Off the Kuff Rotating Header Image

Lyceum poll: Romney 58, Obama 39

This one is just nuts.

Republican Mitt Romney has a commanding lead over Democrat Barack Obama in the presidential race in Texas, with 58 percent to the incumbent’s 39 percent, according to a new Texas Lyceum Poll.

The same survey has Republican Ted Cruz leading Democrat Paul Sadler 50 percent to 24 percent in the race for U.S. Senate from Texas. Those two are among the candidates vying to replace Kay Bailey Hutchison, who didn’t seek another term. About a quarter of the voters were undecided in that race.

The Lyceum’s poll page is here. The basic problem with this poll is demonstrated in the methodology, described on page 7 in their poll summary:

Texas Lyceum Poll
From September 10-26, 2012, The Texas Lyceum conducted a statewide telephone survey of registered voters. The survey utilized a stratified probability sample design, with respondents being randomly selected at the level of the household. The survey also employed a randomized cell phone supplement, with approximately 16% of completed interviews being conducted among cell phone only or cell phone dominant households. A Spanish-language instrument was developed and bilingual interviewers offered respondents a chance to participate in English or Spanish. On average, respondents completed the interview in 19 minutes. Approximately 6,500 records were drawn to yield 1,175 completed interviews. The final data set is weighted by race/ethnicity, age and gender to achieve representativeness as defined by the Texas specifications from the 2010 Current Population Study. The overall margin of error for the poll is +/- 2.83 percentage points.

The ballot questions were asked of a random subset of the overall sample. All told, 666 registrants were asked the ballot question (margin of error = +/- 3.80 percentage points).

Some numbers and analysis—including the ballot items—were produced with a screen for likely voters. Voters were deemed “likely” if they indicated that they were registered to vote, indicated that they were “somewhat” or “extremely” interested in politics, and indicated that they had voted in “almost every” or “every” election in the last 2-3 years. Overall, this screen produced 805 likely voters out of 1,175 registrants, 68.5% of registered voter sample. The overall margin of error for the survey of likely voters is +/- 3.45 percentage points. For the ballot items, 443 respondents out of the 666 were likely voters, creating a margin of error of +/- 4.66 percentage points.

In other words, it’s the same bizarre “likely voter” screen that we saw in the UT/TT poll of May 22, which is not terribly surprising since UT prof Daron Shaw is involved in each. When you take such a restrictive view of the electorate, you get a strange partisan mix, as you can see in the crosstabs, on page 41. The Lyceum sample is 53% Republican, 35% Democrat, and 11% independent. You also get an electorate that’s 67% white, but only 5% African-American (page 42), which is less than half what other pollsters have been projecting. Hell, if that’s what the electorate looks like in November then yeah, Dems are going to get stomped up and down the ballot. But for crying out loud, even GOP pollsters like Mike Baselice tend to estimate the GOP advantage in Texas at about eight to ten points. That Wilson Perkins poll from last month had it at 38% R, 31% D, 29% I. At the risk of being accused of poll un-skewing, if the Lyceum poll had the same partisan mix, the result would be Romney 50.2%, Obama 39.5%, undecided 10.3%; the reason for Obama’s poor showing largely comes from a pathetic 30% share of the Independent vote (Romney got 45% of indies, the rest were undecided). But even this result would portend an election similar to that of 2008, rather than a hard to imagine 2004 Bush-style whipping.

I’m not going to waste any more time on this poll. I am, however, now tracking all of the Texas polls that I become aware of, which you can see on the sidebar. We’ll know soon enough who’s right and who’s crazy. What really annoys me is that the rest of the poll, in which all of the “unlikely” voters were also asked, showed some fairly decent results for Democrats, including a near-even split on the Affordable Care Act and plurality support for expanding Medicaid. Unfortunately, unlike that earlier UT/Trib poll, they apparently didn’t ask the entire sample who they’d be voting for, so we have no basis for comparison there. Like I said, we’ll see what the voters actually have to say.

Related Posts:

5 Comments

  1. Dan McClung says:

    I find you correct in all your comments, Kuff. I, too, am encouraged by the “second” poll’s numbers regarding ACA and Medicaid expansion in the state. That may well produce some additional studies that give the Legislature something to chew on as Perry attempts to throw both programs under the bus over the coming cycle.

  2. Gary says:

    While I do not wish to indulge in the kind of poll denialism so beloved of Republicans (really, just another example of their extreme paranoia and conspiracy delusions), I too find myself more than a little skeptical on this one. Texas may differ somewhat from the national average, but it has tended to move in the same direction with national trends. Recently, polls on such subjects as the war, health care and gay marriages / civil unions show that this is true. The notion that the state has suddenly become an outlier, moving toward redder even as most of the country, including Southern states, are moving toward bluer would strain credulity, even if it were not the case that the Latino percentage of the voting age population creeps up year by year, or that the state becomes steadily more metropolitan. I wish we had several polling firms in the state, so that we could compare. In the meantime, only a fool lets polls keep him from doing what is necessary to turn the state blue.

  3. […] the Kuff deconstructed a truly crappy poll that was nonetheless accepted uncritically by the […]

  4. […] the Kuff deconstructed a truly crappy poll that was nonetheless accepted uncritically by the […]

  5. […] but we don’t know what the screened sample looks like. If it’s anything like that crappy Lyceum poll, with it’s 5% African-American share, you can see how things might get wacky. For what […]

Bookmark and Share