Scientific Jury Selection

Scientific jury selection (SJS) is the use of a survey to decide which jurors to favor in a trial. Prior to the 1970s, jury selection was done by attorneys based on their hunches. The new quantitative method was welcomed enthusiastically by trial attorneys. Social scientists were more reserved. SJS led to the employment of social scientists as trial consultants. This entry describes traditional jury selection as conducted by attorneys, reviews the origin of SJS, presents an example of a 2003 survey used in Florida to implement SJS, and examines evidence on the utility of SJS.

Voir dire is the process at the beginning of a trial when prospective jurors, called venirepersons, are examined verbally to determine their fitness for service as jurors in a particular trial. An unlimited number of venirepersons can be excused for cause—that is, found unfit by the trial judge for reasons of incompetence or prejudice. Venirepersons are also excused peremptorily, that is, by the attorneys without stated reason. The rules for using peremptory excusals vary by jurisdiction and preferences of trial judges. Judges differ widely in the questions they ask or allow the attorneys to ask. Some judges allow venirepersons to complete trial-specific questionnaires constructed by the attorneys. The information about venirepersons available to the attorneys varies greatly, given the court’s voir dire practice. Nevertheless, in U.S. courts, the parties (prosecution, plaintiff, and defendant) have a right to a certain number of peremptory excusals. These peremptory excusals are determined and used during the jury selection process.

Traditional Attorney-Conducted Jury Selection

How do attorneys evaluate venirepersons and decide whom to favor or oppose? Lawyers are sometimes influenced by the published preferences of famous colleagues—the idols of the tribe. These famous trial lawyers have published preferences mainly based on ethnic, religious, gender, and occupational stereotypes that may often conflict with each other. These stereotypes were formed long ago and have no application to modern jurors or cases. Many attorneys hold other stereotypes that may have some limited value, for example, that nurses are unsympathetic to pain and suffering or that people who use newspaper coupons give stingy monetary awards. In addition, a trial attorney’s experience with a particular type of juror may result in prejudice for or against such jurors in future trials. Attorneys also evaluate venirepersons on the basis of “vibes”—their impression of the venireperson’s nonverbal behavior and deportment. Generally, the attorney has a limited profile of good and bad jurors derived from advice from other attorneys, prejudices, speculation, and experience.

Origin of Scientific Jury Selection

In the 1972 Harrisburg Seven trial of Vietnam War resisters, Jay Schulman and colleagues decided to evaluate venirepersons on the basis of a survey. Schulman surveyed 840 respondents in the trial venue, recording diverse attitudes possibly related to juror view of war resisters. Respondents indicated what they thought of various antiwar activities, which historical figures they admired, their level of trust in government and the establishment, and many demographic features such as age, sex, education, occupation, media preference, and residence. Demographic variables were then statistically correlated with attitudes suggesting support for the government and the Vietnam War. Schulman reported that respondent sex, education, religion, and media preference distinguished “good” from “bad” jurors. In principle, jurors at the trial who had features that were bad (male, better educated, Protestant, attended to local media) would be excused peremptorily. Despite the apparently scientific approach, it is clear from Schulman’s article that the Harrisburg Seven jury was chosen by a mishmash of survey findings and old-fashioned practice (including informant information on certain venirepersons; speculation about how a given juror was likely to relate to other jurors; speculation on racism; and deliberation among the multiple lawyers, social scientists, and defendants about the desirability of prospective jurors).

If it is unclear how SJS was employed in the Harrisburg Seven trial, it was abundantly clear to the litigation profession that a promising new tool was available to assist in jury selection. While Schulman’s academic colleagues continued pro bono work on political trials, some of his followers formed the National Jury Project in 1975. In 1979, in the wake of success in the MCI/ATT antitrust case, Donald Vinson founded Litigation Sciences. In 1987, he founded Decision Quest, which was employed by prosecutor Marcia Clark in the 1995 trial of O. J. Simpson. Today, trial consultation is a billion-dollar enterprise, with practitioners in all metropolitan areas.

A Scientific Jury Selection Survey

The following survey was conducted in 2003 in preparation for a trial with charges of vehicular homicide while driving under the influence of alcohol, a felony punishable by as much as 50 years in a Florida state prison. The sample in this SJS survey consisted of 211 Miami-Dade county residents whose sex, age, family income, marital status, education, and race/ethnicity closely paralleled those of registered voters, from among whom state court venirepersons are subpoenaed. These people’s responses should be similar to those of people who show up for duty at this trial.

The survey begins with the case summary, a one-page description of the most essential features of the case, to wit, that an elderly lady was hit and killed by a speeding BMW. The driver failed to stop. A witness with a cell phone described the driver as a black or dark-skinned male. A police BOLO (be on the outlook for) broadcast this description. The police found the abandoned, damaged car within 20 minutes at a distance of 1.5 miles from the accident. About 1 hour after the accident, the police went to the apartment of the registered owner of the car, who lived 8 miles away. He was a light-skinned White man who was wearing his bathrobe. He had been in bed and in the shower and appeared intoxicated. There were glass shards in his robe and in the shower. He admitted having been out drinking all night at two gay bars. He claimed his car was parked below in the garage. When it was not found there, he agreed with the police that it must have been stolen. He was interrogated at the police station and continued to deny guilt. He claimed to have blacked out in a bar and to be unable to remember anything thereafter. His blood alcohol level 4 hours after the accident was 0.17, far above the legal limit. At that point he was arrested. The summary concludes with a statement of a defendant’s right to be presumed innocent and stresses that the entire burden of proving guilt rests with the prosecution.

