07 November 2012

Election post-mortem: How I was wrong (and why)

Let me be clear upfront: I was wrong, and I have no problem admitting it.

On Friday, I projected that Mitt Romney would defeat President Obama on the strength of Republican turnout and relative strength among independents. I based this projection upon the estimate that the +7D turnout of 2008 (39D/32R/29I) would look more like +2D/+3D.

In fact, the exit polls from last night showed 38D/32R/29I, or +6D (not sure where that extra percent went!). This seems to suggest that 1) the Republican edge in enthusiasm didn't transform the electorate enough; and conversely, that 2) the relatively less enthusiasm among Democrats didn't reduce their share of the vote as much as I expected.

Among self-described Independents, Romney won 50-45. Indeed, polls have suggested a high single digit lead among such voters, and here the projection was pretty close. But it couldn't be overcome by the relatively higher turnout than expected among Democrats.

Nate Silver will get well-deserved props for his projections, but the much-maligned pollsters (especially from those on the right) deserve props, too. My case for Romney rested on the assumption that polls were oversampling Democrats; and they were, but just ever so slightly. As a result, I thought the numbers going into Nate's model were not accurate and thus likely to edge the model toward the President.

I have tried to make it clear that the race was very close (and indeed, by measure of the popular vote, it was much closer than 2008). A few points here or there may have made a significant difference. Indeed, as I stressed on Friday, it doesn't now surprise me that Obama was able to exceed 300 EVs.

More to come on other election-related issues...
Post a Comment