This article was originally posted on LinkedIn.

By Greg Srolestar, Director of Research and Technical Assistance

Daniel Kahneman, the Nobel-winning psychologist, once faced a difficult challenge: predicting who would become a great leader. As a young psychologist, he and a colleague designed an exercise to identify leadership potential in the military. They believed in it, used it with confidence, and made selections accordingly.

Then Kahneman did something critical—he tracked what happened to those young leaders. What he found was humbling: their predictions weren’t accurate. The exercise wasn’t actually identifying strong leaders.

What happened next is both strange and completely normal. Despite now knowing the exercise wasn’t predictive, he and his colleague continued using it. The process felt meaningful. Each time, it seemed like they were making sound judgments. Kahneman later described this as the “illusion of validity”—a confidence in our own judgments and processes, even when data shows otherwise.

Prosecutors face a similar challenge. Case-by-case decisions feel rigorous and thoughtful, but without stepping back to examine broader trends, it’s impossible to know whether those decisions are achieving the right outcomes. And even when data suggests a need for change, established ways of doing business can create their own illusion of validity, making it hard to shift course.

So what should prosecutors actually know about data?

1. Data won’t give you all the answers, but it will help you ask the right questions.

Numbers alone won’t tell you why your office is dismissing so many simple assault cases or why racial disparities exist in diversion offers—but they can highlight patterns that demand deeper inquiry. Once you ask the right questions, you can begin to see what the problems are and can generate ideas to improve processes and outcomes for you and your office.

2. Process feels like justice—but it can also be an illusion.

Lawyers are trained to believe in process. It’s what we rely on to ensure fairness. But sometimes, the process itself needs to be questioned. Data can help pinpoint where alternative approaches might yield better results.

For example, some jurisdictions have found that automatic diversion for certain low-level offenses—rather than requiring a prosecutor’s case-by-case approval—reduces racial disparities and improves outcomes. That insight only emerges by looking at the numbers.

3. Public safety isn’t just about sentencing—it’s about outcomes.

People increasingly expect prosecutors to consider the real-world impact of their decisions. Research has shown that even short jail stays can increase recidivism. Without tracking what happens after sentencing or diversion, it’s impossible to know whether a decision truly enhances public safety. Often that means looking at research or partnering with corrections since prosecutors’ offices don’t have that data on hand. But ultimately, learning about outcomes is essential for offices that want to protect public safety.

4. Don’t take the data personally.

The data itself can give you insight on what’s working and what’s not, but that doesn’t mean you or your unit are doing a bad job. Good ideas and good quality work don’t always pan out. It’s hard to remember now, but people used to think that scared straight programs would persuade at-risk young people to fly right. Only when scientists studied the programs did we learn they were counterproductive and left young people more likely to get into future trouble. Getting things wrong is inevitable; refusing to learn from it is not.

At the end of the day, most prosecutors didn’t go to law school to crunch numbers. As I’ve heard many times, “I became a lawyer so I wouldn’t have to do math.” But taking time to review data—regularly and with an open mind—ensures that our decisions are grounded in reality, not just in tradition. It helps us move beyond what feels right and toward what actually works.