Maiello: Defeat the Press
Wolraich: Obama at the Gates of... Gates
There was a bit of drama in Poll Land on Sunday, as a well-respected source in Ohio released The Ohio Poll, showing a dead heat between President Obama and Governor Romney. It was the third poll that ever showed Romney hitting 49 in Ohio (Obama has hit 50 19 times, including in the other two polls that showed Romney at 49). Indeed, two polls posted late Sunday night corroborated the theory of this piece that The Ohio Poll was taken during a modest recession in the President's poll numbers. Moreover, other state polls released Sunday were consistent with the race being roughly even in popular vote, with the President leading narrowly in key states. This race remains on course for a narrow Obama victory, most likely as a result of the President winning either Ohio (where he has led this week and most of the last two months) or Virginia (where he has seemingly moved into a tie). Here is why the data continue to favor that narrow Obama win:
Cincinnati Enquirer's Poll Showing Ohio Even Is Sound, Not Cause For Concern
Today the Cincinnati Enquirer released the results of its Ohio Poll, showing Ohio all tied up at 49. The poll drew from a very large sample, but has been criticized by some Obama supporters for the fact that it was drawn from interviews on October 18-23, before the final Presidential debate. Nate Silver brushes that criticism off, because he suggests that Obama has seemed to him to be ahead by 2, and that when you are ahead by 2, you will draw some ties, and also because (Nate argues) the second debate only led to a half-point or so national drift back to Obama.
I agree with the first half of Nate's response to The Ohio Poll -- you should draw some results a point or two away from your lead if you lead. But I differ with his second point, and draw some comfort from seeing The Ohio Poll as of a piece from a slightly different period in the Ohio race. To understand why this is so, it is helpful to look at the recent chronology of polls in Ohio in several phases.
First, from the Democratic National Convention through the day of the Denver debate, there were 16 polls of Ohio. President Obama led all 16. Second, in the week after Denver, there were 8 polls. President Obama led 5 of them (by 1, 1, 4, 6, and 1), while Governor Romney led in three (by 1, 1, and 1). Third, in the next full week, there were three polls, with President Obama again leading all of them (by 5, 3, and 1). Fourth, you get into the period of time in which The Ohio Poll was taken -- October 18-23. During that period of time, there were seven polls. The majority -- four -- were ties, with the other three coming from right-leaning pollsters (Gravis, Rasmussen, and Suffolk), with The Ohio Poll agreeing with them. The other three were Obama leads of 1 (PPP), 3 (SurveyUSA), and Time (5). Fifth, since that stretch, there have been three polls conducted entirely after the third Presidential debate, and President Obama has led all of them: ARG (by 2), CNN/Opinion Research (by 4), and Purple Strategies (by 2). ARG's is notable, because ARG was one of the three pollsters that showed Romney +1 just after Denver (and has also given Romney good numbers in Colorado and New Hampshire).
Thus, at one level, this week's polling represents a movement back to the time when the President clearly led, even if The Ohio Poll accurately reflected the week before. I would point to two corroborative trends. PPP, the Democratic-leaning pollster (whose numbers lean a bit blue, much as Rasmussen leans a bit red), polled into Colorado, New Hampshire, and Virginia before and after the third debate. Consistently, Obama moved a few points positive in the wake of it. Additionally, IBD/TIPP, which doesn't release its daily samples but is one of the two national tracking pollsters with results that can be squared with the states (WaPo being the other), noted that President Obama rose after the second debate (which he also won, but more narrowly), and that his numbers reverted essentially to Mitt Romney's post-Denver landing point late in that week. The time-space of The Ohio Poll and the three polls by Republican-leaning pollsters finding ties in Ohio all fell then -- late in the week after the second debate and before the third. I think Nate's model, which smooths differences day to day and draws on longer timeframes to estimate November 6 vote share, lulls one into thinking Obama will win by 2.5% or so in Ohio, and thus has always led by 2%. I tend to think Obama's lead was narrower in Ohio during the range in which 4 of 7 polls were tied, and is either largely or more clearly in existence this week.
