I was concerned about sample bias. However, I found this paper.
In Virginia, a cohort of 634 cases of sexual assault and/or homicide dating from 1973 to 1987 was discovered to have retained physical evidence. Since most state legislation that requires evidence storage was enacted in the post-DNA era...the Virginia cases provides a unique opportunity to determine how often DNA testing can be used to identify wrongful convictions. The results can be generalized (with caveats) because the physical evidence was retained for reasons unrelated to the case outcome, and the cases were assigned to the serologist who retained the evidence in a way that did not introduce bias.
They found DNA evidence that was “supportive of exoneration” (versus “inculpative” or “exonerative but insufficient”) in 8% of sexual assault convictions (33 in 422) . However, many of the cases provided no determinate evidence, (which means something like too little DNA or no reference DNA). Among cases that provided determinate evidence, 15% (33 in 227) seemed supportive of exoneration.
They present quite a bit of other statistical analysis, but this isn’t quite my skill set and I’ll need to set aside time to go through it later. I didn’t find any other data sources of similar quality, but I looked for <30min. There was a more recent project in Arizona that tried to apply DNA testing more broadly, but that was aimed at doing a cost/benefit for current testing.
Are those two sets of 33 the same 33? If so, then that deserves calling out, because at first reading (at least to me) it sounds like some of the 8% are only “supportive of exoneration” because of a lack of DNA evidence.
Would the natural experiment of DNA evidence furnish the false positive rate you are interested in?
Yes.
I was concerned about sample bias. However, I found this paper.
They found DNA evidence that was “supportive of exoneration” (versus “inculpative” or “exonerative but insufficient”) in 8% of sexual assault convictions (33 in 422) . However, many of the cases provided no determinate evidence, (which means something like too little DNA or no reference DNA). Among cases that provided determinate evidence, 15% (33 in 227) seemed supportive of exoneration.
They present quite a bit of other statistical analysis, but this isn’t quite my skill set and I’ll need to set aside time to go through it later. I didn’t find any other data sources of similar quality, but I looked for <30min. There was a more recent project in Arizona that tried to apply DNA testing more broadly, but that was aimed at doing a cost/benefit for current testing.
Are those two sets of 33 the same 33? If so, then that deserves calling out, because at first reading (at least to me) it sounds like some of the 8% are only “supportive of exoneration” because of a lack of DNA evidence.