Forum

(dans l’article suivant “USAT” veut dire USA Today)

Gallup Is At It Again

by Steve Soto

If you watched “Countdown” and other cable shows last night, perhaps you noticed the media make a big deal about a new Gallup poll done for USAT over the weekend. Unlike their tracking poll of registered voters, which showed Obama with a 8-point lead, the poll Gallup did for USAT of likely voters over the same period of time, blasted across a large international media platform like the USAT showed McCain with a 5-point lead, based on a ready-made narrative (which sounds like it could have been written by the McCain camp) that Obama’s overseas trip “has not broadened confidence in his ability to be commander of the U.S. military.”

Since the same company took these polls at the same time, how can this be? Moreover, haven’t we been here before with Gallup, when a GOP candidate for president is facing a tough patch?

The answer to both questions is yes; Gallup is making some large assumptions in building its likely voter sample, assumptions that reduce the number of independents and inflate Republican numbers in that sample, in a year where we’ve been told by other pollsters that the GOP is less than enthused with McCain as their nominee and in a year where many independents will turn out. And who is around to question Gallup’s assumptions? Not the media, as evidenced by the sorry Jill Lawrence story in the USAT.

Gallup Likely Voter Sample
July 25-27 USAT Poll
McCain Leads 49%-44%

Republican: 34%
Independent: 29%
Democrat: 36%

Keep in mind that no other pollster has shown McCain getting this much support, and in fact other likely voter samples by other pollsters show Obama with a lead, in some cases a large lead. The registered voter numbers from this same poll that Gallup uses shows Obama leading by 3 points. And for comparison, take a look at the sample composition from their three-day tracking poll done at the same time:

Gallup Registered Voter Sample
July 25-27, 2008 Three-Day Tracking Poll
Republican: 29%
Independent: 33%
Democrat: 36%

In other words, Gallup is making large assumptions about who will be a likely voter, and driving those assumptions through their polls and into the media once again through many outlets. Gallup wants you to believe that if the election were held today, in the midst of McCain’s recent gaffes, that there would be a surge of GOP voters to his cause and a drop-off of independent voters. To their credit, Gallup notes that their assumptions about a “short-term energizing of the GOP base”, caused in their minds by the right-wing narrative that the media is unfair to McCain, could be only a snapshot in time, and not indicative of what will happen in November. But who besides Gallup is actually crediting this right-wing storyline of unfair media coverage with having any effect at all on a likely voter sample? Have other pollsters reached the same conclusions on this?

Perhaps Gallup has some data that validates their assumption that the storyline has led to an upswing in GOP support for McCain, like specific questioning of GOP respondents on this topic at this time. Has anyone seen any of this?

As I noted back in 2004, Gallup did the exact same thing with their samples to inflate the support for Bush at a time when other pollsters were not showing the same level of GOP support. However, since this was Gallup, and was being blasted around the world on CNN or USAT, it became the narrative used by the media.

Oh, and just for comparison, this is the sample composition of a Gallup poll just three weeks ago:

June 5-July 6, 2008 Poll
Republican: 21%
Independent: 38%
Democrat: 35%

Sure, one is a rolling tracking poll of registered voters and the most recent one is a Gallup-assembled sample of likely voters, based on their assumptions of newfound motivated GOP voters and disappearing independent voters. But tell me, what changed in the last three weeks in this race for Gallup to depress the independent numbers in their likely voter sample and inflate GOP numbers? Well, what happened was an international trip by Obama with great visuals, and a virulent right-wing pushback built upon a “the media is unfair to McCain” storyline. In short, Gallup built its likely voter sample this weekend assuming the right wing narrative as fact, affecting the whole electorate.

Yet because they’re Gallup, the media will believe that McCain actually has a five-point lead nationally among likely voters when other pollsters don’t show this, simply on the basis of a supposed surge in GOP voters and a falloff in independent voters built upon a right-wing narrative that the media is unfair to McCain.

It’s going to be a long campaign if they are starting this stuff weeks earlier than they did in 2004.

Email this • Technorati Links • Add to del.icio.us
Steve Soto :: 8:05 AM :: Comments (16) :: Spotlight :: Digg It!