Wednesday morning after the election, it’s clear to me that there’s going to be a lot of post-mortem criticisms of where the commercial popular polling went wrong this election year. Some of the criticisms will be, shall we say, a bit heated, but I am more interested in the criticisms that look pretty specifically at polling methodology and analysis and where these two factors may have led pollsters astray.

The first place to I’ll be looking is Prof. Sam Wang’s “Electoral College Meta-Analysis” site. Prof. Wang, who’s a molecular biologist at Princeton University, took up poll analysis as (I guess) a sort of hobby, and predicted an electoral-college landslide for Kerry, based on aggregating and analyzing poll results on an ongoing basis. What makes Prof. Wang’s page refreshing even in the aftermath of an incorrect prediction is his very specific admissions about where his model seems to have gone wrong. That’s the mark of a good scientist — when your theoretical model goes wrong, acknowledge it quickly and pick up the pieces.

At the same time, I think it’s important to stress that polling, even though a flawed means of learning collective opinions, is arguably pretty important in a democratic society, because it can indicate if the election process itself has stumbled in some way — for example, it was exit polls that indicated something was problematic about the balloting process in Florida in 2000.

—–