| Yesterday we wrote about a Texas Poll Watch poll, perhaps too soon. After a day of looking at numbers and talking to pollsters, we should not have printed the poll because none of the numbers seem correct.
First let's go through the poll county by county.
Texas Poll Watch - Aransas
Hunter- 48%
Garcia- 40.80%
Texas Poll Watch-Calhoun
Hunter- 33.75%
Garcia - 48.75%
Texas Poll Watch-Nueces
Hunter- 52.06%
Garcia- 39.69%
Texas Poll Watch-San Patricio
Hunter- 45.20%
Garcia- 32%
Now let's look at 2006.
Aransas
Seaman- 47.55%
Garcia- 44.94%
Calhoun
Seaman- 46.13%
Garcia- 49.59%
Nueces
Seaman- 50.97%
Garcia- 42.84%
San Patricio
Seaman- 40.90%
Garcia- 54.61%
To take the poll on face would mean that Garcia is - 4.94% in Aransas, - 1.16% in Calhoun,
- 3.15% in Nueces and - 22.61% in San Patricio.
As Kuff pointed out in the comments, "Garcia won San Patricio County by a huge margin in 2006 (7142 to 5350, or 57.2% to 42.8%), so any poll that shows him trailing there by double digits is a little suspicious to begin with. San Patricio did go mostly R in 2006 overall, though Bill Moody won it with 52%, but the downballot statewide Rs didn't do so well, with Todd Staples and Elizabeth Ames Jones getting only a plurality. My guess is that this is a small subsample with a large MOE, and as such probably not very useful."
To add to this, because of the huge swings in San Patricio and Aransas, it looks like Texas Poll Watch is under sampling Hispanics and over sampling Nueces. The combination of the two would skew any poll dramatically and is an amateur mistake when weighting and creating the methodology.
Ed Martin echoes both points in the comments.
In 2006, an excellent Democratic/Hispanic field effort at the end of the campaign helped drive those San Patricio numbers to 57% for Garcia. In this district, San Patricio's 12.5K votes in a House race in an "off" year make it a major county. Unless a pollster knows how to test an accurate and representative sample of Hispanic voters, their preference will typically be under-represented in most polls in South Texas counties like San Patricio. Bottom line: If you flip the San Patricio numbers, this is still a close race that could go down to the wire, but not one that has Hunter ahead, especially by 8 points.
The most important part is this poll both fails to take in unlikely voters that Garcia's campaign will obviously target and new voters identified by the 2008 presidential primary.
In Aransas that is 873 voters. In Calhoun that is 2,653 voters. In Nueces that is 4,271 voters. In San Patricio that is 6,639 voters.
In 2006, Garcia won with 17,607 votes. The Democratic primary alone has identified 14,436 votes compared to 2006 when Garcia had 5,401 2006 primary voters identified. This poll would not touch a large segment of this 14,436 because it is an automated phone system with no live dials. That same system would disproportionately hit men, voters over 35 years old, and excludes minorities.
Just looking at the county by county numbers this poll is strikingly off, and a correction in San Patricio would completely eliminate the 8 point lead Hunter has in this apparently faulty poll. |
