Nate Silver/538 Presidential Now Projections
It seems like every day I turn news on these days (which, I'll admit, is quite frequently), the newscasters always try to make it seem as though the world is coming to an end. With our election cycle coming up, I don't think that's likely to change anytime soon. And while there are many important elections to pay attention to, this post will pay attention to the Big Daddy of them all, the Presidential election.
Like it or not, the election this year features two of the most un-liked candidates in memory and has turned the election season into non-stop circus. Somehow, that has gotten people to care about politics a little more than usual, which, in turn, has gotten people to care quite a bit about polls.
Now, people tend to talk to me quite a bit about polls. Maybe it was because I made the mistake of 'lecturing' some folks about the realities of polling several months ago (I promise, it was with good intention!), maybe it's because I live and breathe statistics. Whatever it is, I've gotten a lot of comments from people about some polls from last week after the RNC that yielded projections such as the following:
Now I get why this is a sensational simulation but allow me to put the brakes on before you get too worked up. First, this simulation is based on if an election were to be held today. Fun, but ultimately meaningless. Second, this is Nate Silver 538's. Perhaps, the Internet's favorite statistician. Unfortunately, while I'll admit he had some success early on his career with political forecasts, I feel he's been a little weak as of late. The results just don't seem to be there and I'm not for the 'secret sauce' methodology that he/they apply behind closed doors.
Enter my alternative: Sam Wang from Princeton University. Sam has been doing some political forecasts for quite a few years now and in my opinion, has quite a bit of success. It's hard to go a detailed breakdown of his methods (though they are available freely at his website) but I will say that I truly feel that he has the right ways to strategies to predict these complex events.
To preface his current simulations, Sam correctly predicted 3 of the 4 spots on the Democratic and Republican tickets last year (the only one he missed was the Republican VP). He pretty much was one of the first people to predict Donald Trump because his models reflect statistical analysis, not pundit magic.
And his current predictions? A 85% Bayesian win probability and a 65% random drift win probability for Clinton in November (Bayesian is a probability based on the events of that have happened so far and random drift is based on where the nation could drift by November).
Definitely would recommend checking out his blog at the link below if you are interested in politics or statistics or both!