation that have special significance in voir dire, such as assertions of fact versus opinions.

2. Coding & Reliability

The adapted RIAS coding scheme included forty-six mutually exclusive codes for utterances. (94) To code the voir dire transcripts using this system, we hired law student graduate research assistants with an eighteen month commitment and invested in extensive training in RIAS coding. (95) The coders, blind to our hypothesis, (96) were assigned individual target jurors to code. Coders began by reviewing an entire voir dire transcript to evaluate every conversation in which a target juror appears. Conversations with the target juror were easily identified and separated from other conversations because these cases all involved individual voir dire. Once the coder identified every conversation, the coder coded every part of the transcript relevant to this juror.

Before undertaking coding, every coder completed a multi-day training, a day-and-a-half of which was taught by an expert in RIAS. We required that all coding be done by two independent coders until each coder had achieved reliability and consistency, and held weekly oral reviews to correct mistakes and to review difficult transcript sections. We also maintained the RIAS Legal Codebook containing a detailed explanation of each code and examples of its proper application. Coders had to review this codebook weekly and to make precise reference to it when raising questions with the project manager. Coders worked on site under the close supervision of the project manager.

As coding improved, we developed a systematic "spot check" on coding accuracy. We asked each current member of the team to code several pages of transcript in preparation for the weekly coding meeting. We ran "spot check" exercises approximately six times over three months. We then reviewed the transcript pages together at the meeting, correcting coding, and providing guidance to improve coding precision.

Finally, we randomly selected twelve juror transcripts for double coding. This is about 1.5% of the jurors in the study (12/792). This review analyzed 9,299 lines of code pertaining to approximately 5,870 coding decisions. We found substantial agreement, with a kappa of .78.

3. Global Affect Rating

Coding of utterances is a fine, micro-level look at the interactions among the parties in voir dire. To assess the more general tenor of the exchanges, coders also rated the tone of the exchanges more globally. After each section of the transcript (that is, when the court handed off the questioning to the prosecutor, and then to the defense), coders completed a Global Affect Ratings Code Sheet to note their perceptions of various affective dimensions of the exchange, such as the parties' warmth, frustration, nervousness, and level of engagement. In contrast to the relatively strict rubric of the RIAS coding explained above, the global affect reports are necessarily subjective. They capture the coders' opinions about the tone of the conversation. Moreover, the use of transcripts instead of videotaped exchanges naturally limits the ability of the coders to gauge the tone of the exchange. Nevertheless, transcribed conversations often convey feelings of irritation, friendliness, and boredom, among others, through cues such as clipped answers, interruptions, and phrasing.


Section IV.A describes the study sample and compares it to the universe of jurors from which we drew the sample. Section IV.B introduces some dimensions of the database by demonstrating ways the data can be used to understand the voir dire process.


Our study includes 792 potential jurors across twelve cases. As noted in Part III.A above, we did not include the whole venire for these cases. Rather, we drew a purposive sample in which all the potential jurors in our study fell into at least one of four target groups: (1) potential jurors with death penalty reservations (46.0%, 365/792), (2) those who had been accused of a crime (8.3%, 66/792), (3) those who had connections to police or a prosecutor's office (18.6%, 147/792), and (4) those who expressed concern that jury service would impose a hardship (42.9%, 340/792). (97) Table 1, Column B shows the total venire of the twelve cases, and Columns C-F show the number of potential jurors in each relevant category that constituted our sample.

Our prior research focused on strike decisions. (98) In this study, we expand the analysis to include removals for cause, which is how most potential jurors in these cases exited the path to serving on a jury, as shown in Table 2, Column D. Table 2 shows the outcome (seated, struck, or excused for cause) for the 792 potential jurors included in the study by each of the twelve cases.

Of these potential jurors, 63.5% were white (503/792), 28.5% were black (226/792), 1.3% were Native American (10/792), 0.3% were Asian (2/792), 0.1% were Latino (1/792). We are missing race information for 6.3% (50/792). Women made up 58.3% of the potential jurors (462/792), and men were 41.7% (330/792). Table 3 shows the outcome for the potential jurors in the study by race; the racially-disparate strike patterns roughly mirror the patterns for the full population of venire members in these twelve cases.

We coded every utterance exchanged by judges, prosecutors, defense counsel, and other court officials with the 792 potential jurors for a total of 193,869 utterances. As noted above, we selected these jurors for inclusion because their particular characteristics were most likely to elicit focused questioning. Every utterance captures a complete thought, but can vary greatly in length from a one-word response to multiple lines of text expressing a single thought. Table 4 therefore sets forth not only the average number of utterances by each potential juror per case (Column C), but also the average number of characters of text for each (Column D), which provides a sense of the amount of talk.

This study was based on a purposive sample. Sample details provide critical foundation for future work. Section IV.A presented details about the study sample in five domains. First it presented the purposive distribution of jurors among cases and target groups. It then documented jury selection outcomes and strike and cause removals rates by race and compared the sample to the universe in this respect. Finally, it documented the scope of conversation codes and variance in rate of conversation by case. The next section shifts attention to the coding scheme.


While the previous section presented basic information about the study sample, this section provides details about the capacities of coding scheme. The first subsection presents analysis using qualities of speech, such as whether a question is open or closed. Next we consider ways of evaluating speech in the aggregate to compare the amount different speakers speak or the ratio between subsets of speakers. This can help evaluate dominance or variance in treatment. The final subsection demonstrates how conversation profiles can be used to identify cases for qualitative work. Previous research on jury selection and charging and sentencing decisions in these cases provide rich data to complement these analyses. (101)

1. Qualities of Speech

Coding tags qualities of speech as well as content. Table 5 provides a simple example of this. The table starts in the least exciting part of the coding scheme. Basic compound codes mark interactions with very little emotional content. Basic codes note whether questions are open or closed, whether they seek or give fact or opinion, and whether they focus on general legal matters, case specific matters, personal legal matters, or personal matters.

More than 60% of the utterances in the databases are marked with basic compound codes. For example, if a prosecutor asks a juror, "Do you go to church?" This would be coded as a "Closed Personal Question." The response, "Yes," would be coded as "Gives Personal Information." Slightly different codes focus on the content of speech--general legal, case specific, personal, or personal legal--in these compound codes. A second set of codes capture affect more precisely. Imagine the prosecutor prefaced the previous question with some apologetic language, saying instead, "Please don't think I am prying into your personal life, but do you go to church?" The first phrase would be coded separately, noting the reassuring tone, "Reassures/Optimism" and the second as a Closed Personal Question.

Table 5 presents basic counts of utterances by only one aspect of general compound code--the fact of asking an open or closed question. It disaggregates this information by voir dire section. Coders identified sections as they coded; sections follow a pattern. Typically individualized voir dire begins with the judge explaining the process and the case to a potential juror. This is the court section. The prosecution follows and holds the floor while he questions the juror. Finally, the defense counsel takes over voir dire if the prosecutor is satisfied and passes the juror. (102)

You can see in Column 3 that generally speaking, most general code question utterances were made during the prosecutor's part of voir dire. This section accounts for 61% of the utterances (23,578/38,854). Defense counsel, on the other hand, goes last and speaks less. Only 22% of the utterances were made by defense counsel (6,773/38,854). This pattern holds across open and closed questions, shown in Columns B and C.

Comparing Columns B and C shows the dominance of closed questions in voir dire. Ninety-four percent of these questions are closed, a pattern that holds throughout voir dire. The overwhelming dominance of closed questions in a context in which the speaker intends to learn about qualified jurors raises general questions about the utility of voir dire. Future research may evaluate the evidence presented here about how voir dire actually unfolded in these cases to test the use of best practices in this kind of questioning.

More specific codes document behavior with a higher level of emotional content. For example, coders watch for a special kind of narrowly defined leading questions. These codes mark information where the speaker introduces new information to the juror that would either increase or decrease the juror's likelihood of qualifying for the jury. Leading questions often address the venire members' opinions about the death penalty. Other codes document each time a speaker asks for reassurance or ask for understanding. Analysis of patterns of speech around expressions of empathy or situations in which an attorney discloses something about himself to the juror ("I play golf there too") may tag areas that merit close analysis of the ways stereotypes influence voir dire speech and limit its value.

2. Moving Beyond Individual Variables to Look at Patterns or Systemic Variation

The coding system also tracked the amount and order of speech to allow us to look beyond individual codes at patterns or systems. Previous scholars have used this capacity to document the amount of talk by speaker or the ratio of one kind of speech to another. (103) In subsequent work we could also use this capacity to analyze the polarity or subjectivity of the language in an effort to capture emotive content in the highly ritualized context of voir dire. (104) Our data allows us to analyze the amount of talk by each speaker, the sequences of speech within and between speakers and respondents, and the nature of the speech.

Calculations of the ratio of talk between medical providers and patients have been used as an indication of verbal dominance. (105) We adapted that measure here to evaluate the relationship between different speakers and jurors. Table 6 presents this information for each of the twelve cases in our study. Each section shows the ratio of the amount of prosecutor, defense attorney, and judge talk to juror talk for black and other jurors. These calculations are based on a sum of the number of characters used in the voir transcript during every utterance by each speaker. We also disaggregated the data and ratios by gender, but did not see meaningful differences in this level of analysis. Subsequent work may look more closely at the intersection between race and gender with respect to verbal dominance.

