Many theorists hold that outright verdicts based on bare statistical evidence are unwarranted. Bare statistical evidence may support high credence, on these views, but does not support outright belief or legal verdicts of culpability. The vignettes that constitute the lottery paradox and the proof paradox are marshalled to support this claim. Some theorists argue, furthermore, that examples of profiling also indicate that bare statistical evidence is insufficient for warranting outright verdicts.I examine Pritchard's and Buchak's treatments of these three kinds of (...) case. Pritchard argues that his safety condition explains the insufficiency of bare statistical evidence for outright verdicts in each of the three cases, while Buchak argues that her treatment of the distinction between credence and belief explains this. In these discussions the three kinds of cases – lottery, proof paradox, and profiling – are treated alike. The cases are taken to exhibit the same epistemic features. I identity significant overlooked epistemic differences amongst these three cases; these differences cast doubt on Pritchard's explanation of the insufficiency of bare statistical evidence for outright verdicts. Finally, I raise the question of whether we should aim for a unified explanation of the three paradoxes. (shrink)