Estimating the Prevalence of Wrongful Convictions

The purpose of this research is to conduct a follow-up to the Urban Institute's 2012 study entitled Post-Conviction DNA Testing and Wrongful Conviction funded through an NIJ contract: 2008F-08165. This study analyzed the results of post-conviction DNA testing of 715 defendants convicted of homicide or sexual assault in the Commonwealth of Virginia from 1973-1987.
In this study, the Urban Institute (UI) will collect court case data on cases involved in post-conviction DNA testing that were used in the study above. UI specifically seeks to determine: 1. The prevalence of four conviction-level testing outcomes identified in the first study (i.e., where the DNA evidence was either: inculpatory, exculpatory but insufficient for exoneration, exculpatory and supportive of exoneration, or indeterminate); 2. The case, victim, and convicted offender attributes that are correlated with these outcomes; 3. How the Virginia conviction rates compare to other states during the study time period; and 4. The utility of post-conviction DNA testing as a tool to detect wrongful convictions. UI will collect case, victim, and convicted offender attribute data from court records of 575 cases. This includes all cases with a determinate post-conviction DNA test and all indeterminate cases that are located at a courthouse with at least one determine case (356 indeterminate cases).
UI will run logistic regressions on data from all 575 cases to identify factors that are likely predicators of a case yielding a determinate DNA test. This analysis hopes to identify case characteristics that may point to the presence of definitive DNA evidence for jurisdictions seeking to engage in post-conviction DNA testing. In regards to the analysis on the prevalence of wrongful conviction, a nested logit analysis will be performed. The nested logit design will determine the likelihood that the DNA test would be favorable to the defendant, and if favorable, the likelihood that the DNA test provides exculpatory evidence that may be supportive of exoneration. This analysis builds on other NIJ funded research that suggests that certain types of cases may contain similar errors that lead to a wrongful conviction.
In an effort to enhance the utility of this analysis for jurisdictions outside the Commonwealth of Virginia, UI will attempt to determine how the Virginia rate of convictions during this timeframe compared to other states. Using the BJS Census of State Felony Courts from 1985, UI will test if Virginias conviction rate was significantly different from the average of state conviction rates at the time. This would allow UI to develop a confidence interval around any wrongful conviction estimate produced by this research. It will also provide a context for other jurisdictions seeking to understand the magnitude of the problem in their own state.
ca/ncf