As I wrote last night, liberal analysts are right when they point out that the preponderance of state polls have greatly strengthened President Obama’s hopes for re-election. But a couple of the latest ones published this morning contradict that conviction, which caused New York Times columnist Paul Krugman to claim only stupid people think the election is not a cinch for Obama. One Democratic-leaning pollster has Romney ahead by one point in supposedly deep-blue Michigan, while a new Pennsylvania poll shows the race there deadlocked.
These may be outliers, but even a Nobel laureate (and, as the Wall Street Journal’s James Taranto likes to say, “former Enron advisor”) like Krugman is smart enough to understand that if Romney wins Pennsylvania and Michigan, Obama has virtually no chance to get to 270 electoral votes. The point here is that while we are all rightly focused on who will win Ohio, the president’s hold on a number of states that were thought to be likely Democrat wins is far from secure. What’s happened in the last month since the Denver debate turned the race around is not just a surge of Republican strength in the South and the West but a surprising comeback for the GOP in the rust belt and the Midwest.
The Michigan poll is from the Democratic firm of Baydoun/Foster sponsored by WJBK Fox Channel 2 in Detroit, and has a sample that has a nine percent edge for the Democrats in terms of partisan identification. More tellingly, it is a fairly large number of respondents for a state poll — 1,913 likely voters — and a relatively low margin of error at 2.24 percent. Yet shockingly it shows Romney up by more than half a percentage point: 46.86 percent to 46.24 percent.
It should be specified that most other Michigan polls are still showing the president with a lead there. Another Democratic pollster, Public Policy Polling, has Obama up 52-46 percent in their latest poll. Just to confuse things, that poll has a smaller Democratic edge in partisan identification at only six percent but it is also the product of a much smaller sample — only 700 likely voters — and therefore has a margin of error that is nearly double that of the Baydoun/Foster poll.
In Pennsylvania, a Susquehanna poll sponsored by the Pittsburgh Tribune-Review shows the race in Pennsylvania a virtual tie. Indeed the poll’s sample of 800 likely voters showed 378 say they would vote for Romney and 372 for Obama. Again, Susquehanna is a bit of an outlier in that it has shown more strength for Romney throughout the campaign than other polls. The Real Clear Politics average of polls for Pennsylvania still shows the president up by more than four points. But it should also be pointed that a clear difference between Susquehanna and the others is the same one that has been stirring discussion about virtually all the presidential polls on both the state and the national level: partisan identification. Susquehanna (whose sample is larger than that of the other Pennsylvania polls) shows a six-percentage point advantage for the Democrats. By contrast, two other polls that show Obama ahead in the state, PPP and Franklin & Marshall, had samples with 10 and nine point edges for the Democrats.
Those numbers make the contradictions between these polls more explicable. It can’t be said often enough that turnout is the key to this election. Those polls that are assuming a large advantage for the Democrats are pointing toward an Obama win. Those that are not are favorable to Romney. It’s as simple as that. If the Obama campaign machine can manufacture a replica of the 2008 electorate, the polls and the analysts predicting and Obama win will be vindicated. If not, then Romney may be on his way to victory and Krugman will be the one sitting in the corner wearing the dunce cap.