Sunday, November 23, 2008

Nate Silver is Still a Genius ...

... even though the election is (mostly) over.

What makes 538.com so cool is not the political analysis. There are many others (this brilliant blog included) who do that as well or better. What makes 538.com so good is the interesting and inventive ways Nate brings statistical methods to the places where the otherwise widely reported data ends.

Today's piece is a great example of what I'm talking about. Many are watching the Minnesota Senate recount with interest and know that Norm Coleman's lead has narrowed by some amount (the exact amount is in dipute but the Star-Tribune has as decent a count as any and they have Coleman's lead at 180 as of this writing). But the problem with the running tallies is that a) knowing the total doesn't tell you anything too useful without know which precincts have been recounted or not and, more importantly, b) a lot of ballots are being challenged by the rival campaigns and those votes a candidate would have had are then deducted from a candidate's total. They will be added back in or not after the challenge is reviewed by the statewide canvassing board later on.

So, who's winning? From the data that has simply been reported as is, it seems like Franken is not narrowing Coleman's lead rapidly enough to move ahead. But is there a systematic way in which the challenges are understating one or the other candidate's support? Nate Silver does some fancy schmancy statistical analysis and finds ... yes. As a result, he rates Franken a "very slight favorite" to win in the end.

And that's pretty cool stuff.

1 comment:

Recovering political scientist said...

While I generally agree that Nate is a genius, I think the confidence intervals on the 27 votes are probably pretty large. Hard to say without seeing the data...

I agree that Franken may be a slight favorite, but it is not a done deal (which Nate wouldn't argue)...