All Intensive Purposes

Ian Ayres offers a testable hypothesis:

[A] disproportionate number of false confessions will occur when the confessing defendant is alleged to have a codefendant.

The logic for this derives directly from the game we all know:

The prisoner's dilema is at core a story of how prisoner's may be induced to confess -- irrespective of whether or not they are guilty. If you face a high enough prospect of conviction if an alleged codefendant finks, and if you will be given a lower sentence if you confess first, it can be entirely rational to falsely confess.

Ayres further suggests that the hypothesis finds some empirical support.