The results of the Democratic straw poll were ridiculous: Edwards 38%, Obama 21%, Hillary 20%. Reduced to rounded numbers, this makes the mathematical tally, for the 8,101 votes cast:

Edwards 3,078
Obamaa 1,701
Clinton 1,620

No chance that this is an accurate reflection of the Texas Democratic electorate. None.

Some media outlets reported the results as if they actually meant something. At least Robert Garrett of the Dallas Morning News mentioned the methodological shortcomings of the “poll.”:

Online surveys are not considered an accurate reflection of public opinion because participants are a self-selected group. Reputable pollsters carefully screen respondents to ensure they are potential voters, and they take a sample that matches the breakdown of a population by age, race, sex and location.

Mr. Richie [Boyd Richie, party chairman] said Texas Democrats tried to block attempts to vote more than once or from out of state. But the party could do little to ensure that respondents were even eligible to vote, let alone registered to vote or likely to do so in a Democratic primary.

In the real world, the latest state polls I could find for Texas, covering the period July 9-16, showed these results:

UPDATE: Please see the comments, below, for the results of the August 29-30 IVR poll. IVR is associated with the Burnt Orange Report blog. Thanks to BOR for the information.

IVR Poll 7/9

Clinton 41
Obama 20
Edwards 12

Lyceum Poll 7/14

Clinton 33
Obama 21
Edwards 10

IVR Poll 7/16

Clinton 42
Obama 24
Edwards 12

IVR is a relatively new player on the polling scene. Like SurveyUSA and Rasmussen, it uses automated calling technology. Its self-description of its methodology includes the name of the person who runs the company: Ralph Bordie. The name meant nothing to me until I read further: He is the “son of a Texas Senate parliamentarian.” That would be, of course, Camilla Bordie, parliamentarian to Lieutenant Governor Bill Hobby and my onetime colleague at the Texas Legislative Council.

IVR Polls uses automated IVR technology to survey voter populations with greater speed and lower cost than traditional phonebank polling. IVR stands for Interactive Voice Response, a technique for interaction with a computer through a telephone keypad, usually in response to a recorded or synthesized voice.

There has been much discussion on the reliability of automated polls. Skeptics say that the computer can’t tell whether a child is randomly punching buttons. This is true, but it is rarely the case. Adults have been known to lie to live pollsters as well, but in a full size sample, the rare outlier has little effect on the overall numbers. Independent reviews comparing automated polls to phonebank polls have shown that a well designed poll can produce accurate results using either method. Slate has a widely cited article comparing various pollster’s numbers on the 2004 race. In that case, the automated polls beat the phonebank polls.

Phonebank polls can vary in quality from call to call. After a few hours on the phone, fatigue can affect the interviewer’s delivery. With IVR polls, a question can be recorded several times until the inflection is perfect. Every call is made using this perfect recording, resulting in a uniform survey free of interviewer bias.

VoIP technology allows calls to be made at an extremely high rate. Voice is digitized and sent in small packets allowing many calls to be processed simultaneously. Equipping and staffing a call center to match our call volume would require significant overhead and raise costs. While residential VoIP calls can vary in quality due to the many ‘hops’ they must make on the way to their destination, IVR Polls uses dedicated servers located in datacenters with direct access to the main internet trunks. These trunks are then connected to phone company switches, allowing our calls to completely bypass the end user congestion that causes most VoIP quality issues.

IVR Polls is run by Ralph Bordie. While relatively new to the polling industry, Ralph has a background in performing statistical and data analysis for litigation, where multi-million dollar settlements depended on his methodology withstanding the scrutiny of opposing experts. He has worked on a number of campaigns going back to 1980, and as son of a Texas Senate Parliamentarian, he was soaking up political ‘inside baseball’ for years before that.

To get back to the Texas Democratic straw poll: My first thought was that the result was manipulated. The trial lawyers own the party. More likely, the kind of folks who are likely to participate in an online poll are under-35 populists who love Edwards’ anti-corporate rhetoric and regard Hillary and Obama as too centrist.

I haven’t mentioned the Republican straw poll. Not worth mentioning. Only two candidates showed up and neither has a ghost of a chance for the nomination.