Heading into Tuesday night's town hall presidential showdown, Romney appeared to be gaining a little separation in some key polls, pulling ahead by four points in both the Gallup likely voter and the latest Public Policy Polling polls, and holding steady with a two-point lead in Rasmussen.
On Wednesday, the Rasmussen poll slipped back to a one-point Romney lead, while Gallup jumped ahead to an unprecedented six-point Romney lead.
As Charles Cook at National Review observed, "These statistics are there to be broken, but it is worth pointing out for the record that, in the history of Gallup, no presidential candidate has ever been over 50 percent in mid-October and gone on to lose."
Bucking the trend was the IBD/TIPP tracking poll, which showed Obama edging back to a one-point lead on Tuesday and stretching that to two on Wednesday.
As with so many polls this year, the poll internals is where things really get interesting.
The Gallup poll's most dramatic internal indicator was an age bracket. In 2008 at this point, Obama led McCain 53 to 47 percent in the 30-49 year age bracket. In the current Gallup poll, Romney leads Obama with this group by a decisive 55 to 45 percent.
The IBD/TIPP poll sampled 37 percent Democrats to 30 percent Republicans, a D+7 that is closer to the 2008 returns than most observers think likely.
Even odder is that the IBD/TIPP results showed that 70 percent of "conservatives" were supporting Romney, but 20 percent were supporting Obama. The president, meanwhile, got the support of 89 percent of "liberals," giving up just 6 percent to Romney, a much more reasonable split.
The PPP poll showed 39 percent Democrats and 37 percent Republicans, but only 24 percent independents. The D+2 between the two parties is, if anything, generous to Romney by a point or two. But the independents appear to be underrepresented, and they tilted to Romney by 14 points.
Markos Moulitsas, proprietor of the Daily Kos, was clearly uncomfortable with the PPP results, and focused on the Sunday numbers, which skewed sharply in Romney's favor.
"That Sunday sample, about a quarter of the total, was entirely responsible for Romney’s favorable numbers. That’s why the good pollsters collect data over multiple days, to smooth out such irregularities. And at 400 respondents (or so), Sunday had a single-day MoE of 4.9 percent. Lots of polls float around with worse. On the other hand, Saturday’s sample MoE was 3.92 percent, while Friday’s was 3.97 percent. And with no external news even suggesting the big Sunday collapse, it certainly smells like an outlier."
You wouldn't know from this passage that the Daily Kos co-sponsored the poll, or that PPP itself is a Democratic polling outfit.
Eric Schulzke writes on national politics for the Deseret News. He can be contacted at email@example.com.