Before Hurricane Sandy derailed the Presidential race, I examined the possibility of a Romney Electoral College victory without winning the crucial battleground state of Ohio. I concluded that there were paths to victory that did not include Ohio and reasonable expectations that those paths are viable within the current polling framework.
But while an Ohio-less Romney victory is reasonable, the easiest path to the White House still captures the Buckeye State. Can Romney still win Ohio? The polling, on its face, is not in Romney’s favor; he is currently trailing by 2.3 percent in the RealClearPolitics average of polls from the last week. While this number seems small, it cannot be explained by random variance – the polling in Ohio has been too deep and too consistent. If the “true” state of the race is Romney leading in Ohio than one would expect to see a spread of results that show both Romney and Obama up a couple of points; instead you see a something closer to a bimodal distribution of the polls – that is some polling firms are consistently finding an Ohio race that is close to tied and a second group of polling firms are consistently finding the president with a three to five point lead. A graph, courtesy of Jay Cost at The Weekly Standard, demonstrates this fairly clearly.
As I said a week ago, there are reasons to believe that President Obama has some institutional advantages in Ohio compared to the national electorate and, therefore, reason to believe the polls that show a slight Obama lead. Unemployment has dropped by a much larger percentage in Ohio than nationally over the past three years and (as I confirm later) the auto bailout is popular in the state while remaining relatively unpopular nationally. This all suggests that Romney still has ground to make up.
But there is also reason to suspect that the group of polls showing a tied race, in line with the national polls, is more likely to be accurate.
To see this, I examined the internals of the most recent SurveyUSA poll of Ohio that showed President Obama up by 3 points. I chose this poll both because SurveyUSA transparently breaks down its results by in-state geography and because the poll falls into the “slight Obama lead” portion of the polling distribution.
SurveyUSA found Obama leading Romney in Ohio 48%-45%. It divides Ohio into six different geographic areas: Cleveland, Columbus, Cincinnati, Dayton, Toledo and Southeast Ohio. Here is their data.
Unsurprisingly, Mitt Romney is winning in the southwestern parts of the state (Dayton and Cincinnati) and Barack Obama is winning in the northeast (Cleveland) with Columbus in the middle, both geographically and politically. President Obama has unusually strong support in the Northwest – supporting the conventional wisdom that the auto bailout is helping him in the state (most of Ohio’s automobile-related manufacturing is centered in this region). Comparing these results to the results in both 2008 and 2004 confirm this.
In Toledo, President Obama is polling almost five points better than his overwhelming victory in 2008 – I’d venture a guess that the Toledo metropolitan area has seen the highest increase in Obama popularity nationwide.
But the strength of the president in Toledo is disguising softness in his support in the rest of the state. As you can see, the president is running behind not only his own numbers from four years ago, but also John Kerry in 2004 in each other geographic region. He is 4.7 points behind his 2008 result in Cincinnati, 4.3 points behind in Cleveland, an enormous 6.1 points behind in Columbus and 3.2 percent back in Dayton.
The difference lies in the relatively high number of undecided voters, between 5 and 7 percent. This is probably indicative of two things: first, an electorate that has been bombarded by negative political advertising and has grown unhappy with both parties. This factor could potentially lead to a late break in undecided voters for Gov. Romney if the anti-political mood manifests itself as an anti-incumbent mood.
That, however, is speculative. More importantly, this probably represents a loose “Likely Voter” screen by SurveyUSA – their standard to determine if someone being polled is going to vote or not is accepting too many people who will end up not voting. This favors Mitt Romney on its face, since Republicans have almost always had better results when “likely voter” filters are applied and, in particular, in this election because of an enthusiasm gap that favors Republicans.
And where the undecided voters are helps Romney, as well. Only Toledo exhibits the natural Obama advantages I spoke of earlier (relatively low unemployment and a popular auto bailout) and it sees only 3 percent undecided voters. Dayton, Cincinnati have experienced marked decreases in unemployment but are traditionally Republican areas and, as such, are not wooed by the bailouts. Columbus has had something of a drop in unemployment but the lack of youth enthusiasm is more likely to aid Romney here. And Cleveland, while a traditional Democratic stronghold, is still mired in significantly higher unemployment than both the rest of the state and the rest of the country.
All told, this suggests that the non-Toledo portion of Ohio is probably behaving more like the nation as a whole.
While this supposition is drawn from only one poll, it does explain the disparity between the “tied” polls and the “slight Obama lead” polls. A stricter “likely voter” screen would show less undecided voters but also 2004-esque breakdown in the non-Toledo areas of the state.
So Romney’s path to victory is narrow, but relatively simple. If he wins levels of support akin to George W. Bush in 2004 across the state (with the exception of the Toledo area), he can eke out a win. And in this scenario the “true state” of Ohio is probably tied. The math is fairly easy to demonstrate; even if Obama wins the remaining 3 percent of undecided voters in Toledo and we assume that current Obama support in the other counties will increase to John Kerry’s level in 2004, Romney still garners 49.7 percent of the vote. Accounting for one to two percent for third party candidates, this scenario would be enough for a Romney victory.
This would also mean a very late election night and the possibility of recount sequel …