These ratios presented in Table 6 suggest that the extent of verbal dominance varies by race (black or other) and actor (judge, prosecutor, or defense counsel). Looking first at prosecutor speech, the average ratio between prosecutor speech and juror speech (prosecutor/juror) is 0.92. This number averages the ratios for prosecutor/juror speech for all jurors in Column B. Columns C and D present the ratio in amount of speech disaggregated by race. Column C presents speech ratios for black jurors and Column D presents ratios for all other jurors. Columns E and F compare the information in Columns D and E. Column E reports the arithmetic difference between them (Column C-Column D). Column F presents the ratio between the columns (Column C/Column D). These measures provide two estimates of the magnitude of any disparity observed.

Looking first at Column C, in eight of twelve cases the prosecutor dominates the juror by between 0.20 and 1.5 points more when the juror is black than when the juror is any other race. The ratio between them varies from 1.27 and 3.14 across these eight cases. In two cases, the ratio does not differ meaningfully by race. In the final two, the ratio flips in the other direction. These two cases are the oldest cases in the study and both involve black defendants and white victims.

Turning to defense counsel speech, it is noteworthy that the defense follows the prosecution in capital voir dire in North Carolina and, therefore, is unlikely to have spoken with any juror that faced a state strike. (106) As a result, defense counsel has less speech overall in the dataset. The average ratio between defense counsel speech and juror speech (defense/juror) is 0.46. Previous research comparing prosecutor and defense decision-making in jury selection found that defense counsel struck a higher percent of non-black jurors than black jurors and, in fact, seemed to operate as the mirror opposite of prosecutors. (107) The mirroring effect is present in ratios in Table 6. The ratio of defense to juror speech with jurors of other races exceeds the ratio for black jurors in ten of twelve cases. The difference in ratios ranges from -0.12 to -0.80 (Column E). The ratio goes from infinity (0.57/0) to .06. The ratios in two cases move in the opposite direction (319 and 330). One of these is a case in which the prosecutor also switched directions.

Table 6 also reports ratios with respect to judge speech. The average ratio between judicial speech and juror speech is 0.64. This analysis excludes any sections where the judge is speaking to a large group of jurors collectively. The judge speech ratio is higher in voir with black jurors in eight of twelve cases. It flips in the other direction for two and is equal in two.

The disparities reported here provide a first look at the way the speakers interacted with each other with respect to holding the floor. Future work will allow us to look more carefully at how these disparities appear in different target groups or in the context of different parts of voir dire conversations.

3. Working with Single Cases, In Depth

This research ultimately seeks to document ways that the voir dire process itself might contribute to the improper influence of race. While the data is poised for analysis, the next phase requires identifying patterns of codes that would help to identify areas where information seeking behavior may present. We need to hypothesize about what the codes might look like in these situations. Precisely, we need queries that we can operationalize and test.

The RIAS coding system lends itself to this part of the process well. By coding utterances individually, that is coding every complete thought separately, and by making the codes "mutually exclusive and exhaustive," the codes work well as "building blocks" that we can combine "to summarize the dialogue" with respect to content, tone, or even sequence. (109)

This subsection presents one example of this kind of work. The analysis here relied on conversation profiles that we built by grouping individual codes around function and, to some extent, tone. The profile has four mutually exclusive components. Use of similar categories in previous RIAS coded studies provided guidance in defining meaningful components. (110)

Education/orientation speech, the first, includes utterances that provide facts and information about the law and the case at hand. Data gathering speech, the second, includes questions about facts and opinions relevant to the proceedings. Relationship building speech, the third, involves utterances in which the speaker reaches out to the juror with expressions such as empathy or approval, self-disclosure, or reassurances. Finally, the fourth, conflict speech arises when there is disagreement, criticism, or leading questions that seem designed to control the juror's response.

We assigned each individual code to a single component by polling our coders. Our coders had read and coded hundreds of pages of transcripts before we asked them to do this project. They had developed nuanced understandings of the codes and the transcripts. Coders assigned each code to one of the four pre-defined mutually exclusive categories. We then combined their results into a "conversation profile" with four distinct components: education/orientation speech, data gathering speech, relationship building speech, and conflict speech. This profile can be refined to reflect different speakers and contexts. (111)

The profile describes what percent of the speech in each case falls in each component. For example, in the case of John Badgett (14.0), 53% of speech was spent on education and orientation, 26% on data gathering, 11% on relationship building, and 12% on conflict. Contrast this with the overall conversation profile in the case of Lesley E. Warren (319.0) where 63% of the speech was spent on education and orientation, 28% on data gathering, only 4% on relationship building, and 5% on conflict. Comparing different profiles suggests variance the tone of voir dire and perhaps the strategy or technique of the attorneys.

We also analyzed communication profiles disaggregated by strike decision (strike versus pass) or by race (black versus white). Figure 1 presents the voir dire conversation profile in John Badgett's case disaggregated by race. Each component of the conversation profile appears in a set of bar graphs. The dark grey bar on the left represents black venire members. The light grey bar on the right represents all other venire members. Black venire members in this case face less education and orientation, but more data gathering. They participate in fewer utterances that build relationships and more that reflect conflict.

Similar patterns appear in some of the other cases as well. Without more, however, it is difficult to know how to interpret these differences. The next step is to analyze the underlying codes to see if disparate patterns of questioning underlie the disparate levels of conflict or relationship building profiles. Complementary research on jury selection and charging and sentencing decisions in each case in the study and in North Carolina during the time of the study provide information that can shed light on the voir dire process and suggest avenues for inquiries within RIAS codes. (112)

                       Black  All Other

Education/Orientation  50%    53%
Data Gathering         26%    24%
Relationship Building   9%    11%
Conflict Conflict      15%    12%

For example, we can look at Badgett's case from multiple angles. We can consider the facts in the case, charging and sentencing decisions, and jury selection. Badgett is a white man; he killed Grover Arthur Kizer, a white man. (113) Badgett had a history of substance abuse and violence. This was not his first homicide, or even his first stabbing. He stabbed a man in the throat before committing this murder and was convicted of voluntary manslaughter. Some evidence suggests he suffered from psychological illnesses. At the time of this crime, Badgett was homeless. Kizer, who also suffered from a mental disability, took him into his house on a particularly cold night. When the victim became delusional and agitated, Badgett became frightened and stabbed the victim in the throat with a pocket knife and slashed his arm. Badgett later stole items and cash from the victim and his home on at least two occasions before the police began investigating.

The jury found the case death-eligible because of his prior record and evidence that he committed the murder for pecuniary gain. (114) The case file does not suggest the crime garnered particular attention in the press. (115) Earlier research concluded that cases with at least one white victim were more likely to result in a death sentence than all other cases, even after controlling for other theoretically and statistically important information about the circumstances of the crime and the defendant. (116) Other than that legally impermissible factor, this case does not stand out as particularly aggravated among other death-eligible homicides. (117)

Despite potentially strong mitigation, the conversations during jury selection in Badgett's case largely focused on issues that are fairly standard in voir dire--the potential jurors' familiarity with the case and potential witnesses. Because this was a capital case, the court and lawyers spend a substantial amount of time on the potential jurors' attitudes about the death penalty. However, besides a few questions about the potential jurors' feelings and prior experiences with mental health professionals, the conversation focused primarily on general attitudes and the potential jurors' backgrounds and occupations. Defense counsel's voir dire strategy foreshadowed their trial strategy, as they did not present any mitigating evidence to the jury at trial. This seems out of step with the kind of defense representation that would have been standard practice in 2004 when this trial took place. (118)

Another issue not raised in voir dire was race, but race is nevertheless an important backdrop to understanding this case. Badgett's case was tried in Randolph County, North Carolina. Randolph County was almost 6% black at the time of Badgett's trial. (119) The representation of African-Americans was lower on Badgett's venire, including only three African Americans among 67 people, 4%. Otherwise, the venire was 94% white (63/67), and 1% other (1/67). (120) Not one of the black venire members in Badgett's case was excused for cause. The prosecutor, however, struck two of the three during jury selection (68%). He struck an even higher number of qualified white venire members (9), but a lower percent of this group's total, 26% (9/35), and did not strike the venire member whose race was "other." The seated jury including alternates was 86% white (12/14), including one African-American. (121)

The conversation profiles in Badgett's case are based on the actual conversations in voir dire that led to this result but they are limited in that they offer the broadest possible view of the data. Looking more closely at information we have about the case and individual jurors in the case may provide a focus with the RIAS coded conversations. For example, our previous study documented that the each of the black jurors--the same jurors struck by the prosecution and included in this study--had characteristics of concern to prosecutors. (122)

One juror expressed reservations about the death penalty. Not enough concern to be excused for cause, but enough that her transcript includes questions about her opinions about the death penalty. We suspect from earlier work in these transcripts that questioning about death penalty opinions may be an area requiring particular attention.

Compare, for example, the following two transcript excerpts. They involve the same prosecutor, but different potential jurors, concerning their views about the death penalty. The first except comes from voir dire of a white juror. (123)
Prosecutor: Do you have feelings about either [life imprisonment
without parole or the death penalty]?

White Juror: I'm kind of against the death penalty, but it's in our

Prosecutor: Can you tell me a little about what makes you feel that

White Juror: I don't know. Just--I mean to kill somebody seems wrong.
But if somebody killed somebody... they could get the death penalty
if that's what the finding was.

Prosecutor: If I'm hearing you correctly, it sounds like you're
opposed to killing, period. And that if you were a jury member, even
though you were generally opposed to killing, you could consider
both possible punishments?

