The American Action Forum is back with a new batch of House polls. (They released their first dozen last week.) One major caveat, though: as pointed out in the comments section (!) of Hotline on Call, the ballot head-to-head questions in each poll were buried under several thermometer-style questions on Obama, Pelosi, and healthcare reform:
Those American Action Forum polls are designed to prime respondents into selecting Republican candidates. Instead of starting off with the ballot-test question, they start by asking if people think the country is on the right or wrong track, do they support/oppose health care reform, and their fav/unfav opinion of Obama & Pelosi (not Boehner and Bush, of course). Look, the Democrats are going to lose a lot of seats, and they will lose some of these seats in this poll, but the most credible way to conduct a general election poll is to start by asking who the respondent is voting for, because that is the one question that will actually be on a ballot. At least I give credit to these Republicans for showing the questions they asked.
That’s absolutely right. For any poll to maintain its credibility, the toplines need to be asked at the start (or at least, no later than right after the favorables, as PPP does it).
That said, we can still take a look at the toplines with a grain of salt.
Ayres, McHenry & Associates (R) for the American Action Forum (8/16-21, likely voters, MoE: ±4.9%):
IA-03:
Leonard Boswell (D-inc): 41
Brad Zaun (R): 51
IN-02:
Joe Donnelly (D-inc): 46
Jackie Walorski (R): 44
MI-07:
Mark Schauer (D-inc): 40
Tim Walberg (R): 50
MO-03:
Russ Carnahan (D-inc): 54
Ed Martin (R): 38
OH-01:
Steve Driehaus (D-inc): 45
Steve Chabot (R): 47
OH-13:
Betty Sutton (D-inc): 43
Tom Ganley (R): 41
OH-15:
Mary Jo Kilroy (D-inc): 44
Steve Stivers (R): 49
OH-16:
John Boccieri (D-inc): 35
Jim Renacci (R): 49
Note: Check out the downright geriatric sample for the OH-16 poll — 4% of the voters are between the ages of 18 and 34, 22% are between 35 and 49, and 74% are 50 and up. SUSA’s crosstabs from 2008 had a sample breakdown among those age brackets of 20-28-52. Other polls in this batch have similar shifts, but this one is the most glaring.
WI-08:
Steve Kagen (D-inc): 39
Reid Ribble (R): 49
Aren;t these Ohio results better than most other polls. I thought Driehaus was considered DOA?
That’s an improvement on previous polling here done by Republican pollsters (way to torpedo your own narrative).
Who will commission me to survey the nearest old folks’ home and call it a poll?
And how the hell is he up on two term Democratic incumbent Steve Kagan?
mainly because it says that there will be a person in the United States Congress named “Roy Ribble”.
It still is nice to see that we wrote off Driehaus too early.
especially since they asked other crap before they got to the main question of who they’d vote for.
OH-1 obviously is a very good number, and who knows, maybe if the topline question was asked first Driehaus would be ahead. It’d be nice to get some corroborating evidence from elsewhere, but maybe we were all too hasty in giving up on this one.
OH-16 is ridiculous and should be thrown out completely. 4% under 34 vote and 74% over 50. I’ve spent plenty of time in places like Medina, Wadsworth, Wooster, Orrville, Ashland, etc, and while those places may be slightly older than the state as a whole, such an age breakdown is 100% bullshit. Take 10% away from the 50-plus and add 10% to the 18-34 and that eats up all of Renacci’s lead and maybe even puts Boccieri in the lead.
OH-13 is a good result for Sutton. If you’re leading in what amounts to a GOP internal for the most part, you’re in good shape. Ditto for Russ Carnahan and Joe Donnelly.
OH-15 is actually a slight dissapointment to me, but it shows that Mary Kilroy isn’t to be written off either. Part of the problem Stivers is having is that this seat has flown right out from under the GOP politcally. Back in 2000, it was an R+3, maybe even R+4 seat. Now it’s D+1 and moving further left.
The only unexpectedly bad number here, I think, is WI-8. I have us losing that seat but I didn’t think Kagen would be trailing by much. 10 points is more than I would have seen, even in a poll like this. MI-7 is bad too but I had us losing that one anyways, and if he’s really down 5 or so, that’s about where I had the race pegged anyhow.
Donnelly will have a closer race this year than he did in 2008, but there’s no way he’s in any serious danger of losing to Wacky Jackie. That’s the problem that I have with these results – they’re all over the map. IN-02, OH-13, and (IMO) OH-01 are way closer than they should be. The numbers for OH-16 and maybe WI-08 are ridiculous, and obviously sampling errors. Every other poll I’ve seen of IA-03 has been either a dead heat or within the MOE.
Obviously we’re losing seats. I just think there’s something fishy about these numbers. Some are worse than they should be consistent with other polling, others are way too good to be true. Either this is the one poll that gets it right, or, more likely, it’s crap.
As you noted, typically favorables are asked ahead of the ballot test. PPP does this, as do most polling firms. Asking right direction/wrong track upfront (typically the first question) is also common practice. So the only allegedly “questionable” ordering is asking their opinion of the healthcare bill before the ballot test. While I agree that I’d prefer it asked afterwards, I have a hard time believing it significantly affected the results of these polls.
These aren’t half bad. I imagine most of the plethora of GOP internals are derived from similar practices. And it is important to take note of the fact every digest reports incumbents only now going up on tv. Dems in a world of hurt yes but certainly not as dire as Republicans like to think.
While not saying that polls are totally irrelevant with such a small vote share they are much harder to poll. Given that the GOP continues to tout these polls as evidence that they are going to come in a huge tidal wave here is some recent history.
Burns vs. Critz only 3 months ago in a swing district. Burns releases a poll before the election showing him up by 2%; 43 to 41. Reality is that he lost 54% – 46% a 10% swing. Could those undecided voters really have swung that much to Critz at the end? Did it turn out that Burns have a major skeleton in his closet? No to either. It is just that the poll was either heavily biased, attempting to write a narrative, or flat out wrong.
Link to the “poll”. http://www.politicspa.com/poli…
Dems are going to lose plenty of seat yes. But do I really buy that Brad Zaun is up by 10 points when news of him being a woman beater surfaced? Pool could have been taken before the news but hard to really hard to believe. Is Boccieri really down 14 points and Dreihaus down only 2?
I compared the demographic breakdown in the American Action Forum poll to the actual voter turnout in 2006. They undersampled women, young people and probably Democrats and moderates. They oversampled conservatives and independents.