3 ms·
It makes a nice headline to reduce the prediction to "a 54% chance vs an 84% chance," but most of these sites are predicting not just the statistical probabilit
by pwinnski 10y ago
It makes a nice headline to reduce the prediction to "a 54% chance vs an 84% chance," but most of these sites are predicting not just the statistical probability of a future event in "happens or doesn't" form, they're predicting the relative vote percentage for each candidate in each state.
So if Trump beats Clinton, that really embarrassing for 538, but if she wins the popular vote by less than 6.3%, that's also slightly off. If he wins Florida, that's slightly less off, but still a miss for 538, and so on.
IMO, the only way to really know which model is closer to reality is to wait until November 9 and find out. That's not helpful for deciding who to pay attention to in 2010, though.