White Juror: Yes, sir.

Prosecutor: And you could vote for the death penalty even though
you're not really in favor of it, if that's the law.... You could
still vote for it?

White Juror: Yes, sir.

The prosecutor passed this juror, but struck the black juror whose exchange with the prosecutor appears in the next transcript excerpt. (124)
Prosecutor: Could you tell me what your thoughts or feelings are about
both of those punishments, life imprisonment without parole and the
death penalty?

Black Juror: I believe if you commit a crime, you deserve your
punishment. And if it's the death penalty, I think you should get it.
I really--I don't feel it's right, but if it's the law, it's the
law. That's how I feel.

Prosecutor: So personal preferences, as opposed to what the law is,
you don't feel that the death penalty is an appropriate punishment?

Black Juror: um, I...

Prosecutor: Or I shouldn't say that it's "appropriate." You just don't
think it's right to have the death penalty?... From your personal
point of view, regardless of what the law... is, do you think that
you have some personal reservation about the death penalty?

Black Juror: No.

Prosecutor: If you had your druthers, would you rather not sit on a
case where the death penalty might be an appropriate punishment or
might be a punishment that you had to consider?

Black Juror: It wouldn't matter.

The prosecutor is faced with very similar assertions by jurors in the very same case: they personally do not favor the death penalty, but believe that the law must be followed regardless of their personal beliefs. The prosecutor follows up these assertions, however, with quite different tactics. In the first instance, the prosecutor engages in a process known as "rehabilitation" of the juror, providing the juror through leading questions with an answer that reconciles the juror's discomfort with capital punishment with the prospect of imposing it. The prosecutor focuses on the juror's stated respect for the law as it stands. In response to the second juror, however, the prosecutor focuses on the juror's personal beliefs and suggests that she would prefer not to serve.

This work with conversation profiles focuses our attention again on those parts of these transcripts. A next step might be to look for the codes included in the "conflict" component of the conversation profile within the section of the transcripts addressing death penalty reservations to see if patterns emerge. The codes capture expressions of concern, criticism, disapproval or disagreement. They include leading questions. The transcript sections excerpted above include utterances that would be coded within this category.

The second juror reported that he knew the defendant. Unlike death penalty reservations, we did not target jurors for inclusion in this study based on knowledge of trial participants. We would not have known to look for "conflict" codes in these parts of the conversation. This too may be fruitful in understanding how trial participants seek information and how the behavior varies by race.

It might also be fruitful to compare the conversation profiles in cases with a larger proportion of black venire members to those with a smaller proportion. Profiles might also vary between especially aggravated cases and those, like Badgett's, that are less aggravated. Each of the subprojects presented here are intended to demonstrate the capacity of the coded text to address the question that underlies the project as a whole.


This research makes two significant contributions to the literature. First, it adapts a rigorous and exacting method of conversation analysis to the legal setting. This method provides the tools to subject the elusive process of jury selection to systematic review. Second, by using this method, this research then provides a first look at the actual questioning of potential jurors that precedes the decision to seat or strike a juror, or to excuse her for cause.

We undertook this project in light of both clear evidence of racial disparities in jury selection in the underlying data and prosecutors' earnest claims that race does not enter their minds when picking a jury. Our intensive work with the jury selection transcripts brought subtle instances of disparate questioning to our attention; these instances in turn raised a question about the role of subtle psychological processes in exacerbating and perpetuating race effects.

Consider the transcript sections above--did the prosecutor engage in this disparate behavior knowingly? Did the judge and defense counsel hear this difference and not intervene? We suspect not. This shift seems exceedingly subtle, especially when compared to the differential questioning the Court addressed in Miller-El, which may serve as a kind of public benchmark. (125) Comparing transcript excerpts like those raised our concern that more subtle disparities in the process of questioning jurors may contribute to the disparate exercise of the kinds of peremptory strikes documented in our earlier research. Moreover, the differences in how prosecutors talk to jurors of different races may be more subtle, but also more pervasive, than the example we present above, and a better understanding of disparate behavior may strengthen judicial oversight.

RIAS coding provided a method for systematic analyses of discourse. RIAS characterizes precisely the verbal interactions that occur in a conversation and the context in which they occur. (126) RIAS, however, was developed to capture nuances of medical conversations rather than those examining jurors for criminal trials. (127) Traditional RIAS coding categories had been tailored to analyze exchanges that involved different objectives and motivations than those of jury selection.

We revised the codes to reflect the content and context of legal proceedings. This methodology provides a systematic look at the actual process of jury selection. This paper presents the project as a whole. We expect to present additional findings in the future. In the meantime, we expect that the adapted RIAS codes may provide useful to legal researchers.

The RIAS methodology has been tested and refined for decades in the public health context; (128) ours is the first study that adapts it to the legal setting. The new methods open new avenues for research, but also suggest caution. The data presented here is a first step in untangling the nuances of conversations leading to the decision to strike, pass, or excuse a juror for cause, and to provide an important foundation for future research.


(*) O'Brien and Grosso are Associate Professors of Law at Michigan State University College of Law. Taylor is the Assistant Dean for Student Engagement. This paper is based upon work supported by the National Science Foundation Law and Social Science Foundation under Grant No. 1318723. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. We thank our coders: Erica Walle, Robert Wright, Paige Szymanski, Alexa Herlosky, Caitlin Vanderkarr, Courtney Hawkins, Jennifer Shaltry, Charles Rogers, and Rebecca Stanislaw. Much of this work would not have been possible without the support of our colleagues at Michigan State University generally and, in particular, Barbara Bean, Amanda Olivier, Scott Nagele, Brian Kalt, and Rhonda Bleisner. We also appreciate the helpful feedback we received on this project from many others including Mona Lynch, Richard Lucas, Neil Vidmar, Mary Rose, Sam Gross, and Phoebe Ellsworth, and at presentations at Chicago-Kent College of Law, Northwestern University Pritzker School of Law, the Law and Society Annual Conference in 2015 and 2016, and the 11th Annual Conference on Empirical Legal Studies.
(Adapted from Roter Interaction Analysis System Codebook.

