Off the Kuff Rotating Header Image

Yet another wacky poll about Latino voting in Texas

Just when I thought I was out…they pull me back in.

Not Ted Cruz

Not Ted Cruz

Sen. Ted Cruz lost the Hispanic vote in Texas by about 20 percentage points, but out-performed GOP presidential nominee Mitt Romney, according to post-2012 election polling shared with the Washington Examiner.

Cruz defeated Democrat Paul Sadler last November by nearly 16 percentage points on his way to garnering 56.5 percent of the vote. Romney bested Obama by nearly 17 percentage points while garnering 57.2 percent of the vote. But a survey taken about six weeks after the election and made available by the Republican senator’s political team, shows Hispanics favored Sadler over Cruz 60 percent to 40 percent and Obama over Romney 59 percent to 33 percent.

The survey offers a unique window into voting patterns of Hispanics, the nation’s fastest growing voting bloc, in conservative-leaning Texas, where exit polling is hard to come by given its reliable Republican voting record in statewide races and the high cost of gathering data in such a large state. With Cruz, son of a Cuban immigrant, exploring a 2016 presidential run, the poll could shed light on how the senator compares with other Republicans when it comes to winning Hispanic votes.

[…]

The poll, conducted by Cruz pollster Wilson Perkins Allen Opinion Research, surveyed 601 Texas Hispanics who voted in the 2012 general election, and has an margin of error of 4 percentage points.

I don’t know if this horse is dead yet, but if it isn’t it’s definitely in the ICU. A few basic observations, which I will try to keep fresh, since even I’m getting tired of this:

– There’s no polling data available for inspection, so it’s that much harder to do any intelligent analysis of this. Wilson Perkins did a Presidential poll in September and released all their data for it, so it’s not like they have a track record of secrecy. Yes, I know, polls commissioned by candidates/officeholders often have proprietary information in them, but they’re also often released in this manner to advance an argument. Mike Baselice does this a lot. I’m not saying this makes a poll automatically suspect, but you are ultimately taking the pollster’s word for it, and when a pollster has been hired by a partisan, that partisan has an interest in what is taken away from the information provided. Caveat emptor, is what I’m saying.

– That claim that the Latino vote divided 60-40 between Cruz and Sadler is suspicious for several reasons. For one, I’ve never seen a poll result where the totals for two candidates added up to 100%. Even in a runoff situation, there will be some “don’t know” and “refused/no answer” type responses. In addition, as we have seen many times before, the percentage of people who have expressed a preference in the Cruz/Sadler race was considerably less than the percentage that expressed a preference in Romney/Obama. It beggars belief that literally everyone picked either Cruz or Sadler. This number alone makes me want to disregard the entire poll.

– As I have pointed out before, claims about Latino voting in Texas can be checked against actual results, and deviations from the Latino Decisions poll last November can generally be shown to be suspect based on that. It’s also worth noting that claims about a certain level of preference for Latino voters affects what the numbers would have to be for Anglo voters as well. For example, if you assume that the racial/ethnic breakdown in Texas was 70% Anglo, 20% Latino, and 10% African-American, then to say Ted Cruz got 40% of the Latino vote implies he must have gotten about 68% of the Anglo vote in order to get 56.5% of the vote overall. Note that back in September, Wilson Perkins claimed that Mitt Romney was polling at 77% among Anglo voters in Texas. How do they explain that disparity? Everyone agrees that Cruz did a little better than Romney did among Latino voters in Texas, which therefore implies that Romney did a little better than Cruz among Anglo voters – you can see that in the data if you compare the county by county results for each – but not nine points better. Which one of their polls would Wilson Perkins say was inaccurate?

– It’s not quite as easy to draw conclusions about the Anglo vote in Texas from precinct or county results, but a look at the most Anglo State Rep districts in Texas suggests that Cruz did better than 68% overall among that demographic:

Dist Anglo% Cruz Sadler Cruz% Sadler% ================================================ 2 81.4 43,359 13,782 73.9% 23.5% 19 80.7 48,200 15,964 73.4% 24.3% 21 80.0 46,050 17,057 71.5% 26.5% 58 80.4 41,720 12,225 75.0% 22.2% 60 84.7 51,821 11,081 80.1% 17.1% 61 85.8 54,602 11,591 80.2% 17.0% 62 82.2 38,182 14,041 70.8% 26.1% 98 80.9 56,907 17,802 73.9% 23.1%

You get into fuzzy math very quickly, so I don’t want to spend much time on this. The Anglo% is Voting Age Population, which is likely to be less than Citizen Voting Age Population, and is also likely to be less than the share of the actual electorate. Unlike Latinos, who voted fairly consistently for Democrats, there are a couple of places in the state that are heavily Anglo but not very Republican – specifically, much of Austin and some place in Houston like Montrose and The Heights. HD48, for example, is 69.5% Anglo VAP, but voted only 37.9% for Cruz. (See the election numbers here and population figures here.) There’s also a lot more Anglo voters than there are Latino voters, so you can’t get nearly as big a total by looking at the most heavily Anglo districts. All that said, there’s nothing here to suggest Cruz got only 68% of the Anglo vote, and as such there’s nothing here to suggest he got as much as 40% of the Latino vote.

– Along the same lines as above, the cited figure of 32% Latino support for Romney is reasonably in line with other data, and implies a level of Anglo support in the 71-72% range, which strikes me as being plausible. I could buy 35% for Cruz – bearing in mind that some number of people thought Cruz was the Democrat in the race – but 40% is just a stretch. I’ll say it again: Show me the math if you want me to believe. Link via Burka.

Related Posts:

2 Comments

  1. M. Incandenza says:

    The results in El Paso, Webb, Hidalgo, Cameron, and every other border county are consistent with Sadler winning the Hispanic vote by at least 2-1. Sadler also won Bexar by 4%, which seems like it would be pretty impossible if Cruz was getting 40% of Hispanics.

    Of course, it’s theoretically possible that hispanics in other parts of the state are much more conservative than those in San Antonio and the border, but if that’s the case I haven’t seen the evidence for it.

  2. Mainstream says:

    Also possible that the voting behavior of a Hispanic person living in Woodlands, River Oaks, Memorial , Cypress, Clear Lake is quite different from one living in an inner city, majority Hispanic state house district or one down in the Valley. But I have no better data.