Election Forecasting

Finally, I can start obsessively checking FiveThirtyEight and its election forecasts! They released their specific model for the House elections on Friday. (This may or may not actually be a good thing, since I really shouldn't be obsessing over politics, but oh well.)

One of the things I've always liked about FiveThirtyEight is that they're extremely detailed about explaining where their forecasts come from and how they bring together and weight all the data they use. I see a lot of people that like to complain about election forecasts and polls, stating that they clearly know that such things are wrong, but then can't actually explain where that certainty comes from or how they know what they're seeing is wrong. Or, sometimes, I see people that can explain why they're so confident of their side's chances, but who still end up being overconfident because they haven't realized how limited their information actually is.

FiveThirtyEight hasn't generally had either of those problems. I don't always agree with everything they write, but there's no doubt that they have a superb track record - one would be wise to come with plenty of solid evidence if one thinks they know better than these forecasts.