Articles and Analysis


The Gallup Poll's Bouncing "Generic" Ball

Topics: 2010 Election , Gallup , Generic House Vote , Interpreting polls

Anyone following campaigns closely has seen the numbers and read the stories. Democrats are in trouble. One particularly salient point: last week's Gallup tracking poll showing a +10 advantage for Republicans in the generic ballot test. It's the largest Republican advantage in the history of Gallup asking the question. Some calculated what a +10 advantage would mean as far as seat pickup. Others simply rejoiced.

But hold on! Yesterday Gallup released its latest generic ballot test. It's evenly split between Dems and Republicans. We're coming back! Obama's address on Iraq had an effect! Glenn Beck's rally had an effect! Sound the other alarm now! Huh, I guess I missed those stories.

To be sure, Democrats are in trouble. To be sure, according to pollster.com's tracking, the generic ballot is trending Republican when you aggregate all polling outlets. But why is one poll (+10) covered so extensively, while another (+0) hardly at all? Why is the former considered important, and the second, perhaps an outlier? This new data point is receiving far less coverage. (Media Matters has a very good summary of the difference in coverage here.)

It's hard to know exactly what is causing the fluctuation--whether it's simple poll fluctuation, or "real" movement. But looking at Gallup's breakout of the generic by party, we see most of the movement comes from Democrats consolidating the base. In the current poll, 93% of self-identified Democrats say they are voting for the Democratic candidate, up from 88% in the previous wave. Republican support for the Republican candidate dropped just slightly (96% to 93%). The difference in base consolidation is now even, for the first time in a month. The chart below shows this metric since Gallup began nightly tracking in March.

Generic Ballot

That remaining Democratic holdouts would begin to come home as we head into the final stretch is not a surprise. Will this pattern hold, and how it translates into actual House seats, remains to be seen. But for those following campaign twists and turns, the latest Gallup poll is a twist worth a bigger mention.



I think the answer to the media coverage differential is obvious. The media wants to continue to show a "GOP wave" and a "Democratic disaster," so of course when a poll shows that it gets played to the hilt.

If, however, some poll shows the opposite, we'll just ignore it since we'd have to change our headlines, and, heaven knows, we cannot do that!



I do think Gallup isn't consistent about where they poll, and what their party makeup is in each generic poll. I noticed, that it looked like the week the GOP had big leads they interviewed more Republicans and last week more Democrats. The likely model on gallup will probably make progressives cry next week, but it is a good wakeup call to get out the vote.



You're 100% right, nelcon. When the media has a narrative going, anything that contradicts it is to be ignored.


Henry Henwy:

Duh. Because almost all the polls generally agree and it's easy to detect the outliers. Even gallup's +10 wasn't all that outstanding compared to other polls. It was the highest gallup had measured the republican lead to be but quite a few other pollsters have reported results in that area. Media latched on to the +10 number because it was an easy in to writing a story that didn't need that particular data point anyway. Namely that the GoP has a historically huge generic ballot advantage. If Gallup showed a bounce to dem +10 or something, _that_ would be something to talk about since it would be outside the realm of anything anyone else is reporting.


Henry Henwy:

A case of the above in action...look at pollster's aggregate graph for the generic ballot. The gallup shift from +10 to 0 has no visible effect whatsoever and if anything, a glance at the graph would show the gap is growing. You'd be hard pressed to argue that the new gallup result meant anything when it's a single data point that's not backed up by any other pollster. The gallup +10 from a week ago, on the otherhand, mirrored a trend common to half a dozen other polls.



Its amazing how the media coverage was all over the 10pt lead the GOP had lastweek,but not 1 media outlet mention this week Generic poll by Gallup.



Obama's Labor Day speech apparently had a big positive impact for the Democrats. Republicans should be very afraid. Another 2 months of Obama actively campaignining will turn this election completely around. One speech picked up 10 points for the Democrats. Obama is in campaign mode now. Remember 2008, anyone?


Thomas Wellock:

I think Gallup has more problems than these two data points. They have been all over the map for the last several months. I think it was early August or so that they had a statistically significant lead of +6 for the Dems. So they have seen a 16 point swing when other polls show far less. It may be that Dems are coming home, but I think there is something in their method that is exaggerating trends.


Bart DePalma:

Gallup has a history of wild swings in their registered voter polling.

The GOP+10 poll was ridiculous and suggested a 300+ seat GOP majority. The only reason that poll was interesting was because it was the third historic GOP lead in a row.

The swing back of ten points in no less ridiculous. It is contrary to every other poll -including the Dem polls - and nothing occurred during this period to trigger an enormous swing back to the Dems. Gallup's enthusiasm chasm favoring the GOP showed no substantive change.

Ignore Gallup until they join everyone else and put up their likely voter screen.


Mike E:

@ Paleo

"You're 100% right, nelcon. When the media has a narrative going, anything that contradicts it is to be ignored."

You are joking right? The left wing MSM in this country have been Obama cheerleaders since day one. How do you think an unqualified and inexperienced guy like Barak got elected?



Currently the Republicans lead in the generic vote polls (excluding internet polls) by an average of 7.2% (7.3% at RealClearPolitics).

At RCP the average of the 5 registered voter polls gives the Republicans a lead of 5.4%. The average of the 3 likely voter polls has the lead at 10.67%. The difference is consistent with previous midterms where Republicans have done about 5% better on likely voter polls than registered voter ones, so I think a 10-11 point lead for the Republicans is probably where the election stands today.

If I'm correct then the Gallup polls of 10 point lead for the Republicans and a tie among registered voters is consistent with Gallup being first at the high end then the low end of their error range. The results of the preceding two Gallup polls gave Republicans a 5 and a 6 point lead. I'm inclined to believe (among registered voters at least) that those are the most accurate results and the other two are outliers based on random error.



What this really points out is that all polls at this point in time are meaningless. Ras makes up polls to rev up his Republican base. Most other pollsters try to poll, but results mean nothing. Take a look at Ct Dem primary for Governor. Quinnipiac poll had Ned Lamont leading by 3 points two days before he lost by 16. But, they put out their next poll as if that one never happened. Other pollsters just as bad. Ignore them all and wait for election day.


Post a comment

Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.