And—at least in a low-resource environment where the cost of a false positive isn’t high (like a boardgame club) -- it’s not inappropriate to consider base rates. Even if you think the base rate of truthful accusations made to boardgame clubs is only (say) 70 percent, that’s still a thumb on the scale.
Think about the decisions of parking-enforcement hearing officers as an analogy. They start with the assumption that the base rate of parking enforcers messing up is lower than the base rate of people lying to get out of parking tickets. It’d be very difficult to run a parking enforcement program otherwise.
The other uncomfortable truth is that kicking probably innocent people out is sometimes unavoidable. For instance, in the context of classified information, you would absolutely revoke someone’s clearance (which is career ending) if you concluded there was a 1 percent chance that they were passing major secrets to the enemy and could not improve that estimate. Better to end 100 CIA agents’ careers than to allow an Ames or Hanssen type traitor to operate. Kicking someone out of a boardgame club on less than 50 percent likelihood of sexual assault is sensible, because the costs of a false negative are worse than the costs of a false positive.
And—at least in a low-resource environment where the cost of a false positive isn’t high (like a boardgame club) -- it’s not inappropriate to consider base rates. Even if you think the base rate of truthful accusations made to boardgame clubs is only (say) 70 percent, that’s still a thumb on the scale.
Think about the decisions of parking-enforcement hearing officers as an analogy. They start with the assumption that the base rate of parking enforcers messing up is lower than the base rate of people lying to get out of parking tickets. It’d be very difficult to run a parking enforcement program otherwise.
The other uncomfortable truth is that kicking probably innocent people out is sometimes unavoidable. For instance, in the context of classified information, you would absolutely revoke someone’s clearance (which is career ending) if you concluded there was a 1 percent chance that they were passing major secrets to the enemy and could not improve that estimate. Better to end 100 CIA agents’ careers than to allow an Ames or Hanssen type traitor to operate. Kicking someone out of a boardgame club on less than 50 percent likelihood of sexual assault is sensible, because the costs of a false negative are worse than the costs of a false positive.