Adams, Ian T.

Interactive Analysis

The Risk of False Confession Driven
Wrongful Convictions

A Bayesian framework for estimating how often lawful interrogation tactics contribute to wrongful convictions of the innocent.

Based on Scott M. Mourtgos and Ian T. Adams, forthcoming at Journal of Criminal Justice

Key Finding from the Research

Under most reasonable assumptions, the estimated risk of a false confession leading to wrongful conviction clusters around 1%. At the traditional Blackstone standard (λ=10), the maximum acceptable risk is about 9%. This suggests common interrogation tactics may fall within acceptable bounds—but your conclusion depends on your assumptions and choice of λ.

Estimated Wrongful Conviction Risk

0.9%

With these assumptions, fewer than 1 in 100 confessions obtained using this tactic would be false and lead to wrongful conviction.

90% confidence interval: 0.7% – 1.2%

Within Acceptable Range

Your threshold (λ=10): 9.1%

Adjust Your Assumptions

True Confession RateWhat percentage of confessions are from truly guilty people? Higher values mean confessions are more reliable indicators of guilt.95%

80% (more false confessions)99% (very reliable)

Tactic Effectiveness (Guilty)How effective is the tactic at getting guilty people to confess? Higher values mean the tactic is more effective at eliciting true confessions.83%

50% (less effective)99% (highly effective)

Innocent Protection RateHow well does the tactic protect innocent people from falsely confessing? Higher values mean innocent people are less likely to falsely confess under this tactic.85%

50% (risky for innocents)99% (protects innocents)

What's Your Trade-off? (λ)How many guilty people going free equals the harm of one innocent person wrongly convicted? Higher values mean you place more weight on protecting the innocent.10

λ represents how many guilty people going free equals the harm of one wrongful conviction.

1Volokh (5)Blackstone (10)Fortescue (20)50

Presets from Literature

Uses the paper's default assumptions: 95% of confessions are true, tactics work 83% of the time on guilty suspects, and 85% of innocent people resist false confession.More skeptical assumptions: assumes a higher false confession rate (8% vs 5%) and somewhat lower protection for innocent suspects.Worst-case assumptions: 12% false confession rate, highly effective tactics that also pose greater risk to innocent suspects (only 70% resist).

What Your Choices Mean

With these assumptions, fewer than 1 in 100 confessions obtained using this tactic would be false and lead to wrongful conviction.

This is 8.1 percentage points below your chosen acceptability threshold of 9.1%.

Low Est.

0.7%

Middle

0.9%

High Est.

1.2%