A Silver Lining

Ad by Fusion

Today, a little over half of America has a bit of a strut in its walk because of one guy: Nate Silver.

For those of you who weren't watching, Silver is a NYT statistician and author of Five Thirty Eight who gave Obama around 90% odds for winning a second term. He even bet Joe Scarborough $1000 (a bet that Joe ridiculed but didn't take). Gawker has a good summary.

Side-by-side electoral maps
Left: Silver's prediction; Right: Actual turnout. photo credit: @cosentino

Now we know that not only was Silver right, but his model nailed the results in all 50 states rendering him an overnight sensation. Silver is (half of) America's knight in silver (OK, I apologize) armor leading the charge against those evil political pundits who use nothing more than their awfully fallible gut feelings to make predictions. The verdict is in and the pitchforks are out. "Down with gut feelings! Long live logic! Long live the maths!"

There's only one problem with this new mantra: it's plain wrong.

I suspect a majority, say 90%, of the folks chanting this mantra haven't the slightest clue about how Silver's model works. Fortunately, there's a human-readable explanation of his methodology here (note: this one is for Senate forecasts which inherits much of its methodology from that for Presidential forecasts) as well as an informative piece on predictive models by Silver himself.

You don't have to go too far into either of these pieces to quickly realize that his model relies heavily on voter polls. In other words, Silver's model performs a statistical analysis on voters' presidential candidate preferences which — at least each side believes this about the other — are almost always based on gut feelings.

In my view, Silver's model isn't anti-gut. Quite to the contrary, it's very pro-gut. And if you've taken my repeated advice to read Gigerenzer's writings, you know that this is well worth celebrating.