The election was, of course, sweet vindication for the New York Times blogger Nate Silver. He called all 50 states in the presidential race correctly, struck a blow for math, and made Joe Scarborough look more than a little silly.
But today, at a lunchtime talk at Brown University's Taubman Center for Public Policy & American Institutions, another numbers man - Michael Dimock, associate director of research for the Pew Research Center - laid out a compelling two-part critique of Silver and other poll aggregators like Simon Jackman of the Huffington Post.
First, Dimock said, they build a false confidence in the accuracy of public opinion surveys. They have margins of error. They have flaws that are not erased even in aggregation. And to suggest that Obama's numbers improved by 0.3 points on a given day is to make an almost meaningless assertion.
Second, Dimock said, the aggregators' intense focus on the horse race obscures the real value of polling. Yes, we all care about the candidates' shifting standing with voters. But a quality poll gives us glimpses into the whys. Why is President Obama's support holding up in a poor economy? Why is Mitt Romney failing to transform a damaged GOP brand? These insights were lost in the aggregation, Dimock suggested.
Of course, "quants" like Silver and Jackman understand the limits of polling as well as anyone. But their deployment of survey data and the way we consume it, Dimock suggested, leaves something to be desired.
Dimock made his comments about Silver during the question-and-answer session following a broader talk on polling and the 2012 election. Below, some other highlights from his appearance.
The Turnout Surge
If the big story of the election was the shifting composition of the electorate - we've all read plenty of stories about the growing Latino vote - Dimock spoke to another phenomenon of interest: a narrowing of the traditional turnout gap between white and non-white, between old and young. In other words, it's not just that the country is growing more diverse. Latinos and young voters are more likely to vote than they once were.
This phenomenon was first evident in the 2008 election. And, as Dimock noted, some dismissed it as an abberation - chalked it up to Barack Obama's historic candidacy. But the numbers held up in 2012. And this, Dimock said, suggests something more.
When I pressed him on the point - where is the evidence that the trend will hold up post-Obama? - he allowed that it is just supposition at the moment. Indeed, the 2010 election - which saw a reversion to typical voting patterns - could be seen as evidence for the Obama effect, he said. The president, after all, wasn't on the ballot that year. But Dimock said he was skeptical of the argument. Mid-terms are much different elections, he said, with much smaller electorates.
And after summer polling suggested the Latino and youth vote in November 2012 would not match the 2008 turnout, he said, there was something genuinely surprising about the fact that it did - something that suggested more than mere loyalty to this particular president. We'll see in four years.
Some interesting odds and ends from the talk: