The second of two polls from yesterday, both of which are interesting in their own way. The NYT story about the poll, which included results from Iowa (Biden leading by3) and Georgia (tied), is behind their firewall, so I’ll give you a tweet summary and then dive into the data, which is available to me. First, the tweet:
NEW @nytimes/@SienaCollege LV poll of Texas:
@realDonaldTrump 46, @JoeBiden 43
@JohnCornyn 43, @mjhegar 37
— Evan Smith (@evanasmith) 5:03 AM – 24 September 2020
The data for all three polls is here, and you can find the Texas results beginning on page 23. I will present the highlights here.
– The first question is about how likely you are to vote. The five responses (not counting Don’t Know/Refused) are Almost Certain, Very Likely, Somewhat Likely, Not Very Likely, and Not Likely At All. Putting aside what distinguishes those labels, every subgroup – including 18 to 29 year olds, Latinos, and any other group you might consider to be lower propensity – was over 90% for Almost Certain plus Very Likely. Democrats were 65% Almost Certain and 32% Very Likely, with Republicans 62% Almost Certain and 34% Very Likely, and Independents 61% Almost Certain and 30% Very Likely. At 91% for the sum of those two categories, Indies were the “least” likely to vote.
– The second question was about how you will vote: In person on Election Day, In person before Election Day (i.e., early in person), and vote by mail. Fifteen percent of voters overall said vote by mail, which is a lot more than what we’re used to, but shouldn’t be a total that will overwhelm local election administrators. For example, in Harris County in 2016, 7.3% of all ballots were mail ballots, so this would be double that as a percentage, slightly more in real terms since there will likely be more total votes. Putting it another way, there were 101K mail ballots in Harris County in 2016, for turnout of just under 1.4 million. If we have 1.5 million votes, and 15% are mail ballots (the “Houston” region subgroup had 14% saying they would vote by mail), that’s 225K mail ballots. I don’t believe that will cause any major problems in processing.
(The Quinnipiac poll had 13% of respondents say they would vote by mail. That poll is a bit goofy as we’ve discussed, but these two numbers largely agree with each other.)
– The two subgroups that say they will vote by mail the most were those 65 and older (33%, and no surprise) and the 18 to 29 year olds (19%), which I’m going to guess will be a slight overestimation in the end. Democrats (16%) planned to vote by mail more than Republicans (12%), but not by much. However, Dems will be voting early overall more than Republicans – 57% early in person plus 16% by mail for Dems, to 51% early in person and 12% by mail for Republicans. If this is accurate, we could have a bit of a “red shift” on Election Day, which is very much what happened in Harris County in 2008 – Dems voted so heavily during the early period that there just weren’t as many left to vote on Election Day. Something to keep an eye on, especially if various Dem hopefuls have an early lead.
– The list of candidates included the Libertarian and Green nominees in the Presidential race, each of which drew one percent, but just the Libertarian in the Senate race; he took four percent. Both questions allowed the respondent to volunteer that they were voting for someone else, but in each case the number for that was zero percent; a couple of subgroups in each reached one percent for Someone Else. In 2016, the “other” candidates received a collective 4.52% of the vote in the Presidential race.
– Biden carried Democrats 91-2, while Trump won Republicans 93-5. Six percent of Democrats said “Don’t know”, with one percent each specifying the Libertarian or Someone Else. Only two percent of Republicans said they didn’t know, and none gave any other answer. Black respondents were at 20% for Don’t Know, and Latinos were at 8; given that Black respondents went for Biden 71-7 and Latinos went for him 57-32, it seems likely that Biden’s overall totals are a bit lower than they will be in the end. Biden also carried indies by a 41-37 margin.
– There were five regions given as subgroups: Austin/San Antonio/South (presumably South Texas), Dallas/Fort Worth, Houston, Minor, and Rural. No, I don’t know what “Minor” means or how it is distinguished from Rural, nor do I know what specific counties are in the first three groups. Here’s how this shook out:
Candidate Aus/SA/South D/FW Houston Minor Rural ======================================================= Biden 50% 47% 51% 34% 25% Trump 38% 34% 36% 58% 71% Others 3% 3% 3% 2% 0% Don't know 8% 16% 10% 6% 3%
Seems clear where most of the Black and Latino vote is coming from, not that this is a surprise. Given that, these numbers would seem to portend very well for the various legislative and Congressional Democrats in those regions. I wish I knew more about this so I could try to do some kind of comparisons, but I don’t. Sorry.
– The Hegar/Cornyn numbers largely recapitulate the Biden/Trump numbers, with Hegar having slightly softer numbers among Dems and groups that tend to vote Dem than Biden does. She’s 81-6 among Dems (Cornyn is at 84-6 among Republicans), with 2% for others an 11% Don’t Know. Black voters go for her 66-9, but the Libertarian candidate gets six percent with another 18% on Don’t Know. Latino voters are 52-32 for Hegar, with 13% Don’t Know. As I’ve said multiple times, I think this race will closely mirror the Presidential race.
That largely covers it, and for more you can read Nate Cohn’s Twitter thread, in which he adds some thoughts. In particular, talking about the likely voter model, “Texas is a state where turnout is particularly uncertain and the upside is likely on the side of Mr. Biden”.