What is the point of this Matthew McConaughey poll?

I have questions about this.

Matthew McConaughey commands more support to be Texas’ next governor than incumbent Greg Abbott, according to a poll released Sunday by The Dallas Morning News and the University of Texas at Tyler.

However, the film actor and political newcomer could hit potholes in either major party’s primary if he enters next year’s governor’s race, the poll found.

For months, McConaughey has teased political pundits and TV talk show hosts with musings that he might enter politics in his home state.

If he were to take the plunge and run for governor, the poll found, 45% of Texas registered voters would vote for McConaughey, 33% would vote for Abbott and 22% would vote for someone else.

McConaughey’s double-digit lead over the two-term Republican incumbent is significant. The poll, conducted April 6-13, surveyed 1,126 registered voters and has a margin of sampling error of plus or minus 2.92 percentage points.

But 56% of Republican voters said they’d vote for Abbott, compared with only 30% for McConaughey.

While Democrats broke 66% to 8% for McConaughey, and independents 44% to 28%, more than twice as many Democratic primary voters — 51% — said they wanted a progressive candidate for governor than wanted a centrist — 25%.

That could pose a problem. McConaughey, who has criticized both major parties, has suggested he’s more of a moderate.

And in the GOP gubernatorial primary, that’s also not obviously a ticket to success. Solid majorities of poll respondents who described themselves as conservative, evangelical or retirement-age Republican primary voters said they’d vote for Abbott.


Jason Stanford, who managed the campaign of second-place finisher and Democrat Chris Bell in the 2006 gubernatorial race, said McConaughey poses no threat to Abbott.

“There doesn’t appear to be a huge groundswell of discontent for Abbott,” Stanford said. Once McConaughey declares as a Democrat or Republican, reality will set in with Texas voters, he added.

“If you ID as a Democrat or a Republican, you’re going to get different answers about him in polls,” Stanford said. “He’s fun, but once you put him in a political context, things will change.”

Poll details can be found here. There’s some issues and approval polling that I’ll get to in a separate post and which is actually kind of interesting, but as for the Abbott/McConaughey question, the only thing you need to read is what Jason Sanford said, because he’s 100% correct.

The first problem with this poll question is in the question itself, which is worded as follows: “Matthew McConaughey has been talked about as a potential candidate for Governor of Texas. If he ran, would you be likely to support him more than Governor Abbott?” Do you see what’s missing in that question? It’s any mention of what (if any) party McConaughey would be claiming. If he’s running as a Democrat against Abbott, then there’s no way in hell he gets 30% of Republicans to support him. Even getting ten percent would be seismic and likely enough to win, but we can’t tell what kind of actual crossover appeal he might have because the question is asked without that piece of information, leaving the respondent to assume that this is some theoretical, non-partisan race. You know, the kind that we don’t have for state elections.

If McConaughey were to run as an independent, then this would need to be polled as a three-way race, because the Democrats would surely have a candidate as well. One could possibly imagine a scenario in which McConaughey mounted an independent campaign and the Texas Democratic Party decided as a tactical matter to support him, the way Dems have supported independent candidates for Senate or Governor in Maine and Kansas and Alaska in recent years. The problem with that scenario is that while McConaughey could announce his independent candidacy now and start staffing up for it, he can’t begin the petition process to get on the ballot until after the primary election, or after the primary runoff if there was one for Governor, and there’s nothing to stop someone from filing to run as a Democrat in the primary in the meantime. Any Democratic nominee, whether a candidate who might be viable against Abbott on their own or a more marginal type who still has appeal to some part of the Democratic base, will draw enough support to make an independent far less competitive in the general. To put it another way, it’s extremely unlikely Matthew McConaughey gets 66% of the Democratic vote in a three-way race.

Maybe I’m wrong about these assertions. You could ask again and name McConaughey as the Democratic nominee, and see how much Republican support he gets. You could also ask about a three-way race that features Abbott and McConaughey and an actual, named Democrat. And if you’re going to do that, why not also ask the horse-race question about just Abbott and that same Dem? Why not ask the Abbott-versus-Beto and/or Abbott-versus-Julian question, which would allow a comparison to McConaughey as a Dem, then ask again with McConaughey in there as an independent? We all understand that at this point in the calendar all these questions are mostly for funsies, but with some useful information in there if you know how to look for it. At least the Abbott/Beto or Abbott/Julian questions would give a data point about whether Dems have any cause to feel optimistic or not, and the three-way race question might tell us something about how much Republican support for Abbott is softer than it looks. Any of it would tell us more than the actual question did.

And of course, if McConaughey were to run against Abbott in a Republican primary, then asking this question in a sample that includes more non-Republicans than Republicans is going to give you a nonsense answer. Point being, if I haven’t beaten it to a sufficiently bloody pulp yet, identifying McConaughey’s partisan affiliation in this question matters. Not including it makes this whole exercise useless for anything that blog fodder and Twitter posts. Which they got, so mission accomplished.

One more thing, before I end this post and write the other one about approvals and issues polling: For some reason, the sample – which as before is partly phone and partly web panel, and all made up of registered voters – voted in the 2020 Presidential election as follows:

Trump – 36%
Biden – 32%
Other – 1%
Did not vote – 30%
Refused to say – 1%

If you’re thinking that’s an awfully large “did not vote” percentage, consider how the sample from their March poll answered the same question:

Trump – 43%
Biden – 38%
Other – 4%
Did not vote – 11%
Refused to say – 4%

Why so different? I have no idea. Why do we think we can draw reasonable conclusions from a poll sample that includes such a large number of people who didn’t vote in the highest turnout election in Texas history? Again, I have no idea. To be sure, the 2022 election will have smaller turnout, and an RV sample is all that makes sense at this time. But maybe weighting the sample a bit more towards actual voters might make any projections about the next election more accurate.

