It was first mentioned on BOR, but the Jim Henley campaign did a poll of CD07 residents, which was written up in this Bellaire Examiner story. Here are the results:
Henley poll: 515 respondents
Republicans 225
Democrats 99
Independents 137
MOST IMPORTANT ISSUES
War in Iraq
Immigration
Economy
Education
GOVERNOR’S RACE
Rick Perry (R) – 36%
Chris Bell (D) -15%
Kinky Friedman (I) -14%
Carole Strayhorn (I) -12%
Undecided -18%
CONGRESS DIST. 7
John Culberson (R) – 51%
Jim Henley (D) – 39%
Undecided – 10%
Very interesting, and I don’t think anyone could say that Republicans were undersampled or Democrats were oversampled. If this is an accurate result, it would project to something like a 57-43 win for Culberson in a district that went 65% GOP in 2004, 70% in 2002. I don’t know about you, but I’d call that pretty darned underwhelming, and I’d bet that if something like that were to happen, this district would get a lot more attention in the 2008 campaign.
The Governor’s result is also pretty interesting, and it gives a little perspective to all the statewide polls. CD07 is more Republican than Texas is as a whole (again going by 2004 results), so if Perry is only pulling 36% here, maybe he’s worse off than SUSA/Zogby/Rasmussen are suggesting. That’s a pretty high undecided number, too. Let me offer my hope here that when and if other Congressional (and State House) races get polled, they ask about the Governor’s race as well.
More on the survey is here.
If this is an accurate result
It’s really hard to take polls conducted by “volunteers” and not reputable pollsters all that seriously, especially polls conducted by one campaign’s volunteers.
Why? This wasn’t a push poll. It was a straight question of preferences. I don’t see how this is any less useful than a poll commissioned by a campaign.
Why don’t you follow the advice you so often give others and contact the Henley campaign to ask them yourself how they conducted the poll? Maybe they can adequately answer whatever objections you have.
Kevin, I can email you an old copy of the questions we asked while doing the polling. (I say old copy because there were a few small edits to the script, but nothing big). We had our technique reviewed by a polling expert, and the entire process was run by a young lady who had worked on political campaigns before.
On another note, it seemed evident from listening to Henley that there is reasonable hope that these numbers, while accurate, will change for the better for the Henley campaign, and that this campaign is still definitely winable. It is definitely an uphill climb, but we realized that from the start.
Anyone who takes an in-house, volunteer-run campaign poll seriously is a rank amateur — or working for the campaign.
Is anybody going to offer a substantive criticism of this poll, or is it just gonna be one pronouncement from on high after another?
There’s no deep magic in doing a poll. The questioners are reading from a script. Why should it matter who the questioner is? If the script is reasonable, and the sample is reasonable, the result they produce should be reasonable.
Is this a reasonable result? Well, as I said before, it projects out to a 57-43 Culberson win, which would be a drop of eight points in his performance from 2004. But here’s a little secret: 2006 is not 2004. George Bush is not helping anyone here. Poll after poll shows a strong anti-Republican, anti-Congress, anti-incumbent mood out there. There’s even polls to show that approval of one’s own Congressperson, which usually soars high above approval of the institution as a whole, is historically low right now. Why shouldn’t all of this be a drag on John Culberson? And that’s before we even take into account the possibility that things he’s done lately, like his stance on the Universities rail plan, may be a net negative for him.
I say this is a reasonable result. Like any poll, it may turn out to be way off the mark, but it’s close enough to any likely ballpark that I think it ought to be taken as a real data point. You want to suggest otherwise, try giving me some evidence rather than your word for it.