The summary is followed by a 9-point verdict scale asking the respondent to indicate the likelihood of guilt based on what he or she knows. This verdict scale is repeated after each separate additional fact introduced in the second section. A final verdict scale comes after the respondent has heard the case summary and all eight additional facts. This is the most important measure of the verdict.

The second section of the survey records the respondent’s verdict after hearing specific critical facts. These verdicts determine trial themes and are the most important information in an SJS survey. They tell the attorney which facts fly and which crash, what to leave in the argument at a trial, and what to leave out. It is no simple matter to know which evidence and which themes to emphasize in a trial.

Recall that respondents indicated verdicts after hearing eight specific bits of evidence or argument. In terms of the prosecution’s case, respondents did not increase their guilt ratings on the basis of evidence stressing that the defendant’s car was involved or that he was out drinking all night without an alibi. However, evidence that the glass spray pattern in the damaged auto proved that no one other than a driver was in the car at the time of the accident resulted in a sharp increase in guilt ratings. These findings identify the critical defense problem that must be successfully addressed.

Two defense themes proved most helpful. First, the police BOLO identified the driver as a dark-skinned male. Second, sloppy police work such as failing to fingerprint the steering wheel led to lower guilt ratings. The defense was on notice to focus on these points and ignore certain other evidence such as a doctor’s testimony that the defendant had no memory of postblackout events under hypnosis.

The third section of the survey consists of open-ended items answered in the respondent’s own words. Respondents are asked for the best reasons for convicting and acquitting, and for any suggestions they might have. The responses confirmed that the best reasons for acquittal were the legal ground rules favoring the defendant, sloppy police work, and the BOLO. Responses to free-response questions often produce useful articulations. These can be used to make a point in argument.

The fourth section consists of opinion items. Respondents used rating scales to indicate their opinions about alcohol and driving, sloppy police work, eyewitness accuracy, and conventionality. The results indicated that opinions about convicting when the police work has been sloppy, not convicting on strong suspicion but only on proof, and concern about irresponsibility and sexual immorality predicted the verdict. The defense would focus on these opinions in voir dire.

Experience has shown that demographic predictors emphasized in the original SJS research are usually less predictive of the verdict than are experiences and/or opinions. Nevertheless, as in the current case, they are sometimes predictive. This is extremely helpful when court procedure limits voir dire as in the federal courts. The current study indicates Cuban Whites reach verdicts similar to those of married jurors, jurors with lower incomes and less education, male jurors, and jurors who never drink even moderate amounts of alcohol when driving. These characteristics are associated with conviction.

In the current case, the survey has distinguished critical and less significant evidential issues. Nine opinion and demographic predictors have been identified. In juror selection, venirepersons with more of these features will be favored for defense peremptory excusals.

The disposition of this case is instructive in evaluating the utility of SJS. Preparation for jury selection is only one aspect of trial outcome and usually far from the most important. In this case, as noted, the defense knew from the survey that the police BOLO indicating that the driver was a dark-skinned male was crucial to introducing doubt that the fair-skinned defendant was driving the car. The prosecution argued that the BOLO was in error because the eyewitness in fact told the police that it was the car that was black. Unfortunately for the defense, the judge disallowed mention of the BOLO. Faced with 50 years in prison if convicted, the defendant agreed to plead guilty and serve 12 years in prison. In this case, SJS succeeded in identifying crucial evidence and juror features predictive of the verdict.

Utility of Scientific Jury Selection

Informed opinion differs, but most academic reviewers are skeptical that SJS improves attorney-conducted jury selection. Eight traditional published studies of the statistical utility of SJS indicate that on average only 11% of the factors that determine a juror’s verdict are explained by SJS opinion and demographic predictors. This is, however, better than it may sound as it suggests an improvement of 17% in the accurate prediction of a venireperson’s verdict. If, for the sake of argument, we assume that attorney jury selection is no better than guesswork—50% accurate (it is unknown but probably better on average)—then SJS improves this to 67% accurate prediction. Eleven (nine unpublished) additional SJS studies at Florida International University in recent years suggest that verdict prediction accuracy can be further improved (up to 76% accuracy). SJS is of no value if the evidence makes conviction inevitable. SJS is more frequently used in civil trials, where the evidence is more likely to be balanced. Finally, it must be recognized that the survey must capture the essence of the evidence to be presented at the trial as well as case-relevant opinions, which are known to predict juror verdicts better than demographic characteristics or general attitudes do.


  1. Kovera, M. B., Dickinson, J. J., & Cutler, B. L. (2003). Voir dire and jury selection: Practical issues, research findings, and directions for future research. In A. M. Goldstein (Ed.), Handbook of psychology: Vol. 11. Forensic psychology (pp. 161-175). New York: Wiley.
  2. Lieberman, J. D., & Sales, B. D. (2007). Scientific jury selection. Washington, DC: American Psychological Association.

Return to the overview of Trial Consulting in Forensic Psychology.