Related Posts:

This entry was posted in Election 2022 and tagged , , , , , , , , , , , , . Bookmark the permalink.

6 Responses to What is the point of this Matthew McConaughey poll?

  1. brad says:


    I believe you are mistaken on when McConaughey can begin petitioning to be an independent candidate.

    His petition team can begin petitioning on the date of the primary and continue on until the petition deadline. Its just that folks signing the petition can’t vote in either the Rep or Dem primary, or any run off, or the signer invalidates their petition signature.

  2. Brad,

    We’re in agreement about that. My point is that McConaughey can’t guarantee that he will have sufficient petition signatures before the primary, which makes a scenario where the TDP agrees to back him as an indie that much less likely.

  3. Kibitzer says:


    Like the recent UH Hobby School polls, this one covers a lot more ground than is the focus of the current story. There is much to agree here with Kuff’s analysis regarding a possibly forthcoming McConaughey candidacy, but there are also some more fundamental concerns regarding the sample and the accuracy of the reported results that extend beyond the questionable utility of the McConaughey item.


    As for the former, a random sample of registered voters (as distinguished from voting-age population) would be great for the purpose of ascertaining the distribution of relevant attitudes going into the next election season, and a sample size of 1,126 is of decent magnitude.

    If respondents were randomly drawn from the actual statewide registration database, the demographic breakdown, voter participation in 2020, and voting patters (D/R breakdown) could then be compared to the known parameters of the statewide electorate to validate the quality (representativeness) of the sample to double-check representativeness.

    Indeed, it would in theory be possible to verify the respondent’s history of participation / nonparticipation in past general and primary elections at the individual level (though not their actual vote choice in general elections, given the secrecy of the ballot).

    What we have here, however, is not a random sample, as becomes clear from the methodology section at the end of the 16-page report (Kudos to Kuff for posting the link). The sample was much rather assembled from two different sources and adjusted to match known parameters of the Texas electorate (i.e., stratified); and statistical weighting was also used. Under these circumstances, additional caveats apply, and the reported margin of error is not as useful as would be the case with a true randomly drawn sample.

    That said, many of the reported differences in the cross-tabulations are quite substantial so as to mitigate concerns over a larger margin of error and other sources of imprecision. Most noteworthy are the huge partisan differences in public approval ratings of national and state level politicians (Biden, Abbott), and crass disparities in specific areas of public policy (though not really surprising).

    And the parallel phrasing of questionnaire items regarding job approval for different leaders (Abbott, Patrick, Paxton) allows for meaningful comparisons even if the sample is not fully representative of the population of registered voters. These dis/approval findings are also more meaningful than the responses regarding a hypothetical candidates such as McConaughey because people know more about them, given that these politicians have a track record and are — for better or worse – the subject of ongoing media coverage.

    The findings regarding specific policy attitudes are fascinating and would merit in-depth articles or blog posts of their own, including COVID mask-wearing and vaccination-related items, gun control, election “reform”, and abortion. The data on the winter storm power and water supply crisis, and preferences for remedies, could be compared with the UH Hobby School poll.


    As for accuracy of reported results, at least one questionnaire item raises a red flag: On the current employment status question, the “Retired” rate is reported as 27% for “Total”, but as 1% and 9% for Men and Women, respectively. How can that be? On the ‘gender’ item, 51% are shown as Female, 48% as Male. 1% went missing even though the total N on the table for gender breakdown matches the sample size. There are no other gender categories and all respondents are above age 18 by virtue of sample frame. The incongruence in the employment-status table could be a clerical error limited to the particular table, but it doesn’t inspire confidence in overall accuracy of the reported results.

    As to the age variable, the brackets are inconsistent ranging in breadth from 6 years for the youngest segment of registered voters to more than 20 for the oldest. The largest age cohort accounts for 34% of Total, but encompasses a range of 19 (or 20, depending on precise definition) years (45-64) while the two younger cohorts below it each encompass only 9 (or 10). There are five age categories in all, but they are neither equal in the width of the intervals in years, nor are they quintiles.


    As for race and ethnicity, there were a total of 8 response categories on offer (including ‘Other’ and ‘More than Two Races’ [though not ‘More than One Race’ or ‘Mixed Race’], but the tables report the break-down only for Hispanic, Black, and White (in that order). Nothing on Asians and indigenous Americans, presumably because of small numbers and related statistical-significance concerns.

    Hispanic, of course, is also not a race, and Hispanic White, Hispanic Black, and Hispanic Mestizo were apparently not available choices for self-identification. Nor does it appear that “White, Non-Hispanic” were permitted to identify their ethnic heritage as Anglo, Irish, Scotch, Italian, German, French, Polish, Jewish, etc.., not to mention multiple/mixed lineages.

    Disaggregation for the Hispanic category is provided as follows: Mexican [comma] Mexican American (64%), Tejano, Puerto Rican, Cuban, and Other. Interestingly, the ‘Other’ category comprises 23% and is larger than the three non-Mexican categories taken together (13%).

  4. C.L. says:

    I’d vote for anyone other than Abbott at this point. Exclusions ? Dan Spacetrippin- Patrick, Ken under Indictment-Paxton, Ted Cancun-Cruz, or John Sergeant Schultz-Cornyn.

    Do I think McConaughey would make a good TX Governor ? Not in the least.

  5. Pingback: More interesting questions from that Matthew McConaughey poll – Off the Kuff

  6. Pingback: Texas blog roundup for the week of April 26 – Off the Kuff

Comments are closed.