Whenever there is a new poll of Texas released, there are a ton of hot takes on Twitter. Old believers of Texas as the great blue whale for the Democrats move to dismiss the poll, saying that Texas has looked good for Democrats in the past, but that they just can’t seem to pull it off. There are others who say that the numbers are real, and are a result of inevitable demographic shifts. Others dismiss Texas numbers as not mattering, because if Texas is close, surely the election is already won for the Democrats.
So let’s look at all of these arguments, why they are right/wrong, what the actual contents of the poll (including the crosstabs, which get very little attention) are saying, and how you can extrapolate that into the broader electorate.
The first argument of new Texas polls, is that polls showing a small Biden lead now is wrong, and Trump will flip it back when he gains in the polls/ when likely voter screens are more prevalent. I’ve written about Likely Voter screens before, and why they may not hurt the Democrats as they have in past years, so I won’t write about that now, so instead I’ll talk about the first argument, that a small lead will not hold. Firstly, polls have underestimated the Democrats in Texas in 2016 and 2018, particularly in the 2018 Senate race, where Republican Senator Ted Cruz was expected to win by high single digits, only to cling to a ~2% win. Additionally, if you only believed the polls, there would be no Democratic representatives in TX-07 and TX-32, as both were polled by the NYT/Sienna, showing small GOP leads, along with a large lead for Will Hurd in the TX-23. Both the TX-07 and TX-32 were won by over 5 points, and the TX-23 turned into a nail biting finish on election night, which has (probably, at least in part) lead to the retirement of Will Hurd in 2020. Other people dismiss those numbers because they expect Trump to claw back some of his losses close to November. The problem with this assumption is that it is the same working assumption that analysts have had since Joe Biden won the nomination, and at every point it has yet to materialize. Since Biden locked up the nomination the pandemic has only gotten worse, and Trump has done nothing but lose ground almost every month. While there is likely a floor for the GOP in modern American politics, and while we are *probably* approaching that, there is no reason to think that floor doesn’t include a loss in Texas.
Other people will argue that the result in Texas doesn’t matter, as it is not a swing state. This argument is mostly correct, as even if Joe Biden were to win Texas in November it would not yet be a swing state, as it would still vote to the right of the nation even in a Biden win, and would surely not be the reason Biden would win the election. However, Biden winning Texas would be an important step for Texas’s inevitable journey on its way to swing state status. Given the swings in 2016, 2018, and very likely 2020, it is only a matter of time until Texas (along with Georgia, Arizona and North Carolina) are given the same important status as Pennsylvania, Wisconsin and Michigan. While that will certainly not happen in 2020, depending on how well Biden is able to do in Texas, that could happen as early as 2024, or even 2028.
So now let’s look at what the actual poll of Texas is saying in the crosstabs. The poll I will highlight is the Quinnipiac poll of Texas released earlier this week. Quinnipiac is one of my favorite pollsters because it breaks down the demographics in a way that is very easy to access and read. When looking at a Texas poll, the first thing I look at is the White voter crosstabs. Again, Quinnipiac is very good with splitting White voters by education, which is crucial to understanding what the poll is saying. Although Biden has made massive strides with White college voters since Clinton in 2016, and is leading them nationally, he is still losing them in Texas by 7 points and losing White non-college voters by a massive 45 points. However, since Texas is a right leaning state, it makes sense that these White voters are more inclined to vote GOP than White voters nationwide. Using the 2016 CNN exit poll (and the 2018 Senate CNN exit poll) as comparison, Biden is at 42% of White college voters, a negative from Beto’s 44% but an improvement from Clinton’s 31%. For White non-college voters, Biden is at 23%, an improvement from Clinton’s 21%, but a net negative from Beto O’Rourke’s 26% in the 2018 Senate election. Now of course, all of these numbers are rough, and we should not take any exit poll as completely accurate, but it is still a good starting point for an analysis like this. On the surface, these numbers make a lot of sense, as Biden is clearly crushing it in the suburbs, so it makes sense he is doing the best of the three of them with White college voters. However, the White non-college voters are somewhat contradictory. This Quinnipiac poll (and many others recently from Texas) show this same trend, of Biden doing barely any better than Clinton and worse than Beto with this key group. Normally, people would dismiss this as part of the global realignment for voters with lower levels of education turning to conservative parties, but this doesn’t line up with the other polling for Biden across the country. The reason Biden has massive leads in the Rust Belt, along with making Ohio and Iowa competitive is because of his strength with White non-college voters nationally, the same group that is apparently abandoning him in Texas. So this suggests two possible explanations. Either Biden is not doing as well with White non-college voters as it seems, or two, Texas polls are systematically understating Biden’s support with these voters. Although the first option is very alluring from an analysis perspective, it is not borne out in polling from literally any other state. Biden is experiencing swings toward him with White non-college voters in the entirety of the Midwest, along with Florida, North Carolina, Georgia, Arizona, etc. So either, somehow, Biden has a very specific White non-college problem in Texas, or option two is correct, and Biden is doing better with this group than his polling is suggesting.
