Evan Smith e-mailed me last night that I would find the poll to be “interesting.” He was right. The previous UT/Tribune poll, in May, had Perry ahead by 49-40. Perry has leaked 10 points since then, and White has lost 7. Perry is ahead–no surprise there–but he is languishing at a familiar number: 39%. The large number of undecided voters is a problem for an incumbent. As a rule, undecideds tend to break in favor of a challenger at the end of a race. Libertarian nominee Kathie Glass has a significant 5% of the vote. This is the lowest number Perry has polled in the entire campaign. The biggest danger for Perry is anti-incumbent sentiment. The fact that he is losing strength suggests that he may be getting caught in the whirls and eddies of an anti-incumbent wave. White has come out for term limits for governor, and he really needs to hammer that home. He should say: (1) The first thing I will do as governor is ask the Legislature to submit to the people of Texas a constitutional amendment establishing a two-term limit for Texas governors. (2) Whether the amendment is approved or rejected, I pledge that I will limit myself to two terms as governor. The poll surveyed 800 registered voters and has a margin of error of +/- 3.46%. I should remind readers that the UT/Trib poll has an idiosyncratic methodology. It is not random. People sign up online to be part of the pool of people that is surveyed. The result is that participants show an atypical degree of interest in politics: Extremely interested 53% Somewhat interested 36% Not very interested 9% Not at all interested 2% Granted that this is a year in which interest in politics is very high. Nevertheless, I find it hard to believe that a true random sample of 800 registered voters would find 89% to be “extremely” or “somewhat” interested in politics. The UT folks defend their methodology vigorously, but I continue to have doubts about whether a self-selecting pool of voters is truly representative of the electorate.s More from the poll: Perry job approval rating Strongly approve 12% Somewhat approve 28% Neither approve nor disapprove 16% Somewhat disapprove 17% Strongly disapprove 25% Don’t know 3% So Perry’s approval rate is 40% and his disapproval rate is 42%. The disapproval rate is a little lower that other polls have shown. The significant comparison lies in the “strongly approve” and “strongly disapprove” categories. Strong disapproval is twice that of strong approval. Downballot races Light Gov: Dewhurst +15 AG: Abbott +17 Land commissioner: Patterson +10 Ag commissioner: Staples +7 Railroad commissioner: Porter +8 Generic congressional/legislative ballot Congress: R 48, D 33 Legislature: R 47, D 32 These are terrible numbers for Democrats. More from the poll: “White has not yet faded and remains in striking distance of Perry,” says Daron Shaw, a government professor at the University of Texas who oversees the UT/Tribune poll with his colleague Jim Henson. “The downside for White is that Perry is up by 18 points among those who say they are extremely likely to vote. White needs a big turnout among young voters and minorities to be competitive.” This is a peculiar way to make the point. It’s that “yet” in the first sentence that seems weird to me. Why not just say, “White has not faded?”