An El Paso Times/News Channel 9 Poll shows Dee Margo with a 43% to 31% lead over Pat Haggerty in the Republican primary battle in El Paso that features a Tom Craddick supporter (Margo) against one of Craddick’s most vociferous critics. The poll surveyed 1,164 likely Republican primary voters–a very large sample for a legislative race with a low margin of error of 3 points. The raw numbers of the poll conducted by the Ruell Group are:

Margo 500
Haggery 364
Undecided 300

The Haggerty campaign immediately released the results of an internal poll that shows Haggerty ahead 46-37 among Republican primary voters and a tracking poll as well that has Haggerty up 48% to 42%. The press release, dated Monday, does not provide any information about how many voters were polled.

The Haggerty campaign offered two possible explanations for the discrepancy in the two polls. One is the method of polling. There is considerable controversy in the polling industry about the effectiveness of automated polls like SurveyUSA and Rasmussen, which use professional announcers to ask prerecorded questions and ask respondents to push buttons on the telephone keypad to indicate their responses. The automated pollsters argue that the professional announcers remove bias in the asking of questions, while live-voice polling may not. Automated polls are much less expensive than live-voice polling and were looked upon with some disdain prior to the 2004 presidential election, when SurveyUSA and Rasmussen proved to among the most accurate polling operations in the country. I am more or less neutral in this debate, although before SurveyUSA lost its media clients in Texas, I found several large month-to-month swings in the favorable/unfavorable rankings of Governor Perry and the state’s U.S. senators for which I could fathom no apparent reason.

The other reason for the discrepancy in the polls given by the Haggerty campaign was that Margo had spent $200,000 on TV and mailings before Haggerty went up on television, which, according to Haggerty, did not occur until last Thursday, “after everyone’s polls had been done.” The Times story confirmed Haggerty’s statement:

Russell Autry, chief of operations for The Reuel Group, said those commercials could be one factor that influenced the poll results. Haggerty’s first TV ad did not air until after the poll.

Update: These are the details of the Haggerty internal polls:

Both Haggerty polls were of republican primary HISTORY voters (unlike the EPT poll which appears to have used an automated “screen” question to qualify voters).

December poll was 350 interviews; +5% margin of error.

February poll was 1,000 interviews; +3% margin of error.

All interviews were live, not “automated robo interactive” calling as the EPT survey.

All three polls were conducted prior to Haggerty running any broadcast advertising versus Margo running an estimated $150,000 of tv and radio through last week.