Right. so the sample size is 12 million in and of itself. You don't copaire it from where the sample came from or else your numbers will get mixed.
In short all you have to remember is this "The majority of criminals are Christians."
Still don't agree. If you assume that any randomly-selected adult from the US population has an equal propensity of going to prison, then so long as you draw enough samples (12 million certainly is enough), then you would expect your prison population demographics to be identitical. But, at 80% versus 59%, it is not, therefore based on this simple argument, Christians have less propensity.
Agian, though, I do beleive there are other mitigating factors outside of declared religion that determine a population's likelihood for imprisionment.