Grandparent wants to want to compare P(found guilty|unattractive) to P(found guilty) to reason about whether attractiveness and the jury's finding are independe

Grandparent wants to want to compare P(found guilty|unattractive) to P(found guilty) to reason about whether attractiveness and the jury's finding are independent. You want to compare P(found guilty|unattractive,guilty) to P(found guilty|guilty) to see the same while taking into account the desirable bias against guilty people, which is of course more informative but has both components unknown. These arguments tend to crop up whenever someone notices dependence between effects desired to be independent. The argument is then about where the the dependence lies. The topic is always politically sensitive. I favor the grandparent's position, since it's the simpler one (and actually tractable). That doesn't mean it's right, and if someone's sufficiently convinced otherwise, they can attempt to estimate the guilt of various convicted (perhaps by waiting 50 years and looking for pardons). The results would have to be extraordinarily convincing, however, and quite possibly they would be harmful in that they would establish a genuine bias against unattractive people, which was the original problem. So this is a situation where the grandparent's position is better whether or not it's right.