ELECTIONATE JUNE 20, 2012
-
Read Later
READ LATERAvailable only to subscribers. SUBSCRIBE TODAY
-
Listen
ARTICLE AUDIO
- Font Size

This morning, Bloomberg News released a poll that sent Twitter into a frenzy, showing Obama up by 13 percentage points, 53-40.
While these results might be promptly dispatched as an outlier, the Bloomberg poll was conducted by Ann Selzer & Co. of Iowa, an exceptionally well regarded pollster best known for conducting the famous Des Moines Register poll prior to the Iowa caucuses. Adding to the intrigue, Obama’s 13 point lead was among likely voters, even though other polls of likely voters have tended to show a tied race, or even a slight Romney advantage.
The internals show that Selzer’s sample is properly weighted by age and race, but the poll does not state the composition of the likely voter universe. Partisan affiliation doesn’t explain Obama’s big lead, as the poll shows Democrats with a 6 point partisan advantage, 39-33. That suggests that either the likely voter screen yields a more Democratic electorate, which seems unlikely given higher GOP enthusiasm, or that Obama holds a surprisingly large advantage among voters who are “totally independent,” a finding largely at odds with recent polling.
Of course, even the best pollsters occasionally produce odd samples. The last Selzer poll showed a tied race, even though Obama led by 5 points in the RCP average. The prior pro-Romney result allows us to dismiss, at least for now, the possibility that Selzer’s vaunted methods produce a pro-Obama house effect. It would be quite interesting if a highly regarded pollster like Ann Selzer consistently showed results that tilted toward Obama, but one poll is insufficient to make such an assertion.
Rather than allow one’s hopes to rise and fall with every poll, the best metric for following the horse race is the simple but accurate polling average. Some pollsters produce polls that lean right, and others lean left. Odd samples produce outlying polls in either direction. The polling average obscures the static and provides a rough gauge of the state of the race, which has been remarkably stable since April. On average, Obama leads by about 4 percent in non-Gallup polls of registered voters, and Romney leads by a hair among likely voters. Perk up when the averages change, not when occasional polls punctuate the race with blaring headlines about decisive movement toward either candidate.
3 comments
It's just a hunch from my gut, but I'll say the President wins by a little or loses by a lot. The former because Romney is such an uninspiring but kinda scary twerp that Obama muddles through on a LOTE (lesser of two evils) basis---especially if Romney really does pick his doppeltwerp Pawlenty as VP for a big initiative in the Midwest to offset Obama's Latino-fest in the Southwest. Uh-huh. But the latter outcome is plausible if the economy takes a deep downturn as some troubling indicators suggest is possible, in part courtesy of those Euro-goofs in the EU. It's all guesswork and occasional weird polls for now, of course, but since not nobody campaigns like the President, my money is still on a second term.
- SteveJudd
June 21, 2012 at 12:38am
The "twerp factor" may come to play a big role, indeed. In the debates, for example.
- ironyroad
June 21, 2012 at 10:40am
"doppeltwerp" - great term, a gem of condensed expression. Fits Paw-Paw to a T, too.
- Haole45
June 21, 2012 at 2:46pm