The numbers this week in Ohio also corroborate the idea that Obama is really ahead now, when you consider where they come from. ARG's new number represents a three point move since the aftermath of Denver (from -1 to +2). Likewise, the CNN/Opinion Research poll finds the same margin (+4) in favor of Obama as it found the week after Denver, a week when Obama led Ohio in the polling consensus. Finally, Purple Strategies' +2 is down from +4 in the heart of Obama's midSeptember bounce, when the range of polling suggested a 6 point or so Obama lead.
Update: Two New Ohio Polls Sunday Night Add Comfort That Obama Leads
With two additional polls released Sunday night, it remains clear that Obama leads, albeit narrowly, in Ohio. PPP has released a poll showing Obama up 4 (51/47), and Gravis Marketing shows Obama up 1 (50/49). This means all five polls postdating the third Presidential debate show an Obama lead in Ohio, with margins of 1, 2, 2, 4, and 4. This averages to 2.6%, which is suspiciously close to the margin Nate Silver predicts Obama will win the state by. There is an additional degree of comfort in seeing that PPP's last poll showed Obama +1 (during the same timeframe in which the Ohio Poll showed the race tied), and now has him +4. Gravis showed a tie (during that same period), and now shows him +1. The polls are consistent with my hypothesis, stated before I inserted this section into the blog at 925 pm ET, that Obama improved modestly from the time when The Ohio Poll was taken.
New Hampshire, Minnesota Tell Us What We Already Know
Meanwhile, some lower-relevance polls tended to confirm what we have been seeing in other states. In tiny New Hampshire, President Obama leads by 2, per PPP (49/47). That is perfectly consistent with the recent average of Granite State polls at +2. Additionally, the polling in New Hampshire is consistent with the point I made about phases in Ohio polling: during the period October 17-23, Romney led in three of four polls (+2, +2, +1) trailing by 9 in the other. The two polls postdating October 23 entirely show Obama up 3 and 2.
We also saw a poll from the Minneapolis Star-Tribune showing Obama up only 3 in Minnesota. This poll continues to suggest a roughly 5 point Obama lead there, as it comes the day after an 8 point Obama lead in Minnesota observed by St. Cloud State. Given that Minnesota, like Wisconsin, Iowa, Pennsylvania, and New Hampshire, was a state Obama won by 10 in 2008, one would expect him to win it by considerably less, but the idea of Governor Romney making up the entire 10 there without advertising or organizing is not credible. I continue to agree with Nate that Romney should have expanded his board by playing in Pennsylvania. Minnesota, not so much.
National Polls Contribute No New Information Today
In national polling today, Gallup moved toward Obama by 1 (now Romney +4), Rasmussen moved to Obama by 1 (now Romney +3), while IBD/TIPP moved toward Romney by 1 (now Obama +1.3). All of this is within the margin of noise, and as I have argued in prior posts, Gallup and Rasmussen' trackers -- because they seem inconsistent with basically all state polling, and in Rasmussen's case, seem inconsistent with Rasmussen's own state polling -- are so much noise themselves.
Conclusion: OH, VA, WI, CO Polling Key; Prediction, Obama 298, Romney 240
As we head toward Election Day, it seems very likely that Barack Obama narrowly leads Ohio. We should be watching the next week's polls to see if they remain consistent with that appearance. Because almost 30% of Ohio's votes are cast, it will be hard for Romney to move that pile, and Obama's 120 campaign offices are continuing to push for every vote. Virginia is a close question, though losing it would almost surely mean a Romney loss as well, so further polling there is key. Wisconsin or Colorado, when added to Ohio, would put President Obama over the top, so the President's consistent but narrow lead in Wisconsin, and his seeming narrow lead in Colorado, remain key.
I would simply ignore the national polls. We will have a better sense on November 6 which ones stunk and why (it is guaranteed that some have stunk), but the motion that matters is in those four states (unless Florida leaps into the Obama column, a possible turn of events that would make all of this moot). The national polls give gentle trend information that correlates pretty well to state motion, but the states and their frequent polls are logically the place to look.
Right now, in order from strongest to weakest states for President Obama, I see him winning with NV, CO, OH, WI, NH, and VA (298 EVs), losing IA, NC, and FL to Romney (240 EVs), a result very close to Bush-Kerry. If I have the states in the right order, then Obama has three states to spare, as a cut line at Ohio would give him a 271-267 victory, precisely the margin of Bush-Gore.