     Code                    Tag          Definition

 1.  Agreement               Agree        Signs of agreement or
 2.  Approve--Direct         Approve      Compliments directed to
                                          the other person present or
                                          showing of gratitude or
                                          appreciation. Any
                                          expression of approval,
                                          praising, rewarding or
                                          showing respect or
                                          admiration directed to the
 3.  Asks for Permission     ?Perm        Questions that ask for
                                          permission to proceed.
 4.  Asks for Reassurance    ?Reassure    Questions of concern that
                                          convey the need or desire
                                          to be reassured or
                                          encouraged. Voice tone,
                                          intonation and emotional
                                          content may help
                                          distinguish questions that
                                          ask for reassurance from
                                          other questions.
 5.  Asks for                ?Understand  Mechanism by which the
     Understanding                        speaker checks with the
                                          other to see if information
                                          that was just said has been
                                          followed or understood
                                          (i.e., in essence asking,
                                          "Do you understand what
                                          I'm saying?").
 6.  Back-Channel            BC           Indicators of sustained
     Responses                            interest, attentive listening
     (Attorney/Court Only)                or encouragement when
                                          the listener does not hold
                                          or take the speaking floor.
 7.  Bid for Repetition      Bid          Request for repetition of
                                          the other's previous
                                          statement. Bids are used
                                          when words or statements
                                          have not been clearly
                                          heard, and therefore need
                                          repetition, and are often
                                          signs of perceptual
 8.  Checks for              Check        Mechanisms by which the
     Understanding                        speaker re-states
                                          information to check for
                                          accuracy or to confirm a
                                          shared understanding of
                                          the facts being discussed
                                          (i.e., in essence asking,
                                          "Do I understand what you
                                          are saying?" "Do I have it
                                          right?" or "Am I on the
                                          right track?").
 9.  Closed [Case Specific   [?]CS        Asks closed question about
     Legal] Information                   case specific facts &
     Question                             knowledge.
10.  Closed [Case Specific   [?]OCS       Asks closed opinion
     Legal] Opinion                       question about case
     Question                             specific facts &
11.  Closed [General Legal]  [?]GL        Asks closed question about
     Information Question                 general legal facts &
12.  Closed [General Legal]  [?]OGL       Asks closed opinion
     Opinion Question                     question about general
                                          legal facts & knowledge.
13.  Closed [Personal        [?]PL        Asks closed question about
     Legal] Information                   personal legal facts &
     Question                             knowledge.
14.  Closed [Personal        [?]OPL       Asks closed opinion
     Legal] Opinion                       question about personal
     Question                             legal facts & knowledge.
15.  Closed [Personal]       [?]P         Asks closed question about
     Information Question                 personal facts &
16.  Closed [Personal]       [?]OP        Asks closed opinion
     Opinion Question                     question about personal
                                          facts & knowledge.
17.  Compliment (General)    Comp         Compliments directed to
                                          the group in general or to
                                          others not involved in the
                                          exchange showing
                                          gratitude or appreciation.
                                          Any expression of
                                          approval, praise, or
                                          showing respect or
                                          admiration directed to the
                                          group in general.
18.  Concern                 Concern      A statement that indicates
                                          that a condition or event is
                                          serious, worrisome,
                                          distressing or deserving
                                          special attention and is of
                                          particular concern now
                                          during the encounter.
                                          Voice tone, intonation or
                                          verbal content may
                                          disclose that worries,
                                          concerns, events or
                                          uncertainties are of
                                          immediate concern.
                                          Includes statements that
                                          ask for pardon and indicate
                                          concern for the other.
19.  Criticism (General)     Crit         Any indication of
                                          disapproval, complaint,
                                          rejection, coolness, or
                                          disbelief directed toward
                                          the group in general or to
                                          others not involved in the
20.  Disapprove or           Dis          Any indication of
     Disagreement (Direct)                disapproval, criticism,
                                          complaint, rejection,
                                          coolness or disbelief
                                          directed to the other person
                                          present. Statements that
                                          contradict or refute
                                          something said by the
                                          other, or imply
                                          disagreement with or
                                          rejection of the other's
                                          hypotheses, ideas or
21.  Empathy/Legitimizing    Emp          Statements that paraphrase,
     Statements                           interpret, name or
                                          recognize the emotional
                                          state of the other person.
                                          Statements that indicate
                                          that the other's emotional
                                          situation, actions, or
                                          thoughts are
                                          understandable and
22.  Gives [Case Specific    GivesCS      Gives case specific
     Legal] Information                   information.
23.  Gives [Case Specific    GivesOCS     Gives case specific
     Legal] Opinion                       opinion.
24.  Gives [General Legal]   GivesGL      Gives general legal
     Information                          information.
25.  Gives [General Legal]   GivesOGL     Gives general legal
     Opinion                              opinion.
26.  Gives [Personal Legal]  GivesPL      Gives personal legal
     Information                          information.
27.  Gives [Personal Legal]  GivesOPL     Gives personal legal
     Opinion                              opinion.
28.  Gives [Personal]        GivesP       Gives personal
     Information                          information.
29.  Gives [Personal]        GivesOP      Gives personal opinion.
30.  Laughs, Tells Jokes     Laugh        Friendly jokes, trying to
                                          amuse or entertain, light
                                          banter. Laughter and
                                          nervous laughter.
31.  Leading Question--      LNeg         Suggests an answer.
     Negative                             Usually calls for yes/no
                                          response. Adds/assumes
                                          information that the juror
                                          had not already provided,
                                          makes the juror appear less
                                          fit to serve
32.  Leading Question--      LPos         Suggests an answer.
     Positive                             Usually calls for yes/no
                                          response. Adds/assumes
                                          information that the juror
                                          had not already provided,
                                          makes the juror appear
                                          more fit to serve.
33.  Open [Case Specific     ?CS          Asks open question about
     Legal] Information                   case specific facts &
     Question                             knowledge.
34.  Open [Case Specific     ?OCS         Asks open opinion
     Legal] Opinion                       question about case
     Question                             specific facts &
35.  Open [General Legal]    ?GL          Asks open question about
     Information Question                 general legal facts &
36.  Open [General Legal]    ?OGL         Asks open opinion
     Opinion Question                     question about general
                                          legal facts & knowledge.
37.  Open [Personal Legal]   ?PL          Asks open question about
     Information Question                 personal legal facts &
38.  Open [Personal Legal]   ?OPL         Asks open opinion
     Opinion Question                     question about personal
                                          legal facts & knowledge.
39.  Open [Personal]         ?P           Asks open question about
     Information Question                 personal facts &
40.  Open [Personal]         ?OP          Asks open opinion
     Opinion Question                     question about personal
                                          facts & knowledge.
41.  Orient                  Orient       Gives instructions.
                                          Statements that tell the
                                          other person what is about
                                          to happen, what is
                                          expected, or serve to orient
                                          the other to the major
                                          topics of discussion or
42.  Partnership Statements  Partner      Statements that convey
     (Attorney/Court Only)                alliance with the other in
                                          terms of help and support.
43.  Personal Remarks/       Chat         Greetings and goodbyes.
     Social Chit Chat                     Social conversation.
44.  Reassure, Encourages    R/O          A statement that indicates
     or Shows Optimism                    optimism, encouragement,
                                          relief of worry or
                                          reassurance. Includes
                                          statements that show an
                                          awareness of the other's
                                          feelings in a positive
                                          upbeat way, or respond to a
                                          request for reassurance.
45.  Self-Disclosure         SDis         Statements that describe
     Statements                           the speaker's personal
     (Attorney/Court Only)                experiences in areas that
                                          have legal and/or
                                          emotional relevance for the
                                          venire member. Self-disclosure
                                          is the revealing of a
                                          non-public personal
46.  Transition Words        Tran         Sentence fragments that
                                          indicate movement to
                                          another topic or area of
                                          discussion, train of thought
                                          or action. Includes
                                          statements or fragments
                                          that are place-holders, if
                                          the utterance stands alone.


    Component              Variables Included

1.  Data Gathering         Open [Case Specific Legal] Information
                           Open [Case Specific Legal] Opinion Question
                           Open [General Legal] Information Question
                           Open [General Legal] Opinion Question
                           Open [Personal Legal] Information Question
                           Open [Personal Legal] Opinion Question
                           Open [Personal] Information Question
                           Open [Personal] Opinion Question
                           Closed [Case Specific Legal] Information
                           Closed [Case Specific Legal] Opinion
                           Closed [General Legal] Information Question
                           Closed [General Legal] Opinion Question
                           Closed [Personal Legal] Information Question
                           Closed [Personal Legal] Opinion Question
                           Closed [Personal] Information Question
                           Closed [Personal] Opinion Question
                           Checks for Understanding
2.  Education/Outreach     Gives [Case Specific Legal] Information
                           Gives [Case Specific Legal] Opinion
                           Gives [General Legal] Information
                           Gives [General Legal] Opinion
                           Gives [Personal Legal] Information
                           Gives [Personal Legal] Opinion
                           Gives [Personal] Information
                           Gives [Personal] Opinion
3.  Relationship Building  Approve (Direct)
                           Asks for Permission
                           Asks for Reassurance
                           Asks for Understanding
                           Compliment (General)
                           Empathy/Legitimizing Statements
                           Laughs, Tells Jokes
                           Partnership Statements (Attorney/Court Only)
                           Personal Remarks/Social Chit Chat
                           Reassure, Encourages or Shows Optimism
4.  Conflict               Concern
                           Criticism (General)
                           Disapprove or Disagreement (Direct)
                           Leading Negative
                           Leading Positive

(1) A significant body of experimental work has looked at the role of race in mock jury selection. For a review, see Samuel R. Sommers & Michael I. Norton, Race and Jury Selection: Psychological Perspectives on the Peremptory Challenge Debate, 63 AM. PSYCHOL. 527, 527 (2008). Anecdotal evidence of discrimination also exists. In a 1986 training video, for example, Philadelphia prosecutor Jack McMahon emphasized the importance of striking certain kinds of jurors, such as "blacks from low-income areas" and blacks who are "real educated." David C. Baldus, George Woodworth, David Zuckerman & Neil Alan Weiner, The Use of Peremptory Challenges in Capital Murder Trials: A Legal and Empirical Analysis, 3 U. PA. J. CONST. L. 3, 41-43 (2001) (citing Jury Selection with Jack McMahon, 1986 (transcript of DATV Productions video tape, n.d.)) [hereinafter Baldus et al.].

(2) Catherine M. Grosso & Barbara O'Brien, A Stubborn Legacy: The Overwhelming Importance of Race in Jury Selection in 173 Post-Batscm North Carolina Capital Trials, 97 IOWA L. REV. 1531 (2012).

(3) NEIL VIDMAR & VALERIE P. HANS, AMERICAN JURIES: THE VERDICT 89-90, 93-95 (2007) (describing this aspect of jury selection).

(4) 476 U.S. 79 (1986).

(5) See generally Swain v. Alabama, 380 U.S. 202 (1965) (holding that a defendant could raise a claim that the prosecutor had discriminated in jury selection if he or she could show systematic evidence of discrimination in selecting trial juries); Strauder v. West Virginia, 100 U.S. 303 (1880) (holding that the West Virginia statute limiting jury service to whites violated the Constitution).

(6) See, e.g., Baldus et al., supra note 1, at 121-30 (summarizing findings concluding that venire member race was "a major determinant in the use of peremptory challenges by both prosecutors and defense counsel" in Philadelphia County capital murder trials); Grosso & O'Brien, supra note 2, at 1554 (presenting the results of a fully controlled logistical model showing that qualified black jurors face odds 2.48 higher than all other jurors of being struck); Mary R. Rose, The Peremptory Challenge Accused of Race or Gender Discrimination? Some Data from One County, 23 L. & HUM. BEHAV. 695, 697-99 (1999) (finding that race played a significant role in the exercise of peremptory challenges by both prosecutors and defense counsel in one county in North Carolina); Billy M. Turner, Rickie D. Lovell, John C. Young & William F. Denny, Race and Peremptory Challenges during Voir Dire: Do Prosecution and Defense Agree?, 14 J. CRIM. JUST. 61, 63 (1986) (presenting findings showing that race played a significant role in the exercise of peremptory challenges in Calcasieu Parish, Louisiana) [hereinafter Turner et al.]; see also Foster v. Chatman, 136 S. Ct. 1737, 1747-56 (2016) (reviewing the record of discrimination in the selection of Foster's jury).

(7) A party raising a Batson challenge must first establish a prima facie case of intentional discrimination by offering evidence that the prosecutor used peremptory strikes to exclude potential jurors because of their race or gender. If the prima facie case is established, the prosecutor must provide a neutral explanation for the strike decisions at issue. The proffered explanation must be more than a bare assertion of good faith, but it need not be a compelling reason or one that impacts members of all races evenly. The trial court must then determine whether the proffered reason was genuine or merely a pretext for racial discrimination. Snyder v. Louisiana, 552 U.S. 472, 478 (2008); Miller-El v. Dretke, 545 U.S. 231, 278 (2005); Batson, 476 U.S. at 93-99.

