Another new poll of Texas, from YouGov:
Republican Presidential hopeful Mitt Romney holds a solid, eleven-point lead over Democratic President Barack Obama in Texas, 52% Romney to 41% Obama, in a poll of 1,254 registered voters statewide, conducted by YouGov.
In the race for Senate, Republican Ted Cruz holds a 50%-31% over Democratic candidate Paul Sadler in the race to replace the retiring Sen. Kay Bailey Hutchison.
[…]
Sampling method: Respondents were selected from the YouGov’s panel using sample matching. A random sample (stratified by age, gender, race, education, and region) was selected from the 2005–2007 American Community Study. Voter registration, turnout, religion, news interest, minor party identification, and non-placement on an ideology scale, were imputed from the 2008 Current Population Survey Registration and Voting supplement and the Pew Religion in American Life Survey. Matching respondents were selected from the YouGov panel, an opt-in Internet panel.
Weighting: The sample was weighted using propensity scores based on age, gender, race, education, news interest, voter registration, and non-placement on an ideology scale.
Number of respondents: 1,254 registered voters statewide.
Their poll data are here. For those of you thinking that Internet-based polls are necessarily bogus, let me refer you to Steve Singiser:
It would seem appropriate, given that they were solely responsible for half of the data collected today, to say a few words about YouGov. The firm is a British firm, that moved into the American market in 2007 when they bought out Polimetrix, a California-based firm that had done a lot of internet-based polling in 2006 in an arrangement with Stanford University.
Their polling is based on internet samples, a method which some find problematic (some aggregators of polling, indeed, refuse to utilize their data). It is a methodology that I also confess to qualms about, because when you have a sample that is essentially volunteering to participate, and a smaller universe from which to draw from, the potential pitfalls are pretty self-evident.
However, the true measure is performance. Some internet-based polling has a track record of missing the fairways. Indeed, the only poll that is specifically barred from inclusion in the Wrap is an internet-based survey: polls from JZ Analytics (once called “Zogby Interactive”). John Zogby is an established veteran who has some impressive hits on his resumé, but his foray into internet polling was pretty awful. 2006 was a particular low water mark, whether it was the insistence that Democrat Bill Ritter was enmeshed in a coin flip for the Colorado governor’s race (he won by 17), or that Herb Kohl was being hard-pressed for re-election (he wound up winning by 38). The lack of movement in the polls also made clear that Zogby was trying to make do with what were very small collections of voters in each state.
YouGov, however, has earned at least a cycle’s worth of benefit of the doubt. Their 2010 track record was more than reasonable. Indeed, of the 18 pollsters that offered up a substantive number of polls, YouGov came in fourth place in terms of their accuracy (defined as the percentage of races where they came within three percent of the final margin).
So we’ll see how we do. It would be nice to have another outfit regularly polling Texas, if they’re halfway decent at it.
My thoughts on the poll data:
– Because of their weighting, I can’t say for sure what they project the racial makeup of the electorate to be. Whites make up 74% of their actual sample, and I’m pretty sure no one expects that. I fiddled around with the numbers a bit, and if I set the white share of the vote to be about 67% and distributed the difference to black and Hispanic voters in proportion to their actual shares, I get pretty close to their totals. All pollsters make guesses about the makeup of an electorate, so there’s nothing particularly odd about this exercise.
– White voters go for Romney by a 66-26 margin. Assuming the 3% of “not sure” voters are unlikely to vote, and that the “other” vote is probably overstated at 5%, that puts Romney close to 70% of white voters once you reallocate. That’s higher than what pollsters other than Wilson Perkins have found, and if true it puts the GOP in the safe zone for continuing to win statewide.
– Obama leads among Hispanic voters by a 58-36 margin. That’s a smaller lead than what other pollsters have found, and it feels understated to me. One reason why I think this is that the poll sample is heavily female (55%), and Hispanic women are among the strongest supporters of President Obama nationwide. Perhaps they’ve weighted out the gender disparity, and perhaps Texas Latinas aren’t like their sisters elsewhere, but still, this feels understated to me.
– The results by age are almost the complete opposite of the Wilson Perkins poll, which had Obama doing best (though still losing) among respondents 55-64 while getting stomped among younger voters. In the YouGov sample, Obama wins the 18-29 and 30-44 age groups, but gets creamed among older voters. Make of all that what you will.
– The Senate result is nearly identical to the poll released by the Sadler campaign, though obviously without the push component. I strongly suspect that a lot of the “don’t knows” in this result are Democratic voters who just don’t know much about Sadler but will ultimately vote for him. 23% of black respondents and 32% of Hispanic respondents were “don’t knows”. In the end, I suspect Sadler’s percentage will mirror Obama’s fairly closely. Ted Cruz didn’t do any better among Hispanic voters than Mitt Romney did, by the way – Sadler led him by a 36-31 margin.
That’s all I’ve got. I’m a bit dubious of this poll, but as Singiser said, they’ve earned some benefit of the doubt based on their 2010 track record. I’m glad we’re getting more data. We’ll see how good it turns out to be.
