After Romney gained about 3 points over the weekend, Obama made gains in today’s Rasmussen and Gallup approval tracking polls, suggesting that Romney’s bounce had peaked and was beginning to fade. At least until Pew interrupted.
The possibility that Romney’s bounce was fading began when the Rasmussen and Gallup approval trackers dropped the first day of interviews following the debates and found Obama making clear gains, suggesting that Obama performed much better on Sunday than he did immediately after the debates. And both also showed Obama returning to post-DNC levels, as Obama averaged a .7 point lead in Rasmussen after the DNC; today it’s tied. Similarly, Gallup showed Obama’s approval averaging plus-4.8 points following the DNC, but today Gallup shows Obama’s approval rating at plus-7.
A turn back toward Obama was also consistent with PPP's tweets that their samples on Saturday and Sunday showed Obama performing at or near pre-debate levels in Wisconsin and Virginia. And, in absolute terms, Obama fared well in state polls released on Sunday or Monday—including leads in Colorado, Iowa, and Virginia—according to Rasmussen, PPP, and Ann Seltzer. The Iowa and Colorado polls were particularly noteworthy: They were conducted yesterday and found Obama performing better than their prior poll in September.
But then there was Pew.
Right when it looked like Romney's bounce was small and sliding back, Pew came out with a new poll showing Romney ahead by 4 points among likely voters. The new data has conservatives celebrating and liberals in a panic, and understandably so. Throughout the entire cycle, Pew has had a strong Obama-lean, making the result particularly surprising. And Pew Research has irreproachable credentials, both in terms of past results and methodology, so this is a very important and powerful data point in the direction of a Romney bounce.
But the Pew poll swung 12 points in Romney's direction, from Obama plus-8 to Romney plus-4—that kind of a swing is not especially plausible. In fact, every other poll has shown a swing of 6 points or less since the first debate. That doesn't mean, though, that the new poll is necessarily out-of-whack. The first poll, and not the second, may have been the inaccurate one. But while it's not especially productive to dig into the weights to guess which poll is most representative (party-ID! party-ID!), it is important to recall that there can be outliers: remember the vaunted Ann Seltzer showing Obama leading by 13 in June? And although Pew's last poll was certainly good for Obama, it was not necessarily an outlier.
Analysts and horserace watchers face a choice between trying to reconcile Pew with today's trackers or waiting for more information. Although one could argue that Pew's Saturday and Sunday samples would have shown Obama surging back into the lead, it's hard to imagine that Pew found Romney ahead by 8 points on Thursday and Friday. Thus, in my view, the Pew poll is largely irreconcilable with the other information. The answer is probably somewhere in the middle, but whether Romney has gained 3 or 5 points makes a huge difference. Given the extent to which the data conflicts, the right choice is probably to wait a few days before drawing conclusions. By later this week, the Pew poll will probably appear as an outlier—or, it might be the first sign of a new race, with Romney in the lead.