The second group to look at is the minority voters, specifically Black and Hispanic voters. In this poll, Biden is winning Black voters 89-6, and Hispanics by a much smaller 53-29 margin. The CNN exit poll has Clinton with 84% of the Black vote, and Beto with 89% of the Black vote, so this crosstab checks out, which also suggests a minor swing against Biden with this group given the national environment. Concerning for Biden though is the Hispanic numbers. With only 53% of the Hispanic vote, he lags Beto’s 64% and Clinton’s 61%. The key point to mention is that of the Hispanic crosstab, 6% were undecided, and 4% said they would not vote, and should have been dropped from the sample. So when distributed, Biden’s numbers would be more in line with Clinton and Beto’s numbers. Interestingly though, Trump trails his own numbers from 2016 (34%) and Cruz in 2018 (35%). The question becomes, why is Biden struggling with Hispanics? Specifically in the state that is ground zero for Trump’s border wall? Hispanics do skew younger, so Biden’s lower numbers with Hispanics could be a reflection of his struggles with younger voters who would much prefer to be voting for Bernie Sanders, but this clearly represents an opportunity for the Biden campaign, as if he can convert those same voters that voted for Clinton and Beto, suddenly Texas is not longer a small Biden win, but a likely gain.
Additionally, the Texas poll included numbers for the Senate race, as Democrat M.J. Hegar looks to take down incumbent Republican Senator John Cornyn. Most people almost exclusively look at this race as Hegar doing worse than Biden, which the topline numbers do agree with. Their argument is that while Texas hates Trump’s version of conservatism, it is still fertile ground for the GOP down ballot. To those people, I think Will Hurd (TX-23), John Culberson (TX-07) and Pete Olson (TX-32) would like to talk to you. “Ok” they might say “But outside of those districts Texas is still really red”. To which I would say “Literally Ted Cruz doesn’t agree with you”.
The Senate race is interesting as it shows Cornyn outrunning Trump by about 2 points, but a large number of undecideds. While Hegar does roughly as well as Biden with White voters (hilariously only one point worse with White non-college) Hegar is massively underperforming Biden with Black (73%) and Hispanic (41%) voters, of which Cornyn is almost tied with Hispanic voters at 38%. The issue is that the sample had 15% of Black voters undecided, and 17% of Hispanic voters undecided, for a total of 14% undecided. While looking at those crosstabs, and given how closely Cornyn’s support tracks to Trump’s, it is not hard to imagine that the majority of those voters are Biden voters, of which many of them likely cast votes for Beto in 2018. In this way, the real story of the Senate race is that there is a large pool of minority voters in Texas who are likely voting for Biden, but are undecided as to vote for either Hegar or Cornyn for Senate. If they split, Cornyn easily walks to re-election. If they split similar to their Presidential preference, then suddenly this race will look very similar to the 2018 Senate race, with the possibility of Biden helping drive more voters to Hegar down the stretch.
With Texas, the size of prize drives a lot of hot takes and bad analysis. It is important to look at the demographics of Texas, and what the crosstabs of the poll are saying with each new release. While much of the focus is placed on the toplines, this article highlights many stories that are just under the surface, waiting to be read.
Robert Martin (@RobertMartinLT) is founder and CEO of LeanTossUp.ca and a contributor to Decision Desk HQ.