Per John’s request, I am posting my comment to his post “A logic for rational voting” that discussed a paper by Bryan Caplan and others on a model for voter altruism.

My first impression is that there ought to be a constant tacked on to the formula for expected benefits from voting, to account for benefits that aren’t affected by the probability of casting a deciding vote (like feeling good about carrying your civic duty, etc.). Also, I was confused for a moment before realizing that B-social refers to per capita social benefits (hence the need to multiply by N).

Is it possible for people to make errors in either direction with regard to low probability events? Or is the evidence that they discount them too often? Seems like sometimes we are overly scared of events with extremely low probability (terrorist attacks vs. car accidents).