Final (?) 2012 Election Comment

All the conservative criticism of the election polls last year emphasized the belief that Republicans were being under-sampled in the polls. Had that been true, Romney’s popular vote percentage should have been higher than his poll percentage, which was a mathematical possibility even despite his popular vote loss. So how do the numbers work out? Here’s a selection of the final pre-election results from 7 prominent polls.

2012 Election PollsThat’s not so good for the under-counting Republicans hypothesis. Romney came in pretty much right on target, while Obama actually increased a bit. Both are within the margin of error, but keep in mind that the margin of error is a bell-shaped curve, so the Romney “miss” of .8% is a much more likely than the Obama miss of 2.5%.

This doesn’t indicate a Democratic undersampling by the polls, though. The more plausible interpretation is that late-breaking independents went Democratic more than Republican. But it does indicate that the Republican undersampling hypothesis wasn’t just over-stated; it was dead wrong, 100% absolutely positively indisputably wrong. And for all the hand-wring, all the wailing and gnashing of teeth, and all the actual real-for-sure legitimate qualms about the weaknesses of survey research as a methodology, for predicting election outcomes the polls–at least the average of the polls–has consistently proven itself to be pretty damned good.


About James Hanley

James Hanley is Associate Professor of Political Science at Adrian College and a Fellow of the Institute for Social Policy and Understanding. The views expressed here do not reflect the views of either organization.
This entry was posted in Uncategorized. Bookmark the permalink.

4 Responses to Final (?) 2012 Election Comment

  1. Matty says:

    I would think the interesting comparison would be with an alternative method like prediction markets not with the views of pundits who are themselves ‘in the tank’ for one side or the other.

  2. AMW says:

    I believe prediction markets tend to be more precise than polls. I.e., similar average, lower variance. I don’t believe polls have a systematic bias. Well, maybe a particular polling group has a systematic bias; but if you take the average of a group of polls, the biases tend to cancel out.

  3. Troublesome Frog says:

    This reminds me of Tyler Cowen stating that pundits who make predictions for a living should have to make bets based on them. He called it a “tax on bullshit.” I think that the idea was that you’d have your bonus money tied up in an account and your employer would bet that money on your behalf. You’d state the odds and they would take one side of the bet or the other at random. Good analysists would get most of their money back. Bullshitters would lose substantial sums in the long run.

    Sounds like fun to me.

  4. Troublesome Frog says:

    Oops. It was Alex Tabarrok. My mistake. Awesome idea, anyway.

Comments are closed.