Chron CD22 poll

Obviously, the big news today is the Chron poll of CD22 (reported last night on KHOU), which claims that the race is a lot closer than anyone thought it would be. Here's how the Chron puts it:

Thirty-five percent of respondents said they would vote for a write-in candidate, a statistical tie with the 36 percent support for Democrat Nick Lampson, according to the poll of more than 500 likely voters in the 22nd Congressional District.

Most who say they will write in a candidate plan on naming Shelley Sekula-Gibbs, the Houston city councilwoman backed by the Republican Party. Two lesser-known candidates also are running as write-ins.

One voter in four is still undecided.

Libertarian Bob Smither, the only person besides Lampson on the general election ballot, drew 4 percent support.

"Most" means about 80%, according to the poll's internals, which you can see here and here (both PDFs). Which in turn means that Lampson leads the written-in Sekula Gibbs by a 36-28 margin. Not quite as sexy as 36-35, but there you have it.

There are many points to discuss in this poll. Most of what I'll be referring to comes from one of the two PDFs linked above.

- The poll sampled 504 registered voters, of whom 450 said they were "likely" to vote this year. I'm here to tell you, ain't no way in hell any random sample of 500 registered voters contains 90% likelies. SurveyUSA has been pegging about 58% of its samples as likely voters, presumably based on past voting history. That's too high as well for a non-Presidential election, in which 35-40% turnout is the norm. In this case, it appears the voters self-identified as likely. And a lot of them fibbed about it.

- Where that may make a difference of course is in its mix of partisan identity. The sample is 32% Democratic, 52% Republican, and 16% Independent. But how many of each of those groups is really likely to vote? That matters quite a bit. This is a weird year in many ways, so any method for determining voter likelihood is not much more than a guess. Are the Democrats more fired up here, as is the case around the country? Or are the Republicans excited about maybe winning a race they're supposed to lose? I don't know how you can judge from this poll.

- Forty percent of poll respondents say they normally vote a straight ticket; the rest say they do not. I've looked at straight ticket voting several times, and in 2004, over 70% of the votes Tom DeLay got in Harris County came from straight tickets. As with the likelihood question, I think some of these respondents are not answering truthfully. Most people don't want to admit they don't vote, and most people don't want to be thought of as rigid partisans (this is why there are more self-identified "independents" than there are people who genuinely vote both parties).

- About half of Democrats in the sample say they'll vote a straight ticket; for Republicans it's 42% straight, 56% not straight, 4% not sure. The Dem numbers are in line with historical patterns, the Republicans are a bit low, but understandably so given the advertising telling them to not vote a straight ticket.

- A total of 149 Republican repondents said they would not be voting a straight ticket. A total of 146 Republicans said they'd vote for a write-in candidate. To say the least, that's a high concentration. It suggests that this may be Sekula Gibbs' ceiling of support.

- Conversely, 108 of 161 self-identified Democrats said they were voting for Lampson, with an additional 35 saying "not sure". This suggests that Lampson's support may be understated.

- 26% of Republicans (69 out of 262) and 24% of Independents (19 out of 81) say they're not sure who they're voting for. It's hard to judge what they might eventually do. In a subsequent question that named Sekula Gibbs on the ballot, the 61 "not sure"s were pushed, but only 18 then identified a candidate. No such pushing was done for the Lampson/Smither/Write In question, where there were twice as many (123) "not sure"s. One might surmise that these are the people least likely to vote.

- It's hard to believe that Bob Smither will get only 4% of the vote. Past history suggests that Libertarian candidates, when they share a ballot with only one major party contestant, get 10-15% of the vote. My guess is that Smither will pick up a number of the not-sure voters, probably more Republicans since those are the ones he's specifically targetting.

- One last point to note is that the war in Iraq was by far the most important issue cited by the respondents, easily beating terrorism, the economy, and immigration. Nearly half of Democrats listed it first, as well as neearly half of Independents, while it trailed only terrorism among Republicans. I point this out mostly because it jibes with the recent poll in CD04, where the question wasn't asked but the pollster reported that almost everyone wanted to discuss it.

The bottom line is that I think this article makes the race sound tighter than it is. I do think it's tighter than I thought it would be - while it hasn't spent $3-4 million, the NRCC has spent over one million dollars, and that's had an effect - but to characterize it as a tossup between Lampson and Sekula Gibbs at this point is an overstatement. Kos, Juanita, Vince, and MyDD also discuss this.

Posted by Charles Kuffner on October 30, 2006 to Election 2006 | TrackBack