Is Gallup An Outlier?

by Deacon Blues

Mitt Romney’s last several weeks have not be kind to him. Rick Santorum and the rest of the GOP fringe have taken over the dialogue and rolled it back to the dark ages at a time when the general electorate wants to hear about jobs and the economy. So with Santorum restarting the culture wars and making Obama’s religion an issue and reopening tired issues from 2008, you’d think the polls would show some damage to the GOP in head-to-head matchups. And you’d be right:

In a general election contest against Mitt Romney, the president leads his Republican opponent 51 to 43 percent, according to an Associated Press-GfK survey on Wednesday. This demonstrates a significant lead for Obama compared to the December poll when the two were virtually tied at 47 to 46 percent.

Meanwhile, Obama would lead Rick Santorum 52 to 43 percent, Newt Gingrich 52 to 42 percent, and Ron Paul 53 to 44 percent.

And this was a recent, good-sized poll.

The Associated Press-GfK Poll was conducted Feb. 16-20 among 1,000 adults and has a margin of error of plus or minus 4.1 percentage points.

But to get Santorum seen as a plausible mainstream candidate while waging this holy war against the president, you’d need at least one poll and willing news outlet to seed the ground and set a narrative that these issues matter to the electorate. Eight years ago, this blog was the first to notice that Gallup and the USA Today were doing the GOP’s work for them in selling us the fiction that George W. Bush was more popular against John Kerry than perhaps he actually was.

Is Gallup doing it again? On Monday, Gallup released a new USAT/Gallup poll taken during the same time as the AP poll, and told us that Romney was beating Obama head-to-head, and Santorum was running even with Obama, even after all this despicable rhetoric from the GOP. So at the same time the AP poll tells us that Obama beats Romney by eight points and Santorum by 9, Gallup is reporting that Romney leads by 4 and Santorum is even. Why?

Well, it could be the sample and respondents that Gallup uses in the poll. We obtained the poll background information from Gallup yesterday, and were surprised to find out that according to Gallup, a representative sample of registered voters in the electorate at this time is:

40% conservative
37% moderate
21% liberal

Really? I didn’t know that 40% of the country was conservative.

Twenty-nine (29%) of the sample had incomes above $75,000. Really? Nearly a third of the electorate makes that kind of money? I didn’t know that.

And Gallup used a sample with a party identification that was 27% GOP, 29% Democrat, and 43% independent. When Gallup forced respondents to pick a party, the sample they used conformed to the conventional beltway wisdom that the country was 45% GOP/GOP-leaning, 45% Democratic/Democratic-leaning, and only 8% truly independent.

So, you think that the difference is that the AP poll is of 1,000 adults nationwide, whereas the Gallup poll was of 898 registered voters nationwide? Alright, let’s compare the Gallup results for February 16-19 of 898 registered voters during Santorum’s nuttiness with other polls of registered voters earlier this month, before Santorum was in full caveman mode:

Gallup poll, Feb. 16-19
898 registered voters nationwide
MOE of 4%

Obama 49%
Santorum 48%

CNN/ORC poll, Feb. 10-13
937 registered voters nationwide
MOE 3%

Obama 52%
Santorum 45%

CBS News/New York Times poll, Feb. 8-13
997 registered voters nationwide
MOE 3%

Obama 49%
Santorum 41%

Pew Research poll, Feb. 8-12
1,172 registered voters nationwide
MOE 3.5%

Obama 53%
Santorum 43%

And yet, according to Gallup, Santorum has surged into a tie with Obama while in the midst of ultimate nuttiness.

Again, for Santorum and his retrograde message to be seen as credible, and for the GOP to keep from falling off a cliff in this race, you'd need help from the media and at least one willing pollster to set a narrative that the crazy talk about taking this country back 50 years and about Obama's alien nature were themes that resonate with large blocs of the electorate.