Tuesday, November 1, 2016

The amazing way to predict this election

I'm going to make a decidedly cautious prediction for the 2016 election: the candidates currently ahead in the polling averages are going to win.

While I could do my own polling average, there is no need since RealClearPolitics has been putting together a great one for years. I predict Clinton will win the presidency with 304 electoral votes, and Democrats will end up with 49 Senate seats. So when I get over 90% of states right with this prediction, I want everyone to say how amazing my skills are like they have done for other election prognosticators that claim to use far more complex analytical models.

This post is not meant to simply knock places like 538 for repacking readily available data in a more interesting format. It's mainly to point out how great polling is. There is a reason polling is a multi-million dollar industry full of highly respected organizations and used by so many different industries. There is a reason every political campaign uses polling extensively.

Don't get me wrong -- places like 538 and The Upshot are going to be very accurate next week because polling is so accurate. It will be basically impossible to know if their more complex analyses are actually better because they will diverge so little from the basic polling averages. For example, in both 2008 and 2012 a basic polling average correctly predicted the presidential outcome in 49 out of 50 states.

I bring this up because I think it is relevant to futurism and all predictions. We shouldn't be focused on how the predictions are presented. We should focus on the real sources behind them.

No comments:

Post a Comment