(8) See Miller-El, 545 U.S. at 278 (Thomas, J., dissenting); Jeffrey Bellin & Junichi P. Semitsu, Widening Batson's Net to Ensnare More than the Unapologetically Bigoted or Painfully Unimaginative Attorney, 96 CORNELL L. REV. 1075, 1116-30 (2011); Kenneth J. Melilli, Batson in Practice: What We Have Learned About Batson and Peremptory Challenges, 71 NOTRE DAME L. REV. 447, 483-84 (1996).

(9) See Vidmar & Hans, supra note 3, at 97 (describing the procedure outlined by the Supreme Court in Batson and noting that an attorney must assert that she believes the other side has removed a juror on the basis of race to raise a challenge).

(10) See, e.g., Transcript of Closing Argument at 2590-92, State v. Robinson, 368 N.C. 569 (2012) (No. 91 CRS 23143).

(11) Mona Lynch & Craig Haney, Mapping the Racial Bias of the White Male Capital Juror: Jury Composition and the "Empathic Divide," 45 L. & Soc'Y REV. 69, 71-72 (2011); Neil A. Rector, R. Michael Bagby & Robert Nicholson, The Effect of Prejudice and Judicial Ambiguity on Defendant Guilt Ratings, 133 J. SOC. PSYCHOL. 651, 652 (1993); Laura T. Sweeney & Craig Haney, The Influence of Race on Sentencing: A Meta-Analytic Review of Experimental Studies, 10 BEHAV. SCI. & L. 179, 182 (1992).

(12) See e.g., Foster v. Chatman, 136 S. Ct. 1737, 1755 (2016) (discussing evidence that prosecutors targeted black jurors for exclusion and sought to seat as few black jurors as possible); see also Baldus et al., supra note 1, at 41 n. 133.

(13) Patricia G. Devine, Stereotypes and Prejudice: Their Automatic and Controlled Components, 56 J. PERSONALITY& SOC. PSYCHOL. 5, 6 (1989); Susan T. Fiske, Stereotyping, Prejudice, and Discrimination, in THE HANDBOOK OF SOCIAL PSYCHOLOGY 357, 357-411 (Daniel T. Gilbert, Susan T. Fiske & Gardner Lindzey eds., 4th ed. 1998).

(14) See Fiske, supra note 13, at 362; Lucy C. Johnston & C. Neil Macrae, Changing Social Stereotypes: The Case of the Information Seeker, 24 EUR. J. SOC. PSYCHOL. 581 (1994); Mark Snyder & William B. Swann, Hypothesis-Testing Processes in Social Interaction, 36 J. PERSONALITY & SOC. PSYCHOL. 1202, 1202(1978).

(15) See Andrew R. Todd, Adam D. Galinsky & Galen V. Bodenhausen, Perspective Taking Undermines Stereotype Maintenance Processes: Evidence from Social Memory, Behavior Explanation, and Information Solicitation, 30 SOC. COGNITION 94, 96 (2012) [hereinafter Todd et al.].

(16) See infra IV for an explanation of how we selected the sample of cases to analyze.

(17) Public health researchers originally developed this system to analyze communication in healthcare settings. Debra L. Roter & Susan Larson, The Relationship Between Residents' and Attending Physicians' Communication during Primary Care Visits: An Illustrative Use of the Roter Interaction Analysis System, 13 HEALTH COMM. 33, 34 (2001).

(18) 476 U.S. 79 (1986); see also supra note 6 (collecting studies showing the influence of race in jury selection).

(19) See, e.g., Foster v. Chatman, 136 S. Ct. 1737, 1744-45 (2016) (documenting prosecutors' explicit references to potential jurors' race and suitability for jury service); Baldus et al., supra note 1, at 41 n. 133.

(20) See generally Sheri Lynn Johnson, Unconscious Racism and the Criminal Law, 73 CORNELL L. REV. 1016 (1988) (analyzing the influence of unconscious racism in charging, sentencing, and jury selection).

(21) North Carolina law provides an example of the process described in this paragraph. See N.C. GEN. STAT. [section] 15A-1212 (2009).

(22) See, e.g., North Carolina v. Wilkinson, 474 S.E. 375, 397 (1996) (noting that the "primary goal of the jury selection process is to ensure selection of a jury comprised only of persons who will render a fair and impartial verdict" and that removal for cause is appropriate if the juror is unable to render a verdict in accordance with the law) (quoting North Carolina v. Conaway, 453 S.E.2d 824, 839 (1995)).

(23) See, e.g., N.C. GEN. STAT. [section] 15A-1212(5) & (6).

(24) See, e.g., id. at [section] 15A-1217; J.E.B. v. Alabama, 511 U.S. 127, 129 (1994) (extending Batson's prohibition on race-based peremptory strikes to gender); Batson v. Kentucky, 476 U.S. 79, 89 (1986) (noting that a prosecutor may generally strike a potential juror for any reason but race).

(25) See generally Angela J. Davis, Prosecution and Race: The Power and Privilege of Discretion, 67 FORDHAM L. REV. 13 (1998) (analyzing the risks of discrimination by prosecutors or police officers when they have heightened discretion and proposing reforms that would limit discretion); Johnson, supra note 19, at 1019 (discussing Justice Powell's observation that the "multiplicity of factors that enter sentencing decisions and the consequent need for discretion may make the inference of race-based decisionmaking riskier in the sentencing context than where only a few permissible considerations enter into a decision") (citing McCleskey v. Kemp, 481 U.S. 279, 292 (1987)).

(26) See, e.g., Baldus et al., supra note 1, at 121-30; Grosso & O'Brien, supra note 2, at 1554; Rose, supra note 6, at 697-99; Turner et al., supra note 6, at 63.

(27) See, e.g., Batson, 476 U.S. at 102 (Marshall, J., concurring); Edward S. Adams & Christian Lane, Constructing a Jury That Is Both Impartial and Representative: Utilizing Cumulative Voting in Jury Selection, 73 N.Y.U. L. REV. 703, 706-07 (1998); Alan Raphael, Discriminatory Jury Selection: Lower Court Implementation of Batson v. Kentucky, 25 WILLAMETTE L. REV. 294, 349 (1989).

(28) Batson, 476 U.S. at 102 (Marshall, J., concurring); see generally Albert W. Alschuler, The Supreme Court and the Jury: Voir Dire, Peremptory Challenges, and the Review of Jury Verdicts, 56 U. CHI. L. REV. 153 (1989); William G. Childs, The Intersection of Peremptory Challenges, Challenges for Cause and Harmless Error, 27 AM. J. CRIM. L. 49 (1999); Morris B. Hoffman, Peremptory Challenges Should Be Abolished: A Trial Judge's Perspective, 64 U. CHI. L. REV. 809, 810 n.2 (1997) (listing judges and academics who have voiced strong concerns about peremptory challenges); Vivien Toomey Montz & Craig Lee Montz, The Peremptory Challenge: Should It Still Exist--An Examination of Federal and Florida Law, 54 U. MIAMI L. REV. 451 (1999); Arielle Siebert, Batson v. Kentucky: Application to Whites and the Effect on the Peremptory Challenge System, 32 COLUM. J.L. & SOC. PROBS. 307 (1998); Amy Wilson, The End of Peremptory Challenges: A Call for Change through Comparative Analysis, 31 HASTINGS INT'L & COMP. L. REV. 363 (2009).

(29) Adams & Lane, supra note 27, at 739-62; Nathan Koppel, Three Strikes and You 're Out? Critics Seek Juror-Dismissal Cap, WALL STREET JOURNAL, Mar. 5, 2009, [] (quoting Iowa law professor David Baldus calling for a three strike limit).

(30) For a review, see Sommers & Norton, supra note 1, at 527 for a review.

(31) Id.

(32) See Baldus et al., supra note 1, at 121-30; Rose, supra note 6, 697-99; Turner et al., supra note 6, at 63.

(33) See generally Grosso & O'Brien, supra note 2. This finding was the basis of a North Carolina court's grant of relief to vacate the death sentence of Marcus Robinson and resentence him to life in prison (Order Granting Appropriate Relief at 44-46, State v. Robinson, 91-CRS-23143 (Sup. Ct. N.C. Apr. 20, 2012) and subsequent grant of relief to three more defendants on December 13, 2012 (Order Granting Motions for Appropriate Relief at 2, State v. Golphin et al., 97-CRS-47314-15 (Sup. Ct. N.C. Dec. 13, 2012)). The North Carolina Supreme Court remanded both decisions for new hearings on December 18, 2015. The court found procedural errors in each decision. State v. Robinson, 780 S.E.2d 151, 151 (2015); State v. Augustine et al., 780 S.E.2d 552, 552 (2015).

(34) Cathy Johnson & Craig Haney, Felony Voir Dire: An Exploratory Study of Its Content and Effect, 18 LAW & HUM. B-EHAV. 487, 487 (1994) (noting that voir dire has seldom been studied).

(35) See VIDMAR & HANS, supra note 3, at 89 for a review.

(36) Johnson & Haney, supra note 34, at 491.

(37) Id. at 491-92.

