OK, I want a Mason-Dixon poll now for TX Governor to sort all this out. 
I've never seen a Mason-Dixon poll of Texas, ever. The only polls I've seen do Texas are Rasmussen, SurveyUSA and Scripps Howard (not a quality poll, btw).
OK, who's the best of the groups that do poll Texas, then?
I had to look some of this up and I can't some of the other data (like 2002 Governor)
In 2004, SUSA pretty much nailed Texas, with a prediction of Bush +22. Scripps Howard was at Bush +26, still within MOE, but not as good. Rasmussen's last poll in Texas was August 26, and showed Bush +19. That's quite to long ago for me to count.
In 2002, SUSA said Cornyn +7, it was Cornyn +12. Zogby did even worse, saying Cornyn +1 (one of typically bad ones that year). Two media outlets did their own (I think the Dallas Morning News through Scripps-Howard), with Cornyn +9, and the Houston Chronicle did one with Cornyn +6.
In 2002 Governor, I don't remember the exact totals, but Zogby actually did pretty well here (go figure). So did SUSA.
So, in conculsion, the Zog should poll Texas, but buyer beware. Rasmussen has a non-existant record. Scripps-Howard is better than I remember it being, but SUSA is probably the best.
That still doesn't mean I'll ignore its often wild and strange swings.
