Estimating the Probability of False Conviction for Criminal Defendants Who are Sentenced to Death

Abstract

This study is addressed to one of the great unknowns of criminal justice: the frequency of convictions of innocent defendants. A general estimate of the rate of false conviction may be beyond reach, but the extraordinarily high number of exonerations among defendants who have been sentenced to death makes it possible to estimate a lower bound for the rate of false conviction among death sentences. Previous studies have shown that 2.3% of all defendants sentenced to death in the United States between 1973 and 1989 (Gross & O’Brien 2008), and 3.3% of defendants sentenced to death for rape murder from 1982 through 1989 (Risinger 2007), were exonerated and released because of strong post-trial evidence of innocence. These are underestimates of the rate of false conviction because some innocent death row prisoners are never identified and exonerated. This study focuses on those death-sentenced defendants who remain under threat of execution, as opposed to the large proportion who are removed from death row and resentenced to life imprisonment. Our hypothesis is that only those capital defendants who might still be put to death receive the extraordinary level of attention and scrutiny that makes it plausible to use the rate of exoneration as a proxy for the rate of false conviction at trial. Using survival analysis, we: (1) demonstrate that removing the threat of execution does in fact decrease the probability of exoneration for a death sentenced inmate by a factor of about 8; and (2) estimate that the cumulative rate of exoneration for those defendants sentenced to death in the United States between 1973 and 2004 who remained under threat of execution for at least 21 years is 4%. This too, however, is very likely an underestimate. There are strong theoretical and anecdotal reasons to believe that death sentenced defendants who might be innocent are more likely than other death row inmates to be removed from death row without being exonerated – at which point the probability of exoneration drops precipitously. If so, the population of defendants who remain on death row for long periods of time, or who are executed, includes a smaller proportion of innocent defendants than the entire population of death sentenced defendants, and the true rate of false convictions among all death sentenced defendants in the United States must be considerably higher than 4%.