New polls temper Obama bounce assumptions after convention

While early polls coming out of the Democratic National Convention did show a strong bounce for the president, two new polls out Monday raise questions about how much changed during the two weeks of the conventions.

The strongest indicators of an Obama bounce were the Rasmussen and Gallup tracking polls, which showed identical 5-point spreads favoring the president on Monday. Tuesday's Rasmussen update drops the president's lead to 3 points.

On Tuesday a new ABC/Washington Post poll was released which showed the two candidates virtually tied among likely voters, with Obama leading 49 to 48 percent. These numbers are essentially unchanged since before the conventions.

This poll has the virtue of actually revealing its partisan sample. The D/R/I ratio here favors Democrats 33 to 27, with independents at 36 percent.

Ed Morrissey at Hot Air breaks down the numbers:

"Registered-voter samples don’t mean much at this stage of the election; it’s likely voters that provide predictive data from surveys. They mean even less when only 26 percent in the sample are Republicans. The likely voter sample improves that by a point to 27 percent, but still has a D+6 D/R/I at 33/27/36. The 2010 midterms had a national turnout D/R/I of 35/35/30; the 2008 election was D+7 at 39/32/29. A GOP turnout of 27 percent would be among the worst ever in a presidential race, if not a record. Since enthusiasm measures in other surveys, notably Gallup’s, show an enthusiasm gap favoring Republicans, I’m not inclined to buy this poll’s likely-voter split as a model for this election."

Most observers expect the turnout this year to split the difference, which means that this poll, showing the candidates tied, still significantly over-samples Democrats.

The detailed view of the ABC/Washington Post poll had independents favoring Romney 54 to 43 percent. Thus, assuming a turnout ratio similar to those from 2008 and 2010 puts Romney slightly ahead in this poll.

Less useful is a new CNN/ORC poll, which shows Obama leading Romney 52 to 46 percent but independents favoring Romney by 14 percent.

Unlike ABC/Washington Post, the CNN/ORC poll does not offer partisan ratios, so readers can only make suppositions about how pollsters arrived at these numbers. The pollsters did not respond to Deseret News inquiries for internal data.

To show Obama leading by 6 while trailing among independents by 14, the balance must overwhelmingly favor Democrats and reduce independents to almost nothing.

Some polls do follow a legitimate practice of pressing respondents to identify which way they lean and then "squeezing" them into that party, reducing the number of independents. But the CNN/ORC poll offers no transparency on this front.

And with an unsqueezed 36 percent independent sample in the Washington Post/ABC poll favoring Romney by 11 points, it seems unlikely squeezing that group would sharply favor the Democrats.

In short, while it is still too early to say, the apparent clarity of the past few days of polls favoring Obama may have been a mirage.