Natural Language Processing Software Can Identify Biased Jury Selection, Has Potential to Be Used in Real Time During Voir Dire
by Jo Ellen Nott
The Journal of Empirical Legal Studies published a ground-breaking study out of Cornell University that proves prosecutors question Black venire persons in a hard-to-detect, but significantly different way, than they question other potential jurors.
“Quantifying Disparate Questioning of Black and White Jurors in Capital Jury Selection” was published on July 14, 2023, and shows that natural language processing software and data science can successfully identify when prosecutors question potential jurors based on race or gender.
This type of disparate questioning is illegal, but it still occurs nearly 40 years after the Supreme Court banned it in Batson v. Kentucky, 476 U.S. 79 (1986), in which the Court held that using preemptory challenges in a racially (or gender based) discriminatory manner violates the Equal Protection Clause of the Fourteenth Amendment. In the years since, enforcement by the Supreme Court and state and federal courts has been uneven, according to the study’s authors.
The study analyzed transcripts from 17 capital cases in South Carolina. The researchers examined the number of questions asked of Black, white, male, and female potential jurors as well as the topics covered, each question’s complexity, and the parts of speech used. They found notable differences in the length, level of difficulty, and point of view of the questions prosecutors asked of Black potential jurors compared to white ones.
For example, prosecutors were more likely to ask Black potential jurors about their views on the death penalty, and they were more likely to ask them graphic questions about execution methods. After leading into the hoped-for response, prosecutors would then ask the potential juror if he or she would be willing to vote for a death sentence. When the response was no, the individual was struck from the jury pool.
Of the 17 cases examined in the study, a judge later ruled in six of them that a prosecutor illegally removed potential jurors because of their race. The researchers could correctly identify cases that violated Batson by looking at the combined natural language programming analyses for each case.
The next step in this research is to do similar studies on larger datasets with more diverse types of cases. Once the validity of the method is established, the analysis “could be done during jury selection almost in real time,” according to Martin Wells, professor of Statistical Sciences at Cornell and one of the authors of the study.
Natural language processing software can be a powerful tool to monitor jury selection or provide evidence for an appeal, especially for defendants who are possibly facing the death penalty.
Source: Cornell Chronicle
As a digital subscriber to Criminal Legal News, you can access full text and downloads for this and other premium content.
Already a subscriber? Login
