Romney’s rise

HuffPost’s Pollster tracking model suggested a 4.5-percentage-point swing to Romney, very close to the average 4-point shift I computed from all Gallup data around debates. In Pollster’s model, Obama’s national lead shrank to just half a point.

But four questions present themselves.

First, will the bounce persist? By the time you read this, another debate could have altered the dynamic again, but mounting evidence suggests Romney has plateaued. Pollster’s latest margin is ever so slightly more tilted in the president’s direction than it was earlier. We saw movement toward Romney post-debate, but in some states the surge stopped, while in others the numbers started moving back in the president’s direction — a trend also evident in the HuffPost Pollster’s model. On Oct. 9, they put 263 electoral votes in the president’s column, but by the 15th, the number climbed to 281.

Some national poll-to-poll comparisons reveal the fading impact of Romney’s debate victory. Shortly after the debate, Ipsos put Obama 3 points behind Romney; today he’s 2 points ahead. IBD/TIPP’s poll gave Romney a 2-point lead just after the debate, but he is tied with the president today. Rand’s tracker had Obama’s lead down to 2 points on the 10th, but is back at 5 points today.

These are small shifts at best, but they do suggest Romney’s rise has been arrested, and perhaps even reversed.

Second, will the debate change the outcome? It’s one thing to see a debate narrow margins, something else to say it determined the outcome. Again, this question is more complicated than it seems. Consider the simplest version: At any point after the debate, did it appear probable that Romney would win the general election? No. At no time did Pollster’s model give Romney the lead in the popular vote. Even when they assigned the president 7 fewer electoral votes than he needed for victory, he also had an 83 percent chance of winning Ohio and a 63 percent chance of winning Colorado, either of which would have put him over the top. Romney was then likely to win just 206 electoral votes.

Was it really the debate? There is some evidence more was at work here. Turn again to the Pollster model, which suggests Romney’s upswing began around Sept. 23, well before the debate.

The Rand survey offers up another intriguing tidbit. By re-interviewing the same voters repeatedly, this poll examines the extent to which voters are shifting their allegiances. The number switching to Obama was greater than the number moving to Romney every day during the last weeks in September. However, beginning a couple of days before the debate, those switching to Romney began to outnumber those moving to Obama. Similarly, their horserace data found a sustained upward trend for Romney beginning Sept. 27, peaking on Oct. 10.

If it wasn’t just the debate, what else was it? Here it gets interesting, harkening back to analysis based on fundamentals that I have been preaching for years. In short, somehow or other, the horserace always gets aligned with the fundamentals at some point in the cycle. We attribute it to a debate or a gaffe or an ad, but it’s actually just fundamental reality asserting itself. Ten statistical models developed by political scientists that consider only economic and political fundamentals (not trial heat polls) predict an average two-party vote that is tied at exactly 50 percent for each. This was destined to be a close race. Somehow, sometime, the horserace polls were going to fall in line with the fundamentals.

With the HuffPost model showing a nine-tenths-of-a-point advantage for the president, it could hardly be closer to 50/50, but with 281 electoral votes now likely in the president’s column, he remains the favorite.

Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982. Current clients include the majority leader of the Senate and the Democratic whip in the House.