When policymakers and courts consider which criminal defendants to release before trial, they base decisions on incomplete information—often a handful of signals based on a defendant’s offense, past record, and flight risk. In “Making Sense of Mixed Signals: An Experimental Analysis of Case Factors and Pretrial Decision-making,” SPA Assistant Professor Stacie St. Louis and coauthor Nick Petersen (University of Miami) adapt signaling theory to test how the public synthesizes these conflicting factors into pretrial decisions.

“We realized that almost every study on pretrial decision making looks at factors like offense type and criminal history independently,” said St. Louis. “But all of these characteristics . . . are actually being considered as one portfolio, as one case.”

In a national online survey of 1,092 U.S. adults, they presented respondents with a randomly assigned experimental vignette, varying: (1) offense (robbery vs identity theft), (2) prior arrests (multiple vs none), and (3) flight risk (high vs low). Their analysis, using signaling theory, suggests that reforms built around single factors cannot account for the messiness of real cases.

When signals conflict, perceived risk and preferred pretrial responses often converge in ways that undercut a simple offense‑first approach. Offense type, flight risk, and prior record each significantly predict harsher pretrial preferences. Robbery results in higher odds of bail denial and substantially larger bail amounts than identity theft; high flight risk and multiple priors also raise denial odds and bail.

However, in vignettes pitting a one-time serious violent robbery against repeated nonviolent offending, the public did not consistently assign harsher pretrial conditions to those with violent charges. A severe charge can be muted by a clean history and low flight risk, while a less severe offense paired with a long record and high flight risk can lead to harsher judgments. For example, identity‑theft defendants with multiple priors and high flight risk received supervision choices and bail amounts ($16,371) statistically similar to robbery defendants who lack other negative signals ($10,962), with little differences in type of pretrial supervision or bail amount assigned.

 “Few cases operate at the extremes,” St. Louis said. “The fact that a new offender with a low flight risk can temper the negative reflected signal of a violent robbery is really fascinating.”

Crucially, many reforms and public debates center on offense categories. New York’s high‑profile bail changes, St. Louis noted, initially relied heavily on offense type and drew backlash that led to partial rollbacks. “The initial reform had blanket guidelines based on offense type, largely overlooking criminal history and all the nuance related to other factors. Some of those rollbacks added these factors back into the mix.”

She urged policymakers to heed both the evidence and public sentiment. “As we engage in these reforms, we must collect good data and [evaluate whether] 1) the public is still supportive of this, and 2) these reforms achieving their intended aims,” she said. Courts and reformers should avoid blanket exceptions or reforms based only on offense type, and consider nuance by including the perspectives of stakeholders, researchers, practitioners, and the public.

St. Louis, who is currently completing a meta‑analysis of pretrial decision studies, hopes scholars will expand case‑level work that connects public opinion, practicing court actors, and observed judicial behavior. “With that evidence base together, we could create guides for court actors and policy briefs for lawmakers, all to try to find that good balance point.” As founder and director of AU’s Research & Innovation on Pretrial Processes & Legal Experiences (RIPPLE), she and her team of research assistants study decision-making, jail and court operations, and their impacts on individuals and communities and prepare policy briefs.

The practical stakes are high: whether defendants are monitored, detained, or released for a year or more can hinge on initial pretrial judgments made in the fog.

“There's so much nuance,” St. Louis repeated. “You are a whole person. You are a whole defendant.”