A day after CNN released polling showing Hillary Clinton up just two points on Barack Obama in Texas, a state her campaign has put forward as a must-win, a new poll from SurveyUSA shows largely the same results. The results below, along with the latest Pollster.com trend estimate in the state in parens.

Hillary Clinton: 50 percent (49.6 percent)
Barack Obama: 45 percent (44.9 percent)

Here's some interesting analysis from the folks at SurveyUSA. Note, in particular, the couple of sentences after the "BUT":

In a Democratic Primary in Texas today, 02/18/08, 15 days to the vote, Hillary Clinton leads Barack Obama 50% to 45%, according to a SurveyUSA poll conducted exclusively for KTRK-TV Houston, KTVT-TV Dallas, and KRLD-AM radio Dallas. BUT: there is "give" in these numbers that must be mentioned in the same breath. Among Hispanic voters, Clinton leads 2:1. SurveyUSA estimates that Hispanics make-up 32% of Democratic primary voters in a Primary today. If Hispanics vote in larger numbers, Clinton's lead is larger than the 5 points shown here. If Hispanics vote in smaller numbers, Obama runs stronger than these numbers show. Among white voters, Clinton leads by 12. Among black voters, Obama leads by 57. Race Gap is 69 points. Among males, Obama leads by 20. Among females, Clinton leads by 27. Gender Gap is 47 points. Among registered Democrats, Clinton up 14. Among Independents, