Are confessions enough evidence to sentence a defendant?

Resumo

Confession evidence as a burden of proof has been a source of controversy for the last decades, being continuously questioned as sufficient evidence and by the methods to obtain it. Laboratory research has recently been analysed (Stewart, Woody, & Pulos, 2018), but the results are not valid as the effect sizes have not been weighted. As for this, a new search was made in the scientific databases of reference, Web of Science and Scopus. A total of 17 primary studies were found obtaining 22 effect sizes for a total of 1,704 participants. Effect sizes were computed with Cohen’s h (differences between proportions: accepting false confession vs. not accepting) for one-sample. The results showed a non-significant effect size, h = -0.0077, 95% CI [-0.864, 0.102], non-generalizable, 80% CV [-2.55, 2.57], and mediated by moderators, %Var = 5.05. Succinctly, the probability of accepting a false confession is the same of refusing it (50%). Although these are laboratory results and, for then, with face validity for real context, they are enough to establish that confessions should not have probative value per se, as they infringe the principle of presumption of innocence.