(38) See generally Robert W. Balch, Curt Taylor Griffiths, Edwin L. Hall & L. Thomas Winfree, The Socialization of Jurors the Voir Dire as a Rite of Passage, 4 J. CRIM. JUST. 271 (1976).

(39) Joshua Klayman & Young-Won Ha, Confirmation, Disconfirmation, and Information in Hypothesis Testing, 94 PSYCHOL. REV. 211, 225 (1987); Raymond S. Nickerson, Confirmation Bias: A Ubiquitous Phenomenon in Many Guises, 2 REV. GEN. PSYCHOL. 175, 175 (1998).

(40) Todd et al., supra note 15, at 95-96.

(41) Barbara O'Brien, Prime Suspect: An Examination of Factors that Aggravate and Counteract Confirmation Bias in Criminal Investigations, 15 PSYCHOL. PUB. POL'Y & L. 315, 318-28 (2009).

(42) Nickerson, supra note 39, at 175.

(43) Klayman & Ha, supra note 39, at 225.

(44) See generally James Friedrich, Primary Error Detection and Minimization (PEDMIN) Strategies in Social Cognition: A Reinterpretation of Confirmation Bias Phenomena, 100 PSYCHOL. REV. 298 (1993); Dieter Frey, Recent Research on Selective Exposure to Information, 19 ADVANCES IN EXPERIMENTAL Soc. PSYCHOL. 41 (1986); Eva Jonas, Stefan Schulz-Hardt, Dieter Frey & Norman Thelen, Confirmation Bias in Sequential Information Search after Preliminary Decisions: An Expansion of Dissonance Theoretical Research on Selective Exposure to Information, 80 J. PERSONALITY & Soc. PSYCHOL. 557 (2001).

(45) See generally DAN SIMON, IN DOUBT: THE PSYCHOLOGY OF THE CRIMINAL JUSTICE PROCESS (2012); O'Brien, supra note 41.

(46) Bryan Gibson, David M. Sanbonmatsu & Steven S. Posavac, The Effects of Selective Hypothesis Testing on Gambling, 3 J. EXPERIMENTAL PSYCHOL.: APPLIED 126, 127 (1997).

(47) Nickerson, supra note 39, at 176.

(48) C. Neil Macrae, Alan B. Milne & Galen V. Bodenhausen, Stereotypes as Energy-Saving Devices: A Peek Inside the Cognitive Toolbox, 66 J. PERSONALITY & SOC. PSYCHOL. 37, 38 (1994).

(49) Todd et al., supra note 15, at 95-96.

(50) See generally id.

(51) Fiske, supra note 13, at 362; Johnston & McCrae, supra note 14; Snyder & Swann, supra note 14, at 1205.

(52) Nickerson, supra note 39, at 176.

(53) Todd et al., supra note 15, at 95-96.

(54) Galen V. Bodenhausen, Stereotypic Biases in Social Decision Making and Memory: Testing Process Models of Stereotype Use, 55 J. PERSONALITY & SOC. PSYCHOL. 726, 726-28 (1988).

(55) See generally Samuel L. Gaertner & John F. Dovidio, The Aversive Form of Racism, in PREJUDICE, DISCRIMINATION, AND RACISM 1, 61-89 (Samuel L. Gaertner & John F. Dovidio eds., 1986); Nao Hagiwara, Louis A. Penner, Richard Gonzalez, Susan Eggly, John F. Dovidio, Samuel L. Gaertner, Tessa West & Terrance L. Albrecht, Racial Attitudes, Physician-Patient Talk Time Ratio, and Adherence in Racially Discordant Medical Interactions, 87 SOC SCI. & MED. 123 (2013).

(56) Louis A. Penner, John F. Dovidio, Tessa V. West, Samuel L. Gaertner, Terrance L. Albrecht, Rhonda K. Dailey & Tsveti Markova, Aversive Racism and Medical Interactions with Black Patients: A Field Study, 46 J. EXPERIMENTAL SOC PSYCHOL. 436, 436 (2010).

(57) Bellin & Semitsu state as follows:
[W]e now consider the charade that has become the Batson process. The
State may provide the trial court with a series of pat race-neutral
reasons for exercise of peremptory challenges.... Surely, new
prosecutors are given a manual, probably entitled, 'Handy Race-Neutral
Explanations' or '20 Time-Tested Race-Neutral Explanations.' It
might include: too old, too young, divorced, 'long, unkempt hair,'
free-lance writer, religion, social worker, renter, lack of family
contact, attempting to make eye-contact with defendant, 'lived in an
area consisting predominantly of apartment complexes,' single,
over-educated, lack of maturity, improper demeanor, unemployed,
improper attire, juror lived alone, misspelled place of employment,
living with girlfriend, unemployed spouse, spouse employed as school
teacher, employment as part-time barber, friendship with city council
member, failure to remove hat....

Bellin & Semitsu, supra note 8, at 1090 (quoting People v. Randall, 671 N.E.2d 60, 65-66 (111. App. Ct. 1996)).

(58) John Heritage & Douglas W. Maynard, Introduction: Analyzing Interaction between Doctors and Patients in Primary Care, in COMMUNICATION IN MEDICAL CARE: INTERACTION BETWEEN PRIMARY CARE PHYSICIANS AND PATIENTS 1-21 (John Heritage & Douglas W. Maynard eds., 2006).

(59) Douglas W. Maynard & John Heritage, Conversation Analysis, Doctor-Patient Interaction and Medical Communication, 39 MED. EDUC. 428, 428 (2005).

(60) Id.

(61) Id.

(62) See generally Richard M. Frankel, Cracking the Code: Theory and Method in Clinical Communication Analysis, 13 HEALTH COMM. 101 (2001). For a list of articles applying this methodology, see Resources: Bibliography and Abstracts of RIAS Studies through 2016, RIASWORKS, (last visited July 2, 2016).

(63) Roter & Larson, supra note 17, at 37.

(64) See generally Debra Roter & Susan Larson, The Roter Interaction Analysis System (RIAS): Utility and Flexibility for Analysis of Medical Interactions, 46 PATIENT EDUC. & COUNSELING 243 (2002).

(65) Anssi Perakyla, Two Traditions of Interaction Research, 43 J. Soc. PSYCHOL. 1, 3 (2004).

(66) See generally Mary Catherine Beach, Somnath Saha, P. Todd Korthuis, Victoria Sharp, Jonathon Cohn, Ira Wilson, Susan Eggly, Lisa A. Cooper, Debra Roter, Andrea Sankar & Richard Moore, Differences in Patient-Provider Communication for Hispanic Compared to Non-Hispanic White Patients inHIVCare, 25 J. GEN. INTERNAL MED. 682 (2010) [hereinafter Differences in Patient-Provider Communication]; Mary Catherine Beach, Somnath Saha, P. Todd Korthuis, Victoria Sharp, Jonathon Cohn, Ira Wilson, Susan Eggly, Lisa A. Cooper, Debra Roter, Andrea Sankar & Richard Moore, Patient-Provider Communication Differs for Black Compared to White HIV-infected Patients, 15 AIDS & BEHAV. 805 (2011) [hereinafter Patient-Provider Communication Differs].

(67) Patient-Provider Communication Differs, supra note 66, at 809.

(68) Differences in Patient-Provider Communication, supra note 66, at 685. For other examples of research using conversation analysis to examine racial differences in healthcare provider-patient interactions, see Margarita Alegria, Debra L. Roter, Anne Valentine, Chihnan Chen, Xinliang Li, Julia Lin, Daniel Rosen, Sheri Lapatin, Sharon-Lise Normand, Susan Larson & Patrick E. Shrout, Patient-Clinician Ethnic Concordance and Communication in Mental Health Intake Visits, 93 PATIENT EDUC. & COUNSELING 188, 190 (2013) (using RIAS to compare communication patterns between mental health-care providers and patients of the same racial and ethnic groups to those in which the provider and patient did not belong to the same group); Edward P. Havranek, Rebecca Hanratty, Channing Tate, L. Miriam Dickinson, John F. Steiner, Geoffrey Cohen & Irene A. Blair, Effect of Values Affirmation on Race-Discordant Patient-Provider Communication, 21 ARCHIVES INTERNAL MED. 1662, 1663-65 (2012) (using RIAS to measure the effect of an intervention designed to improve communication between African-American patients with hypertension and their health care providers). Researchers have also used RIAS to examine the relation between other types of bias and patient care. See, e.g., Elizabeth D. Cox, Kirstin A. Nackers, Henry N. Young, Megan A. Moreno, Joseph F. Levy & Rita M. Mangione-Smith Influence of Race and Socioeconomic Status on Engagement in Pediatric Primary Care, 87 PATIENT EDUC. & COUNSELING 319, 320-23 (2012) (using RIAS to examine differences in communication between pediatric primary care providers and patients based on race, ethnicity, and socioeconomic status); Kimberly A. Gudzune, Mary Catherine Beach, Debra L. Roter & Lisa A. Cooper, Physicians Build Less Rapport with Obese Patients, 21 OBESITY 2146, 2147-49 (2013) (using RIAS to show that primary care providers developed less rapport with overweight and obese patients than with patients of a normal weight).

(69) VIDMAR & HANS, supra note 3, at 89 (describing the process of voir dire).

(70) B. Mitchell Peck & Meredith Denney, Disparities in the Conduct of the Medical Encounter, 2 SAGE OPEN 1, 2 (2012); 47 Am. Jur. 2d Jury [section] 181 (2017) (noting trial court's role in supervising the content and structure of voir dire and discretion in limiting topics of discussion).

