As everyone knows, to understand statistics and electoral polls, you need large testicles and a deep, manly voice. Just ask the Conservative Smear Machine. (More)
I have my own issues with New York Times‘ poll maven Nate Silver. Silver’s book The Signal and the Noise, discussed this month in Morning Feature, completely ignored the work of Tuphty MacMaibee. MacMaibee pioneered the field of probability in his quest to predict where best to stash nuts for the winter. Alas, Tuphty’s success allowed him to focus more on eating than on writing, and English mathematician Thomas Bayes watched Tuphty and did the writing. That’s why the most important equation in probability is called Bayes’ Theorem. Still, MacMaibee’s legacy endures in a common English word for something that might happen: maybe.
And that story is at least as true as the charges of bias being hurled at Nate Silver by conservatives in what Paul Krugman called “The War on Objectivity.” All of this is, of course, because Silver’s analytic model of polling data has not bent to the Republican myth of Mittmentum. That tale says Mitt Romney was so awesomely wonderful in the Denver debate that nothing else matters and he will ride an ever-rising tide of public approval right into the White House.
While Romney did see a bounce after Denver, just as President Obama had after the Democratic National Convention, in fact the race has settled back to where it was in late August. The national popular vote looks very close, but President Obama leads in enough state polls to be a clear favorite in the Electoral College. Silver is not alone in that forecast. Drew Linzer’s Votamatic yields similar numbers, as do the Princeton Election Consortium and the Real Clear Politics. Betting markets like Intrade and Betfair also show President Obama between a 3:2 and 3:1 favorite.
Politico‘s Dylan Byers ignores all of those other analysts giving similar forecasts, pretends Nate Silver is all alone in that prediction, quotes him saying “Romney, clearly, could still win,” and adds criticisms from David Brooks and Joe Scarborough. Byers never discusses (nor probably does he even understand) how aggregating polls reduces their margin of error. He cedes nothing to Electoral College math. There are national polls showing Romney leading, and others showing President Obama leading, and others showing a tie … so the race must be a tossup!
The London Telegraph‘s Tim Stanley takes it even further, dismantling the strawman claim that Silver and the New York Times are trying to steal the election from voters. Were Silver and the Times attempting that, Stanley’s argument might make sense.
But Silver is a data geek. His methodology feeds national and state polling data – every poll from every pollster – into an algorithm that weights them for recency and methodology, and adjusts for “house effects” (pollsters whose results routinely diverge from statistical means). His model also incorporates macro-economic data, although the influence of that data is now diminishing as he assumes voters have weighed it into their own decisions and thus it is “priced into” the polls themselves. Finally, he runs 10,000 mock elections, randomizing returns based on polling data and historical correlations between states’ results, and calculates state-by-state and overall win percentages.
Clearly, that’s not enough, as UnSkewedPolls’ Dean Chambers explains:
Nate Silver is a man of very small stature, a thin and effeminate man with a soft-sounding voice that sounds almost exactly like the “Mr. New Castrati” voice used by Rush Limbaugh on his program. In fact, Silver could easily be the poster child for the New Castrati in both image and sound.
Unskewedpolls argument: Nate Silver seems kinda gay + ??? = Romney landslide!
Again, Silver is not out on a limb with his forecast. Every other poll aggregation model shows President Obama a favorite, and most models show the president a wider favorite than Silver. The betting houses are also giving good odds to Romney backers, which raises the question: if conservatives so certain that Silver, Votamatic, the Princeton Election Consortium, and the odds makers are so wrong … why aren’t they betting on Romney?
Maybe they lack the testicular girth to put their money where their oh-so-manly mouths are.
Good day and good nuts.