Consultant: Big data and online studies upending jury selection

It may sound creepy, said John Campbell, but it’s true: The amount of data online about private individuals is so vast, it’s possible for an informed attorney to walk into jury selection with his or her ideal jurors already picked.

“I was recently in a trial in Phoenix where it was clear to me that the defense did not need to talk to the jury [during voir dire],” said Campbell, an attorney in St. Louis and an associate professor at the University of Denver Sturm College of Law. “They asked a few questions that were clearly designed to confirm information they already had.”

Big data is upending jury selection, Campbell explained on June 26 during his CLE presentation for the Bar Association of Metropolitan St. Louis, held at the Missouri Athletic Club.

Dozens of providers have sprung up during the past two years offering lawyers the use of sophisticated search algorithms, Campbell observed. These tools can sometimes turn up a potential juror’s voting record, political donations, marital status, employment, education, property value — even what they watch on TV. Soon, full reports will be available in a matter of minutes.

“I suspect it will become almost malpractice to not do this in jury selection in the next few years,” said Campbell, who also is co-founder of a consultancy, Empirical Jury.

These search results can be useful, Campbell said, because online jury studies now can reveal the characteristics of jurors who, on average, will be most receptive to a lawyer’s case.

While many firms have used focus groups to make such determinations, it’s now possible for plaintiffs’ attorneys or defense teams to test their cases online with sample sizes that are much larger — in the hundreds or even thousands. All they need to do is use sites such as Amazon Mechanical Turk that match them up with internet users willing to perform online tasks for modest sums of money.

The difference between a traditional focus group and an online jury study is this, Campbell said: When you test your case online, you can have 500 responses within a couple of hours. That will tell you whether your sample was representative and free of statistical anomalies.

The way to ensure the online test jurors are taking the study seriously, Campbell said, is to insert questions that verify whether they’re paying attention and comprehending the material.

For example, after signing nondisclosure agreements, test jurors can be asked questions about themselves more than once; if their answers vary, they’re routed out of the study.

If they’re clicking the same multiple choice answer for every question, they might hit “Strongly Disagree” when asked if they’re paying attention, and get booted that way. If they can’t answer obvious questions about the material — such as how an accident victim shown in an evidence video was carried away from a wreck — then they’ll be removed from the study.

To check for bias, a lawyer also can ask the test jurors: Who do you think is paying for this survey? If most of the test jurors believe an academic institution is behind the research, for instance, the data will be more reliable because the test jurors won’t clearly be trying to tell the plaintiffs’ lawyers what they want to hear.

Online jury studies already have revealed a variety of patterns, Campbell said. In one study of a murder case, a male attorney was seen as most effective when he delivered the closing argument angrily, whereas a female attorney was considered most effective when she delivered it calmly.

In another study of a medical-malpractice suit against two doctors, the test jurors found at least one of them liable 69 percent of the time when both went to trial; however, upon hearing that one of them settled pretrial, the test jurors found the remaining doctor liable only 42 percent of the time. Knowing this in advance would change a litigant’s entire strategy, Campbell suggested.

Online jury studies also allow a lawyer to test requests for damages and see whether they affect findings of liability. In other words, the data may point to a dollar amount at which jurors start penalizing a plaintiff’s attorney by declining to find the defendant liable. The goal, Campbell said, is to find the “Goldilocks” amount — one that’s not too little and not too much, but just right.

“We’ve used this with some effect in a number of cases,” Campbell said. “Otherwise you’re sort of pulling [the request] out of thin air.”

Campbell said that right now, the effect of these tools is asymmetric in that they’re predominantly being used by plaintiffs’ lawyers, but once everyone else follows suit, the playing field will level out.