A Formal-Epistemic Perspective on Statistical Evidence

Conclusion
This essay has argued that Enoch et al. greatly overestimate the verdict accuracy rate that statistical evidence could grant. Two reasons have been provided. First, that of base rates that are often disregarded or unknown. A particular instance of a low driver responsibility rate in road accidents has been provided to show that base rates can drive the verdict accuracy extremely low, and accordingly it can not be reasonably upheld that strong statistical evidence must grant high verdict accuracy rate. Secondly, it has been argued that compliance with the legal requirement of proving the defendant’s guilt beyond reasonable doubt in addition to the statistical dependence between a group of suspects permits findings of guilt in 0 to approximately 1/3 of all cases, depending on the exact percentile value that ‘’beyond reasonable doubt’’ is estimated to be. In the particular case discussed, the granted verdict accuracy rate is between 23-43%, depending on how many innocent persons have been convicted and guilty persons acquitted. Once again, the accuracy rate is very low. Hence it must be concluded that Enoch et al.’s assumption that statistical evidence falls under the legal System B, one that is more accurate than systems that heavily base their proceedings on epistemic considerations, is mistaken.
…