Evan Smith sent me this link to a poll in Lubbock County that has Gov. Rick Perry leading Sen. Kay Bailey Hutchison 52-37 among Lubbock County voters, with 11% undecided. The video I saw was vague on the details of the poll done by Mike Stevens, who I don’t know. (I’ll try to send him an email — if you see this, Mike, could you please send me more info on the poll, like confirming what the sample is here and MOE?)
Lubbock is obviously a heavily Republican county and a conservative one to boot — Hutchison ran 21 points stronger than her statewide total there in 2006, and Perry ran 5 points stronger. (Don’t read too much into the 21 v. 5 comparison — remember, his was the four-way race, and Kinky Friedman and Carole Strayhorn got about 33 points between them in Lubbock.) In 2008, McCain/Palin beat Obama/Biden there 68-31.
Not many local polls have been made public, but this one confirms the rumor mill: One, Perry’s lead in conservative areas of the state is a few points above what the UT/Trib poll found statewide. And, two, the impression that, as of now, the Hutchison campaign is floundering if they are really trying to peel conservative voters away from Perry in significant numbers.
The poll’s methodology falls somewhere between unorthodox and dubious.
The pollster is a printer. Michael Stevens is running his busy but offbeat polling operation out of Action Printing in Lubbock (they do some political printing, he says, for candidates on both sides of the partisan line, and also run some campaigns).
He’s got a weird polling method. He gets lists of people who regularly vote and sends them postcards directing them to a polling website that, he says, lets him know exactly who’s responding. (Mr. Orwell? Hello?) For this particular poll, done in advance of the November elections, which included some bond issues in Lubbock, he sent out 18,883 postcards and got 2,087 responses. He says the margin of error on that is +/- 1.89 percent. This particular poll was done October 30 through November 4, but wasn’t released until this week.
He said he included people from four lists: people who’d voted in previous Republican primaries made up the majority of a group that also had registered voters new to Lubbock, people who voted in three elections in the last year, and voters who took part in “Operation Chaos” — a grassroots campaign geared to flood the 2008 Democratic presidential primaries with people who’d vote for Hillary Clinton and against Barack Obama.
He defends the accuracy of his polling, but says he doesn’t have much company in that. “This isn’t accepted by most political analysts yet — they’re skeptical about it.”
Ross, Jim's posting wont allow me to comment for some reason. But is he is flat ass wrong.Then Professor Henson updates his blog...
Based on the info you provided there is no reason to discount that Lubbock mail in survey then an Internet survey or a telephone survey. The reliability will be about the same. All three methods have issues but the Lubbock survey using a combo of mail and the internet is just as reliable as an opt in web based users or one just calling off a voter list. Jim should know better then to trash a poll base on a new concept. More Intel could shed new light on this but the details known to readers so far do not diminish the results of this survey.
This is not a sound sampling method, period.
The sampling method is neither random nor matched, and there is likely significant non-response bias, a classic problem in mail-only solicitation. (In English: That means there might be something about the people that don't respond to your poll — some particular characteristic — that throws off the results and makes the poll unreliable.)Given the problems in both sampling and response, there is no real margin of error. You might do some calculations that might give you the reported +/- 1.89 MOE, but the methods don't meet the conditions to make such a calculation.