With a week to go before the runoff election for congressional district 23 between Republican incumbent Henry Bonilla and Democratic challenger Ciro Rodriguez, a SurveyUSA poll shows Bonilla leading by 53% to 46%. The sample of 471 likely voters has a margin of error of +/- 4.6%.

SurveyUSA uses professional announcers to ask recorded questions, a methodology the organization says results in fewer refusals and less influence on respondents. Automated polls are controversial, but they outperformed organizations using live questions in the 2004 election cycle.

The only eyebrow-raising number here is the low fraction of undecided voters (1%). This seems unusual for a special election at a season when politics is not foremost in voters’ minds.

Here’s what SurveyUSA has to say about the race: “Bonilla gets 70% of White votes. Rodriguez gets 72% of Hispanic votes. In SurveyUSA’s turnout model, 59% of likely runoff voters are white, 36% are Hispanic. If Hispanics, who are 55% of the population in TX 23, make up more than 36% of those who vote in the Runoff, the contest will be closer than SurveyUSA’s numbers here show. Bonilla gets 94% of Republican votes. Rodriguez gets 89% of Democratic votes. Independents split. Bonilla wins by 25 points among higher-income voters, and by 15 points among middle-income voters. Rodriguez wins by 25 points among lower-income voters.”

Bonilla received only 8% of the Hispanic vote in 2002, when his district included Webb County. From 8% to 28% is a big jump. Webb County has a machine-generated Democratic vote; San Antonio is different. The judges who drew Bonilla’s district did him a big favor by getting him out of Laredo and into the south side of San Antonio, where he grew up.

This district is winnable for a Democrat, but Rodriguez is in mathematical never-never land–close enough to have a chance, too far away to win. Getting that last 4% of the vote is incredibly tough, especially since he already has 89% of the Democratic vote and is losing the independents. Where are his votes going to come from? The ray of hope is that SurveyUSA’s turnout model (59% Anglo, 36% Hispanic) is faulty. But, as I have said before–only to be ridiculed by the Burnt Orange Report bloggers–Republicans tend to do a better job of getting their vote out for special elections than Democrats do. I’m surprised that the race is as close as it is. I thought Bonilla would be in the 55-57 range.

The likely Bonilla victory will not have any long-term significance. Bonilla’s party is now in the minority, and that’s no fun at all. He has made it clear that he is a short-termer in the House and wants to run for the Senate if Hutchison gives up her seat to run for governor in 2010 (or is the vice-presidential nominee on a winning GOP ticket in ’08). He may not seek reelection in two years, and he may get beat by a stronger candidate than Rodriguez if he does. By 2011, the Voting Rights Act will make this a Democratic district.