(71) Compare Peck & Denney, supra note 70, at 1 (discussing the three functions of the medical interview: data gathering, communicating information, and relationship building), with JAMES J. GOBERT, ELLEN KREITZBERG ROSE III, JURY SELECTION: THE LAW, ART AND SCIENCE OF SELECTING A JURY, at [section] 10:2 (2013-14):
The legal rationale for voir dire is two-fold: 1) to provide the court
and counsel with the information needed to remove for cause jurors who
cannot be fair and impartial; and 2) to provide counsel with
information that allows the exercise of peremptory challenges in an
informed and intelligent manner. To ensure that voir dire is effective,
counsel should try to elicit as much relevant information as possible
from the jurors. Only then is counsel in a position to determine which
jurors should be challenged. Voir dire also provides a lawyer the
opportunity to inform jurors about the facts, set out the client's
theory of the case, and explain relevant legal principles

and id. at [section] 10:1 ("It also provides a lawyer a few moments to get to know the jurors, to establish a degree of rapport, and to tell them about the case they will be hearing.").

(72) Shari Seidman Diamond, Neil Vidmar, Mary Rose & Leslie Ellis, Juror Discussions During Civil Trials: Studying an Arizona Innovation, 45 ARIZ. L. REV. 1 (2003).

(73) N.C. GEN. STAT. [section] 15A-1214(j) (2009) ("In capital cases the trial judge for good cause shown may direct that jurors be selected one at a time, in which case each juror must first be passed by the State. These jurors may be sequestered before and after selection.").

(74) There were not enough cases involving white defendants and black victims to include this category.

(75) In one instance, the selected case had only one black juror. We eliminated this case and used SPSS to randomly select a new case.

(76) See, e.g., Grosso & O'Brien, supra note 2.

(77) This difference is statistically significant at p < .001.

(78) This difference is statistically significant at p < .001.

(79) See Grosso & O'Brien, supra note 2, at 1556, tbl. 5 (presenting a fully-controlled logistic regression model in which these factors are significant predictors of prosecutorial strike decisions).

(80) Baldus et al., supra note 1, at 41 n.133; Transcript of Closing Argument, supra note 10, at 2590-92.

(81) John K. Cochran & Mitchell B. Chamlin, The Enduring Racial Divide in Death Penalty Support, 34 J. CRIM. JUST. 85, 85 (2006); Robert Fitzgerald & Phoebe C. Ellsworth, Due Process vs. Crime Control: Death Qualification and Jury Attitudes, 8 L. & HUM. BEHAV. 31, 46 (1984); James D. Unnever & Francis T. Cullen, Reassessing the Racial Divide in Support for Capital Punishment the Continuing Significance Of Race, 44 J. RES. CRIME & DELINQ. 124, 125 (2007); James D. Unnever & Francis T. Cullen, The Racial Divide in Support for the Death Penalty: Does White Racism Matter?, 85 SOC. FORCES 1281, 1281 (2007).

(82) See Grosso & O'Brien, supra note 2, at 1554-55, 1556, tbl. 5.

(83) Id. at 1556, tbl. 5.

(84) Id. This difference is statistically significant at p < .001.

(85) See Bellin & Semitsu, supra note 8, at 1096-97 (reviewing commonly proffered reasons for peremptory strikes and noting the frequency with which prior experience with the criminal justice system was accepted as a race-neutral reason notwithstanding its disproportionate impact on specific racial and ethnic groups); Gilad Edelman, Why Is It So Easy for Prosecutors to Strike Black Jurors?, THE NEW YORKER (June 5, 2015), (arguing that "race is an unfortunate but powerful generalization" because "black people are more likely to have been targeted or abused by police" and "to be affected by the extreme racial disparities in arrests, incarceration, and the death penalty").

(86) See generally Vida B. Johnson, Arresting Batson: How Striking Jurors Based on Arrest Records Violates Batson, 34 YALE L. & POL'Y

REV. 387 (2016) (reviewing the practices of striking jurors based on arrest record and concluding that doing so violates Batson).

(87) Grosso & O'Brien, supra note 2, at 1556, tbl. 5. This variable does not include close relatives of the jurors who were accused of a crime. Both this variable and the broader variable predict state strikes, but the broader variable would target too many jurors for this project.

(88) Data on file with authors.

(89) Grosso & O'Brien, supra note 2, at 1556, tbl. 5.

(90) Roter & Larson, supra note 17, at 35.

(91) id.

(92) See infra Appendix 1 for a complete list of text codes.

(93) For instance, RIAS includes specific codes for biomedical information giving and counseling, distinct from psychosocial information giving and counseling. The former involves information specific to the patient's condition and treatment plan, and the latter involves broader information about lifestyle. Those precise codes would not apply in the voir dire setting, but there are other distinctions that are relevant, such as information about the specific case being tried versus the law more generally.

(94) See infra Appendix 1. See, e.g., Lisa A. Cooper, Debra L. Roter, Kathryn A. Carson, Mary Catherine Beach, Janice A. Sabin, Anthony G. Greenwald & Thomas S. Inui, The Associations of Clinicians ' Implicit Attitudes about Race with Medical Visit Communication and Patient Ratings of Interpersonal Care, 102 AM. J. PUB. HEALTH 979, 980 (2012) ("This system assigns each thought the patient and clinician express to mutually exclusive and exhaustive codes that can be combined to reflect categories of exchange, including functions of the medical interview."); Differences in Patient-Provider Communication, supra note 66, at 683:
RIAS analysts assign one of 37 mutually exclusive and exhaustive
categories to each complete thought expressed by either the patient or
provider (referred to as an utterance). These categories can be
combined to reflect four broad types of exchange...: socio-emotional
communication (including emotional talk, positive talk, negative talk,
and social chit-chat), information-giving (including biomedical and
psychosocial/lifestyle information and counseling), question-asking
(including biomedical and psychosocial/lifestyle questions), and
patient activation (such as asking for the others' opinions, confirming
the others' understanding, or clarifying one's own understanding.

(95) We hired separate coders to complete the descriptive information, strike decisions, and removals for cause to identify all targeted jurors in the study.

(96) Coders were typically blind to race, as well, except for a few instances when race was explicitly mentioned by one of the parties in the transcript. We decided against redacting that information as doing so would have drawn attention to the issue of race and might have alerted the coders that we were interested in racial disparities.

(97) The percentages do not add up to 100% because some potential jurors had more than one of the targeted characteristics and thus fell into more than one target group.

(98) Grosso & O'Brien, supra note 2, at 1533 (explaining the scope of the research).

(99) Column G includes all of the jurors in columns C-F but is not the sum of columns C-F. A juror may fall into more than one target group (presented in columns C-F), but ultimately appears only once in the study sample (presented in column G).

(100) This column includes venire members selected as alternates.

(101) See generally Grosso & O'Brien, supra note 2; Barbara O'Brien, Catherine M. Grosso, George Woodworm & Abijah Taylor, Untangling the Role of Race in Capital Charging and Sentencing in North Carolina, 1990-2009, 94 N.C. L. REV. 1997 (2016).

(102) N.C. GEN. STAT. [section] 15A-1214 (2009) (defining jury selection procedures).

(103) See, e.g., Roter & Larson, supra note 17, at 37 (citing studies where these approaches have been taken).

(104) See generally Douglas R. Rice & Christopher Zorn, Corpus-Based Dictionaries for Sentiment Analysis of Specialized Vocabularies, Presentation at the New Directions in Analyzing Text as Data Workshop: September 27-28, 2013 (Sept. 19, 2013) (preliminary draft).

(105) Roter & Larson, supra note 17, at 37.

(106) N.C. GEN. STAT. [section] 15A-1214 (defining jury selection procedures).

(107) David C. Baldus, George Woodworth, David Zuckerman, Neil Alan Weiner & Barbara Broffitt, The Use of Peremptory Challenges in Capital Murder Trials: A Legal and Empirical Analysis, 3 U. PA. J. CONST. L. 3, 51-55 (2001).

(108) Not all columns sum because of missing race information.

(109) Roter & Larson, supra note 17, at 37.

(110) See Debra L. Roter, Moira Stewart, Samuel M. Putnam, Mack Lipkin, William Stiles & Thomas S. Inui, Communication Patterns of Primary Care Physicians, 277 J. AM. MED. ASS'N 350, 351 (1997) (explaining the summary groupings of RIAS codes in a study analyzing doctor-patient communication as including question-asking, biomedical information giving, psychosocial exchanges, positive talk, negative talk, social talk, facilitation, and orientation).

(111) See infra Appendix 2 for a definition of the codes included in each category. Note that this profile assumed a prosecutor as speaker. Subsequent work will prepare profiles by speaker.

(112) See generally Grosso & O'Brien, supra note 2, at 1531; O'Brien, Grosso, Woodworth & Taylor, supra note 101, at 1997.

(113) See North Carolina v. Badgett, 644 S.E.2d 206, 209-11 (N.C. 2006) (recounting the facts of the crime and the trial).

(114) Id. at 223.

(115) Several local news articles printed several paragraphs each on the murder; none seem to have placed it on the front page or in the front section. Kathi Keys, Suspect Arrested in Death of City Man, THE COURIER-TRIB., Dec. 2, 2002, at 2A; Mark Brumley, Ex-Convict Arrested in Homicide: Records Show John Scott Badgett--Charged with Murder in a Weekend Homicide--Killed A Man Years Ago, GREENSBORO NEWS & REC, Bl, Dec. 3, 2002, at Bl; Mark Brumley, Man Charged with Murder in Stabbing: John Scott Badgett, Who Served Four Years in Prison for Another Stabbing Death, is Accused of Killing Grover Arthur Kiser, GREENSBORO NEWS & REC, Bl, Dec. 5,2002, at B1.

