Professor says data mining can improve jury selection
While advertisers, marketing companies and political campaigns routinely glean personal information about consumers and voters from databases, such data remain largely untapped when it comes to jury duty.
These billions of bits of unmined data offer a wealth of opportunity for courts to increase juror diversity and for litigants to acquire information once available only from expensive jury consultants, says Andrew Guthrie Ferguson, a law professor at the University of the District of Columbia.
“Lawyers tend to pick jurors based on things they can see—race, gender, age, sexual orientation—in part because they don’t know anything more,” says Ferguson, who envisions a shift in the jury system as the use of database technology increases.
The nation’s court systems are using a practically antiquated system to summon potential jurors by randomly culling names from driver’s license records and voter registration lists. The process has been used since 1968, when a federal law was enacted to drive racism, sexism and classism from jury selection. Ferguson, who studies juries, describes these records as “dim data.” That’s because, by design, they reveal little about potential jurors.
Using dim data to summon and select jurors, however, hasn’t helped eliminate racism, classism or ageism, according to Ferguson, who authored a recent article in Notre Dame Law Review on this issue. He points out that public records used for summonsing jury pools can be stale, resulting in a high no-show rate among potential jurors. And using dim data in voir dire encourages lawyers to rely on appearance, affect, hunches and stereotypes, which may actually result in discriminatory use of peremptory challenges.
But by turning instead to what Ferguson calls “bright data,” deficiencies and inequities in both the jury summonsing and selection processes could be eliminated. Bright data—more commonly referred to as big data—is about gaining insights using large-scale information and statistics in fresh ways. Big data centers around data brokers, who buy information that private companies collect about ordinary people, much of which is obtained after customers hit the “I agree” button online or use a store loyalty card that records usage. Other data vendors mine social media activity for information. And, yes, it’s all legal.
Before a trial, attorneys for both sides routinely obtain the names of potential jurors on the day of jury selection. It’s now possible using big-data sources to flag or score potential jurors on certain factors—fiscal and social ideology, for examply, or on attitudes relevant to liability or damages—enabling lawyers to make exceedingly nuanced strikes.
Just as important, big data could be used by court administrators ahead of time for more accurate and efficient summonsing by drawing up-to-date information about the jury pool, which can improve the rate of response.
But because they are notoriously traditional institutions, courts are only just beginning to consider ex-perimenting with this kind of technological advancement. The Council for Court Excellence in Washington, D.C., for instance, recommended investigating big data in its December 2015 report, Jury Service Revisited: Upgrades for the 21st Century.
In 2014, the Superior Court of the District of Columbia mailed 150,454 summonses to potential jurors, according to the report. Of those, 70,715 were registered as “failure to respond” and 22,027 were returned as undeliverable, a situation that could be fixed with updated juror information obtained from big-data companies.
Every year, the scale and accuracy of big data gets better and more robust, according to Tucker Willsie, co-founder of Jury Mapping, one company seeking to apply this technology to the court system. “There are thousands of data points that allow us to slice the population into segments,” he explains.
Information purchased from data vendors can quickly reveal, for example, which jurors own a gun, who has a nurse in the family, who votes Republican or who subscribes to parenting magazines. “People are looking for an advantage in the time-constrained jury process. After months of trial prep, voir dire is typically very rushed,” Willsie says.
But just because big data could be used in both summonsing and jury selection doesn’t mean it should be, according to Ferguson. Indeed, several legal and ethical issues are raised by the specter of big-data juries. Most notable is the issue of privacy. Many citizens don’t want the government knowing the magazines they read, the places they shop or the TV programs they watch. Were the government to use that information or supply it to litigants, the result could be a backlash against jury service.
In addition, big data raises existential questions about the role of juries. Specifically, the jury system was established on the very notion that every citizen is equal. “If we’re seeking to create an algorithmically perfect venire, that undermines equality,” Ferguson says.
Related to equality concerns are 14th Amendment equal protection issues raised by the deliberate microtargeting of jurors. For example, seemingly neutral information—subscriptions to, say, Essence or Southern Living magazines—may actually serve as a proxy for race or gender.
Yet some jury experts say the flip side may be true: Nuanced information gained from statistics may provide a reason for keeping a particular juror when a party might have otherwise struck that person on impermissible grounds. “I’m persuaded by the argument that if lawyers had more information than just race and gender, they’re less likely to focus on race and gender,” says Nina Chernoff, a professor at the City University of New York School of Law who studies the composition of juries.
There’s also the issue of cost. For litigants, this kind of data is only available to parties who can afford sophisticated jury consultants. “We hope to help correct the disparity of resources,” says Jury Mapping’s Willsie. “Our services cost a fraction of a jury consultant.” His website uses big data and predictive analytics to help litigants quickly and affordably zero in on desirable jurors.
“Normally, doing research on 300 people in advance would be prohibitively challenging and time-consuming, and it would lack the predictive scoring. With our tool, it’s all at their fingertips,” Willsie says. “The day of the trial, we’ll learn the names of the 50 individuals assigned to the courtroom, as well as the ordering. Armed with a ranking that we helped establish in advance, our client can look at the first six seated jurors and know who to direct questioning to—and who not to—to avoid exposing a favorable juror to the other side.”
While big data is cheaper than jury consultants, it may be too expensive for court administrators, at least at first glance. However, in the long run, it might actually result in savings. Because summons yield rates are as low as 12 percent, courts must mail far more summonses than necessary, which requires postage and manpower.
BIG VS. BAD
“Yield rates are low in part because courts have bad records,” Ferguson explains. “Big-data services find out about [citizens’] moves right away,” in contrast to out-of-date Department of Motor Vehicle and voter records.
But Chernoff insists that whatever the cost, courts simply can’t yet handle large-scale information. “Jury offices are barely managing the basic data they’re required to. They’re struggling to keep pace with technology from 10 years ago,” she says.
In Hartford, Connecticut, for example, a computer error caused the federal jury selection system to interpret the “d” in Hartford to mean “deceased.” So no one from Hartford was called for jury service, eliminating more than 60 percent of eligible African-Americans. Similarly, the computer system in Kent County, Michigan, had an erroneous setting that only selected jurors from certain ZIP codes, particularly those without larger African-American populations.
“Courts do a horrendously poor job of dealing with ‘dim data.’ That’s the baseline: complete mismanagement,” Chernoff says. “It’s not malfeasance; they’re just under-resourced.” So adopting bright-data technologies may be too much to ask of overburdened jury administrators.
As for the future of big-data juries, Ferguson maintains that the two potential uses for big data don’t need to be adopted together. For instance, court administrators could use big data for more efficient venire creation, but then not provide that personal data to litigants for actual jury selection. Either way, the possibilities and potential effects should be studied.
“Big-data companies could do this in a heartbeat,” Ferguson says. “All litigants want more information on potential jurors. Every civil lawyer, prosecutor and defense lawyer would encourage this.”
This article originally appeared in the September 2016 issue of the ABA Journal with this headline: “Big-Data Juries: Professor says drawing from a variety of databases can improve the jury system.”