New PPP poll: Perry 48, White 42

A bigger spread than last time, but still another example of a poll showing a close race.

[Rick] Perry holds a 48-42 lead over [Bill] White with 8 weeks left to go until election day.

The race is confounding the major trends we’re seeing in most contests across the country. White is winning independents 53-34. Republicans have the lead with them most everywhere else. White’s winning 82% of Democrats while Perry’s getting 77% of Republicans. Republican voters are more unified than Democrats most everywhere else. But there are a lot more GOP voters than Dems in Texas so Perry’s still ahead anyway.

At 50% a majority of Texans disapprove of the job Perry’s doing with only 39% giving him good marks. Democrats dislike him (85%) a whole lot more than Republicans like him (63%) and independents split against him by a 25/64 margin.

The full crosstabs are here This poll is similar to the Hill Research Texas Watch poll in that it shows White doing better among Democrats than Perry is among Republicans, but it differs in how independents perceive the candidates. If White had this level of indy support in the other poll, he’d have been in the lead. He might have had the lead in the PPP poll with that, too, except they had a sample that was 47-30 Republican, while the Hill/TW poll was 37/33. The respondents in the PPP poll went for McCain over Obama by a 52-41 margin, though, so it’s not like this is way off base. (I’ll get to the Hill/TW partisan mix in a minute.)

The PPP poll gives Perry a higher level of support than the others from Tuesday, which may make you think that my “soft Republican support” thesis is looking shaky. But there’s this:

To get an idea of just how well White is doing compared to overall Democratic performance in the state we also polled the Lieutenant Governor’s race. There Republican incumbent David Dewhurst leads Democratic challenger Linda Chavez-Thompson 54-34, perhaps giving an indication of how lopsided the Governor’s race might be in this political climate if the GOP had someone stronger than Perry and the Democrats someone weaker than White.

Seems to me that’s another way of saying that there are a significant number of Republicans who aren’t particularly enamored with Rick Perry. Indeed, given the smaller number of undecideds in the PPP poll, you could claim that they’re also leaning towards White. I don’t want to make too much out of one result, but I do stand by my original hypothesis, at least for now.

Getting back to the Hill/TW sample, the following is from an email I received from David Benzion, who did the polling for Hill Research:

There is, of course, no universally accepted definition of what constitutes a “likely” voter or standard set of criteria for defining one… but at a minimum there is agreement that the mere act of being a “registered” voter isn’t enough. There has to be some reason to think the respondent can be expected to show up at the polls.

In this case, probably the most precise way to characterize our statewide sample of 600 is to say that they are “active” voters. By that I mean they are not merely “registered,” but have voted in at least one (often more, of course) of the last three cycles–2004, 2006 or 2008. In other words, it is unlikely that many of them are now dead or simply registered in the course of getting their drivers license, or used to vote long ago but are no-longer politically engaged. They are real, genuine voters–a realistic approximation of the Texas electorate as a whole.

This approach was utilized for a variety of reasons, including the desire to maintain methodological consistency with previous surveys conducted for this client. Also, this early in the Fall campaign (we were in the field pre-Labor Day), screener questions about self-perceived likelihood of voting do you some good, but hardly as much as asking those same questions one month or 10 days out from Election Day. When it’s all said and done, the best predictor of future voting behavior is past voting behavior. By that definition alone, our sample consists of likely voters.

For what it’s worth though, I had our statistician run the numbers for only those respondents who voted in the 2006 gubernatorial election. These are known, proven, off-year, Texas-gubernatorial voters. Among them (n=395, MOE +/-4.93%) the ballot comes out (as of 8/25-29) at:

Perry: 44%
White: 43%
Undecided: 11%

Additionally, rather than rely on self-reported measures of likelihood of voting (people flatter themselves and/or want to look like “responsible citizens” in the eyes of the interviewer), HRC has developed a proprietary likely-voter model based on nearly a dozen different demographic and attitudinal factors.

Running the data through our internal “most likely” voter model (n=441, MOE +/- 4.67%) puts the ballot currently at:

Perry: 45%
White: 43%
Undecided: 11%

Bottom Line: No matter how the numbers get diced, as of August 5-29,
we see Perry with an edge and White with an opening.

This survey was not designed to measure–and is therefore completely silent about–whether Perry can quickly build on and solidify his edge, or White can realistically and effectively exploit his opening.

For that God created blogs.

Can’t argue with that. Every poll is a snapshot, and no one poll is the whole story. I’m just glad we have more data points, especially more non-Rasmussen data points.

Related Posts:

• UT/Trib: Cruz 41, O’Rourke 36, part 2
• Abbott and the Latino vote
• Precinct analysis: Abbott versus Perry in Latino districts