Democratic Primaries: Making Sense of the Texas Mess (Part 1)

n.b. This is Part 1 of 2. I am waiting for Part 2, largely so that I can work from updated pollster and poll aggregator bias calculations after South Carolina’s primary on Saturday. Update: no Part 2. South Carolina was ruinous. You can see what I think will happen in Texas now here.

There have been seven reputable polls in February ahead of the Longhorn Democratic primary next Tuesday, March 1. Four of them have come out in the last twenty-four hours. They are a hot Texas mess.

Screen Shot 2016-02-26 at 1.43.50 PM


Two polls (UT/Texas Tribune & Emerson) suggest that it’s a 10-16 point race in Clinton’s favor. This would be earth-shattering bad news for former Secretary of State Hillary Clinton. Two other polls (PPP & Survey USA) suggest that it’s a 23-29 point race in favor of Clinton. Pegging the middle course between the two would suggest that Clinton is ahead five or six points nationally (in keeping with Real Clear Politics average of polls), that Sanders is at least moderately competitive with Latino voters even when he doesn’t have the combination of time, organization, will, and resources for a serious outreach campaign in the state. An additional two polls (CBS and Monmouth) call it a 32-24 point advantage for Clinton which would mean a 12-15 point lead for Clinton nationally, a more serious but not necessarily dire deficit with Latino voters, and a faint hope that Sanders can keep his head above water past March 8. Finally, the Austin American Statesman calls it a 40 point race for Clinton in which case Sanders has zero shot at the nomination and would do well to fold ’em and send $27 refunds to as many people as possible.

So which is more likely correct?

Is Sanders down around ten, twenty, thirty, or forty?

Let’s did into the numbers, how the polling firms have done thus far this cycle, and the fundamentals of the race. Short answer: Senator Sanders should have put more resources into the Lonestar State long ago. If he had, he could have done surprisingly well. Since he didn’t, he’s likely down 25 points or more. (The Real Clear Politics average of polls pegs him at 26.3. Fivethirtyeight’s “polls only” forecast says he’s down 29.1; their “polls plus” forecast puts Clinton ahead by 36.6. If I had $100, I would bet the RCP average is closest.)

The most contentious issue on the Democratic side of the primary season is where the Latino vote is falling and will continue to fall. It’s a “data driven” dispute that may or not be answered definitively on Super Tuesday. In a way, Texas is a great bell weather since it has a massive hispanic population both proportionally and in terms of sheer numbers. Fivethirtyeight’s Nate Silver has admitted that, heretofore, his poll projections have lumped together black and brown voters with all other people of color in keeping with his narrative that Bernie Sanders could win a small handful of White and Liberal states and lose everywhere else. This is why fivethirtyeight’s “polls plus” projection differs so radically from the Real Clear Politics average.

In another way, Texas will not tell us as much as a combination of Nevada, Colorado, and Massachusetts because the Sanders campaign has invested almost nothing in Texas (see associated chart). It may well be that Sanders was down just 23 points or so in early February (per PPP), gained momentum by winning the Latino vote in Nevada (per Texas Tribune & Emerson), has now been obliterated by a strong advertising campaign and ground game by Team Clinton, and will lose by thirty points or so. In terms of advertising dollars alone, the Sanders campaign has devoted twenty times the resources to Colorado and Massachusetts as it has to Texas. In my interview with the Southwest Voter Registration Education Project’s (SVREP) Antonio Gonzalez, he suggested that Sanders’ lack of investment in Texas made some sense given the sheer geographic size of the state.

So where do Clinton and Sanders stand respectively in terms of Latino voters in Texas? That really, really depends what numbers you consult and may, in fact, be critical to the entire Democratic Primary cycle.

The first set of Texas specific numbers to consult are PPP’s. PPP was in the field for three days beginning with Valentine’s Day. They interviewed a remarkable 30% Latino voters out of 514 respondents by “using a blended methodology with automated phone calls to land lines and online interviews of cell phone only respondents.” Hispanic voters, astonishingly for those who follow fivethirtyeight, supported Sanders at a greater rate (-16%) than either the poll as a whole (-23% for Sanders) or White voters (-24% for Sanders).

It appears that PPP underpolled 18-45 year-olds as a whole at 39% compared to their numbers in Nevada, New Hampshire, and Iowa. The lack of a get-out-the-vote drive by Sanders in Texas, however, may mean that their numbers are severely dampened on Tuesday. It’s definitely a number worth watching on election night.  One of the major disputes from Nevada is whether Latino 18-29 year-olds could have made up 38% of the Latino vote there. According to Pew Research, Latino Millennials (18-34) will make up 44% of eligible Latino voters in 2016. In Nevada, 18-29 year-olds make up 37% of eligible Latino voters in Nevada and 32% of the same in Texas.

Will they turn out in the same or nearly the same numbers in Texas as they did in Nevada?

Likely not.

Beyond the lack of effort from the Sanders campaign, Texas is an open primary where ID is required and registration had to be completed a month in advance where Nevada was open, no ID, with the ability of voters to register online and onsite all the way up through the day of the caucuses. According numbers sent to me by Nevada’s Secretary of State, Nevada registered 12,500 new Democratic voters between January 1 and this Tuesday with around 3,000 of those voters registering on the day of the caucuses (see accompanying chart). These numbers support the claim by SVREP’s Gonzalez that the entrance poll numbers for Latinos in Nevada from last Saturday are totally legitimate and reflect the numbers of eligible 18-29 year-old Latino voters and what a good campaign can reasonably accomplish.

Nevada Voter Registration Jan - Today

Nevada’s “Active Voter Registration” numbers from left to right from January 4, 2016, Friday February 19, 2016, and Tuesday February 22, 2016

… to be continued. (or not)

Correction: a previous version of this post said Texas primaries are closed (meaning members of other parties cannot cross-over).


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s