Lizzie Fletcher
The U.S. House race between GOP incumbent John Culberson and Democratic challenger Lizzie Pannill Fletcher is generally expected to be closer than most in this traditionally Republican enclave of west Houston and the Harris County suburbs.
Now an internal Democratic poll of the 7th Congressional District shows it to be a statistical tie. The poll of district voters, released Friday by the Democratic Congressional Campaign Committee, found Fletcher within 2 points of Culberson, 45 percent to 47 percent. That is within the poll’s 4.9 percent margin of error.
[…]
The DCCC poll shows Fletcher leading Culberson by 8 points among women (50 percent to 42 percent), 20 points among independents (52 percent to 32 percent), and by 28 points among voters under 50 (57 percent to 29 percent).
Further proof that that the district could be in play: The poll found that a generic Democrat is within striking distance of a generic Republican – 46 percent to 47 percent. That’s tighter than the difference between Fletcher and Culberson, but still within the margin of error.
The Democratic poll also gave Culberson a net-negative favorability rating, with 32 percent of voters having a favorable view of the congressman, compared to 39 percent who don’t. Similarly, the poll found that 35 percent of voters approve of Culberson’s job performance, while 39 percent disapprove.
Meanwhile, Trump also remains underwater in a district, which he lost by 1.4 points in 2016. In the DCCC poll, 50 percent of Seventh District voters disapprove of his job performance, while 42 percent approve.
I first heard about this poll via G. Elliott Morris’s Twitter feed, but this story adds some details. Internal polls are generally treated with skepticism – scroll down to see the responses to that tweet for a couple of examples – and I want to talk about why that is first. The main reason why internal polls are looked at differently is because when an internal poll is released, you have no way of knowing how many other polls that particular campaign or committee might have done that they did not choose to release. In other words, the poll that gets released may be the most favorable of the bunch, cherry-picked to present a sunny view of the situation. Media and tracking polls are public, with all their results out there to be seen, so when there’s an outlier it tends to stand out. You just don’t know if an internal poll is an outlier or not.
The other reason why internal polls are different is that they are sometimes used for specific purposes like testing a message or attracting financial support. Polls that take the measure of a race, then “inform” the respondents about one of the candidates and re-ask the original question again at the end, are a common example of this. The Justin Nelson poll from December is in this category. There’s nothing wrong with this – it’s a valuable campaign tool – but since the result comes from an idealized scenario – in a real campaign, both candidates get to “inform” voters, assuming they have the resources to do so – these polls are not very useful as predictive tools.
For those reasons, and because full poll data is often not available, poll aggregators and election modelers tend to give internal polls less weight. All that said, this poll is an example of one we can probably take more seriously. For one thing, given that the runoff was less than two weeks ago, there very likely have not been any other polls done by the DCCC since Fletcher became the nominee. There’s (again, probably) nothing to cherry-pick from. The DCCC, which has now added Fletcher to its Red to Blue group, generally doesn’t try to convince funders to invest in a particular race, and for them to want to include CD07 as a race to target they’ll want accurate horse-race numbers. None of this means that they couldn’t have made optimistic assumptions about turnout or the makeup of the electorate – we don’t have the internal poll data, so who knows what they sampled from – but all pollsters have to make those judgments.
All things considered, I believe we can take this poll more or less at face value. Which is to say, it’s a data point, and we hope to see more of them to get a fuller picture of what may be happening. Given that, the way to think about this is not just for this race, which we believe will be close and competitive, but for how it fits into the bigger picture. For one thing, Democrats swept Harris County in 2016 while John Culberson was winning in CD07 by 12 points. If we’re in an election year where CD07 is truly a tossup, then that strongly implies an even better year for Democrats in the county. Even more than that Lina Hidalgo poll, this should be encouraging for Dems, and downright terrifying for Republicans.
But it’s not just Harris County. There are two big reasons why CD07 is and has been seen as a top pickup opportunity. The main reason is because Hillary Clinton carried the district in 2016, but as we have discussed here before, some of that was because of crossover voters. Like I said above, Culberson still won the district 56-44. The other, equally important, reason is that the national atmosphere is one that favors Democrats and strongly indicates that the Republican advantage in districts like CD07 will be greatly diminished. Put another way, we expect that more Democrats and fewer Republicans will vote than in other similar election years. And that’s not just true in CD07, and in other battleground districts like CD23 and CD32. It’s true across the board, and it’s factored into every election prediction model, like the Morris model. Scroll down to the “Forecasts for every House seat” section and compare his projected margin in each Congressional district to the actual margins from 2016 and 2014.
This is something that I don’t think has been absorbed by media outlets and pundits in this state, all of which comes very much to the fore when a statewide poll like the second one from Quinnipiac comes out. Greg Abbott, who carried Harris County by five points in 2014, carried CD07 by a 60-38 margin in 2014; Culberson won that year by a 63-35 score. Again, if we are in an election where CD07 is a tossup, then the effect of that will be felt statewide, not just countywide. More to the point, if we are in that election, then the same effect will be felt in every Congressional district in Texas. It will be felt more in some districts than in others, and in specific races with specific candidates with strengths and weaknesses that may counter or enhance the national mood. But it will be felt.
The point I’m making is that a poll like that second Quinnipiac poll may be right, and polls like the DCCC CD07 poll and the Hidalgo Harris County poll may be right, but they can’t all be right. If the Q-poll is right, the other two are almost certainly too optimistic about Democratic chances, and if the latter two are right, then that Q-poll is almost certainly understating Democratic statewide support. I wish the people who write about these things would take that into consideration when they do. We don’t know yet which view is right. The fact that these conflicting polls exist is almost certainly because everyone has a different idea of what that national atmosphere will be like, and how big its effect on Texas will be. If you’re skeptical of any effect here you need to explain why. For now at least, all I’m saying is that polls like these don’t exist in a vacuum. Don’t evaluate one without taking into consideration the others.
Pingback: Checking in on the Congressional forecast – Off the Kuff