(116) O'Brien, Grosso, Woodworth & Taylor, supra note 101, at 1998.

(117) See id. at 2027 (presenting fully controlled logistic regression models documenting those factors associated with a highly aggravated case in North Carolina during this time period).

(118) See ABA Guidelines for Capital Defense, Guideline 10.11(A) (2003), available at represent ation/2003guidelines.authcheckdam.pdf ("[C]ounsel at every stage of the case have a continuing duty to investigate issues bearing upon penalty and to seek information that supports mitigation.").

(119) Population of Randolph County, North Carolina: Census 2010 and 2000 Interactive Map, Demographics, Statistics, Graphs, Quick Facts, CENSUSVIEWER, [] (last visited Aug. 8, 2017).

(120) The person coded as other described himself as "other" in the jury questionnaire and to the North Carolina Board of Elections. We do not have additional information about his race.

(121) In comparison, the purposive sample of venire members with at least one of the target characteristics was 6.1% black (2/33) and 93.9% white (31/33). None of the black venire members in the sample were excused for cause, but 61.3% of the white potential jurors were excused (19/31). Among venire members qualified to serve, the state struck all of the black venire members in the sample (2/2), and 33.3% of the white venire members (4/12). Three white jurors from the sample were seated on the jury, and no black jurors in the sample were seated.

(122) See Grosso & O'Brien, supra note 2, at 1531, tbl. 5.

(123) Transcript of Jury Selection at 1018-19, State v. Williams, No. 01 CRS 63278 (N.C. Super. Ct. Nov. 8, 2004).

(124) Id.

(125) In Miller-El, prosecutors deployed a "graphic description" of how death-sentenced prisoners were executed in Texas, intended to increase the potential juror's reservations about the death penalty, with 53% of black jurors but only 6% of white jurors. David C. Baldus, Catherine M. Grosso, Robert Dunham, George Woodworth & Richard Newell, 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, 1436 (2012) (citing Miller-El v. Dretke, 545 U.S. 231, 260 (2005)).

(126) See generally Roter & Larson, supra note 17 (2001).

(127) See supra Part II; Roter & Larson, supra note 64 (2002).

(128) See supra Part II.C.

A      B       C             D        E         F           G
       Total   Death         Accused  Service   Knows       Total
Case   Venire  Penalty       of       is        Police or   Jurors in
               Reservations  Crime    Hardship  Prosecutor  Study (99)

 14.0    67     11            2        16        12          33
 41.0   138     64            9        18        25          88
 88.2   107     23            5        37        14          73
103.0    65     17            8        13        10          37
190.0   275     72           13       106        33         194
220.1   141     27           10        58        13          93
256.0    90     20            6        15        15          48
294.0    70     24            1        14         3          38
319.0   103     19            1        33         8          57
327.0   118     51            6         7         5          65
330.0    74     24            4         7         8          37
363.0    57     15            2        16         3          29
Total  1305    367           67       340       149         792


A       B       C        D            E             F
Case    State   Defense  Excused for
        Struck  Struck   Cause        Seated (100)  Totals

 14.0    6       5        19           3             33
 41.0    9       6        69           4             88
 88.2    8       8        48           9             73
103.0    8       6        16           7             37
190.0   13       6       169           6            194
220.1   11       6        72           3             93
256.0    5       7        32           6             48
294.0    6       3        26           3             38
319.0    7       4        43           3             57
327.0    7       3        51           5             65
330.0   10       3        20           4             37
363.0    6       2        18           3             29
Totals  96      59       583          56            792


         State Strikes               Defense Strikes
         Passed          Struck      Passed     Struck

          26.0%           74.0%      81.8%      18.2%
Black    (13/50)         (37/50)     (9/11)     (2/11)
          64.3%           35.7%      39.4%      60.6%
White    (99/154)        (55/154)    (37/94)    (57/94)
Other      0              (2/2)       0           0
Unknown    0              (2/2)       0           0
          53.8%            46%       43.8%      56.2%
Total    (112/208)      (96/208)   (46/105)   (59/105)

           Cause Challenges
           Passed    Excused

           21.7%       78.3%
Black     (49/226)   (177/226)
           31.0%       69.0%
White     (156/503)  (347/503)
           15.4%       84.6%
Other     (2/13)       (11/13)
            4.0%       96.0%
Unknown   (2/50)     (48/50)
           26.4%       73.6%
Total    (209/792)   (583/792)


A      B          C                        D
Case   Jurors     Average # of Utterances  Average # of Characters
       per Juror                           of Speech per Juror

 14.0   33        246                      459,395
 41.0   88        432                      799,806
 88.2   73        232                      431,777
103.0   37        256                      478,394
190.0  194        192                      351,563
220.1   93        181                      343,197
256.0   48        395                      743,462
294.0   38        208                      394,354
319.0   57        198                      275,690
327.0   65        166                      309,712
330.0   37        338                      626,512
363.0   29        200                      379,400
Total  792        245                      466,105


    A           B               C
                Open Questions  Closed Questions
                n         %     n          %

1.  Court         336     15%    6,437     18%
2.  Defense       597     27%    7,906     22%
3.  Prosecutor  1,259     57%   22,319     61%
4.  Totals      2,192           36,662
                    6%              94%


A                  B           C         D          E           F
                   All Jurors  Black     All Other  Difference  Ratio
                               Jurors    Jurors     (C-D)       (C/D)

14 Badgett         (n = 33)    (n = 2)   (n = 31)
Prosecutor/Juror     0.63        1.59      0.57      1.02       2.79
Defense/Juror        0.53        0         0.57     -0.57       0.00
Judge/Juror          0.95        0.92      0.96     -0.04       0.96
41 Braxton         (n = 88)    (n = 48)  (n = 40)
Prosecutor/Juror     1.11        1.09      1.13     -0.04       0.96
Defense/Juror        0.50        0.39      0.59     -0.2        0.66
Judge/Juror          0.43        0.53      0.34      0.19       1.56
88.2 Duke          (n = 73)    (n = 6)   (n = 53)
Prosecutor/Juror     0.73        1.05      0.72      0.33       1.46
Defense/Juror        0.56        0.55      0.58     -0.03       0.95
Judge/Juror          0.51        0.41      0.47     -0.06       0.87
103 Gainey         (n = 37)    (n = 7)   (n = 29)
Prosecutor/Juror      .53         .78       .48       .3        1.63
Defense/Juror        0.19        0.07      0.22     -0.15       0.32
Judge/Juror          0.70        0.81      0.67      0.14       1.21
190 Manness        (n = 194)   (n = 61)  (n = 123)
Prosecutor/Juror     1.35        1.66      1.25      0.41       1.33
Defense/Juror        0.59        0.28      0.71     -0.43       0.39
Judge/Juror          0.32        0.36      0.31      0.05       1.16
220.1 Al-Bayyinah  (n = 93)    (n = 8)   (n = 85)
Prosecutor/Juror     0.81        1.01      0.77      0.24       1.31
Defense/Juror        0.30        0.20      0.33     -0.13       0.61
Judge/Juror          1.10        0.91      1.13     -0.22       0.81
256 Richardson     (n = 48)    (n = 15)  (n = 33)
Prosecutor/Juror     0.70        2.20      0.70      1.5        3.14
Defense/Juror        0.55        0.35      0.49     -0.14       0.71
Judge/Juror          0.99        1.60      0.88      0.72       1.82
294 Stroud         (n = 38)    (n = 14)  (n =23)
Prosecutor/Juror     0.86        0.88      0.85      0.03       1.04
Defense/Juror        0.47        0.36      0.54     -0.18       0.67
Judge/Juror          0.71        0.74      0.68      0.06       1.09
319 Warren         (n = 57)    (n = 4)   (n = 45)
Prosecutor/Juror     0.76        0.93      0.73      0.2        1.27
Defense/Juror        0.39        0.47      0.41      0.06       1.15
Judge/Juror          0.66        0.78      0.62      0.16       1.26
327 Williams       (n = 65)    (n = 45)  (n = 20)
Prosecutor/Juror     1.02        1.10      0.85      0.25       1.29
Defense/Juror        0.26        0.20      0.39     -0.19       0.51
Judge/Juror          0.62        0.65      0.54      0.11       1.20
330 Williams       (n = 37)    (n = 22)  (n = 14)
Prosecutor/Juror     0.85        0.68      0.90     -0.22       0.76
Defense/Juror        0.45        0.54      0.42      0.12       1.29
Judge/Juror          0.42        0.77      0.32      0.45       2.41
363 Thomas         (n = 29)    (n = 3)   (n = 26)
Prosecutor/Juror     1.66        1.05      1.71     -0.66       0.61
Defense/Juror        0.75        0.02      0.82     -0.8        0.02
Judge/Juror          0.26        0.44      0.24      0.2        1.83
COPYRIGHT 2017 Northwestern University, School of Law
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2017 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:The Death Penalty's Numbered Days?
Author:O'Brien, Barbara; Grosso, Catherine M.; Taylor, Abijah P.
Publication:Journal of Criminal Law and Criminology
Date:Sep 22, 2017

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters