Maiello: Defeat the Press
Wolraich: Obama at the Gates of... Gates
Time is running out on Mitt Romney's campaign for President, which seems to have roared back into a competitive posture, but not the lead its cheerleaders in the media would wish. Here's what the polls showed in another day that provided some good news for Obama, and no more good news for Romney.
Virginia Showing Trend Away From Romney
The only real news in the polling today was that a pair of polls showed President Obama even in Virginia (unreliable Gravis Marketing, 48/48), and +4 (WaPo, 51/47). Because Virginia is an absolute must-win for Romney, these polls are bad news for him. Gravis was last in the field in Virginia on September 8-9, when it found Romney up 5, when everyone else thought the President was ahead. The Post had last surveyed Virginia on September 12-16, in the heart of Obama's post convention bounce, finding him up 8, the strongest result in the fall season there for him. Gravis does tend to have a Republican lean, but its polls do bounce around unpredictably, so standing alone, its conclusion that Virginia is tied is not that useful.
The reason the Romney campaign should be very worried is that the polls taken since President Obama's decisive win in the third Presidential debate average to a tie, and trending toward better numbers, as one sees from listing them oldest to newest: FOX (Romney +2), Purple Strategies (tied), Rasmussen (Romney +2), WaPo (Obama +4), Gravis (tied). Virginia back in the margin of error means a critical Romney state is now a coin flip. This is consistent with my prediction after the third debate that it would be followed by a bit of movement back to Obama.
Yesterday, I noted the disjunction between conservative media attempting to spin that Romney is ahead or "surging" and that the President's campaign is faltering, failing, or recognizing that it is failing. Virginia is a stick in the eye to those folks. For example, Jay Cost of RealClearPolitics wrote on Saturday that Romney has a clear lead in states comprising 261 EVs, and included in that group Virginia. Of course, Mitt Romney could win Virginia. But he hasn't led there most of the year, has probably never led by more than a point or two, and may not lead today. Likewise, the head of Suffolk University's "Political Research Center," David Paleologos, announced on Bill O'Reilly's The Factor back on October 9 that his group would no longer poll Virginia because it had "painted [it] red" -- Romney couldn't lose it. The unscientific lack of curiosity Suffolk and Paleologos exhibited is characteristic of the degeneration of poll data (at least for the incurious and the intellectually dishonest) into advocacy. It's also quite stupid -- President Obama led Virginia outside the margins of error of many polls, but Suffolk didn't stop polling then. Mitt Romney storming into a *statistical tie* is only reason to look away and stop learning in the world of denialists who behave like political advocates rather than political scientists.
Other State Polls Suggest No Real Movement, Belie Romney's Supposed National Lead
The only polls of states that are reasonably close were of Minnesota, where St. Cloud State put President Obama +8 (53/45), and North Carolina, where Rasmussen saw a healthy +6 for Mitt Romney (52/46). The St. Cloud number is a good one for Obama, as Minnesota voted in 2008 almost exactly like neighboring Wisconsin and Iowa, and the +8 represents only a 2 point Obama decline from 2008. Again, it is impossible to square polls like that one with a world in which the President trails by 5, the Gallup fantasy.
The Rasmussen number in North Carolina is a good one for Romney, to be sure, and forms part of an average of polls putting Romney up 3 in North Carolina. That, again, cannot be squared with the Gallup and Rasmussen national numbers, as a national Romney lead of 4 or 6 should result in a double-digit lead in North Carolina. Indeed, Rasmussen's own sampling of swing states cannot be squared with the +6 result claimed for North Carolina. President Obama won every swing state Rasmussen polls as such by more than he won North Carolina. Rasmussen is making the ridiculous claim that all swing states together (CO, FL, IA, MI, NV, NH, NC, OH, PA, VA, WI) are +6 for Romney, but that Romney is +6 in North Carolina. This makes no sense, and cannot possibly be true. (Never mind that the average of Rasmussen's state polls of these states is almost exactly even -- North Carolina's +6 is better than any number Romney has achieved in the great majority of these states all year.
National Polls Stay Silly, Largely Irrelevant
Gallup continues to find Romney decisively ahead (51/46), which simply contradicts all state polling one sees. If so (taking Obama's 2008 margins and subtracting the 12 point difference), Romney should lead by 2-3 in MN, not trail by 8, and should lead in NC by 12, not 6 (or 1 point, the margin a Republican pollster leaning Republican reported yesterday). The same point applies to Rasmussen, which sits at 50/46, and applies in reverse to RAND, which now shows Obama up 51/45. It is comical to see poll denialists like Jay Cost or the right's Nate Silver-lite, Senor Numbersmuncher, who argue that Gallup and Rasmussen's party ID numbers mean Romney leads and will win. They never cross-apply this wisdom to the state polls that are all of necessity (in that worldview0 very badly wrong, nor can they. We will see on November 6, and some of these polling services will be humiliated, unless they conveniently break back to a place the state polls reflect just before November 6. There is still time for Rasmussen's frame to move toward a more Democratic composition, and call the election a jump ball, then to claim that it was close enough either way.
Meanwhile, non-silly national polling by IBD/TIPP showed President Obama up 2.3 points, essentially unchanged, and WaPo (which showed Romney +1 on Friday) did not release a survey today. Those two national pollsters (whose numbers can be square with state polling) alone failed to strain credulity today, and for that we are thankful.
Data Still Consistent With Narrow Obama National Lead
My theory of this race has long been that it exists in an effective range defined by Obama winning by up to 6 to a virtual tie in which Romney could prevail. Obama overperformed for a while in September as Romney ran badly, placing Obama near the limit of his performance, while after Denver Romney stood near his. State polling suggests the election is settling somewhere in between, perhaps closer to the Romney end of my range, but not at the edge of the range, as Romney would need to win. The solidity of Nevada, the seeming solidity of Wisconsin, the drift back to Obama in Colorado and Virginia, seem to corroborate this. It does seem that Romney should win Florida and North Carolina, though Florida seems close enough to be won by Obama yet.
From here, look for polling of Wisconsin, Colorado, Virginia, and Iowa. If Romney fails to pull even in Wisconsin, he will lose unless he sweeps Ohio, Virginia, and Florida (and could lose anyway). Colorado substitutes in for Wisconsin for Obama. He can win with Nevada + Ohio + Colorado and no other swing state. So if motion to Obama continues there, or even flattens at a narrow lead, that's very bad news for Romney.
In sum, if Romney cannot pull of an Election Day surprise in Ohio, the data on the board simply are not consistent with him winning the Presidency.