Resources

  • SNITCHING: Criminal Informants and the Erosion of American Justice
  • U.S. Attorney General's Guidelines on the FBI's Use of Confidential Human Sources
  • Sarah Stillman, The Throwaways, The New Yorker (2012) (article on the use of juvenile informants)

Recent Blog Posts

Wednesday, February 14, 2018

Experiments on the impact of inducements

Scholars at the University of Arizona Law School have published a paper entitled "Incentives, Lies and Disclosure," 20 U. Pa. J. Con L. 33 (2018).  They conduct a number of behavioral experiments and conclude that incentivized witnesses like jailhouse informants are more likely to lie, and that even when potential jurors are told about the incentives, they still believe the witness.  Here is the abstract:  

Prosecutors can force witnesses to testify and use perjury prosecutions to hold them to the provable truth. More controversially, prosecutors also offer witnesses inducements for favorable testimony, including leniency, immunity, and even cash. This ubiquitous behavior would be illegal as witness bribery, except for a longstanding tradition of sovereigns using this power, which legal doctrine now reflects. A causal analysis shows that even if prosecutors use this power only in good faith, these inducements undermine the epistemic value of witness testimony. 

Due process requires, and legal doctrine assumes, that when such inducements are disclosed to the jury, they will discount the witness testimony accordingly. However, juries’ success in doing so is an empirical question. We conducted three randomized experiments with 1,000 human subjects in roles of witnesses and jurors deciding vignettes based on real cases. We find that incentives have large effects on witnesses, allowing prosecutors to routinely procure favorable testimony regardless of its truth. Yet, disclosure has no detectable effects on either witnesses or jurors. 

We discuss two potential reforms. First, courts could borrow from the practice with expert witnesses and use the current rules of evidence to conduct Daubert-like pretrial screening of incentivized witnesses for reliability. We frame the appropriate counterfactual question about whether the incentives would cause a witness to give the same testimony even if it were false. Second, we present the novel suggestion that prosecutors could decide whether to offer benefits to a witness based on whether she will testify to material information, but without knowing whether the information is favorable to the Government. These mechanisms may preserve the value of incentives to produce information, while minimizing false testimony.