Association for Public Policy Analysis & Management

We show how regression methods employing data on DNA exonerations and the number of convictions by state by year by race can be used to estimate a parameter that captures the combination of any racial differences in wrongful conviction rates for a given crime and any racial differences in the likelihood of a DNA exoneration among innocent convicts. We then show how we can back out the parameter of interest – namely the parameter that captures any difference in wrongful conviction rates by race – but only by making an assumption regarding the extent to which the likelihood of DNA exoneration among the innocent does or does not systematically differ by race.

The fundamental complication is that there is little direct evidence regarding whether the likelihood of DNA exoneration does or does not systematically differ across innocent convicts by race. However, our methodology allows us to consider various assumptions regarding possible racial biases in the DNA exoneration process and see how such differing assumptions impact the conclusions one would make regarding racial differences in wrongful conviction rates. With respect to rape, we show that under a relatively wide range of assumptions regarding possible racial biases in the DNA exoneration process (including no bias), our estimates suggest a significantly higher wrongful conviction rate among black convicts than white convicts. Namely, for one to conclude the wrongful conviction rate for rape is not significantly higher among black convicts than white convicts, the likelihood of a DNA exoneration among wrongfully convicted black convicts must be over fifty percent higher than it is among wrongfully convicted white convicts convicted in the same state and same year.

Interestingly, our results for murder differ substantially. Specifically, if one thinks the likelihood of a DNA exoneration among innocent defendants convicted in the same year in the same state is roughly similar across races, then our estimates suggest that the wrongful conviction rate for murder is roughly equal across races. For one to conclude the wrongful conviction rate for murder differs significantly across races, one would have to believe that the likelihood of a DNA exoneration among innocent defendants of one race is over fifty percent higher than it is among innocent defendants of the other race convicted in the same year and same state.

Given these results, one is then left with the question – is there a strong racial bias in the DNA exoneration process? As we have stated above, we have little direct evidence on this matter. However, we find time to exoneration longer among black exonerees than white exonerees, particularly among those exonerated for murder. We argue this would be quite unlikely if the DNA exoneration process somehow favored innocent black convicts over innocent white convicts.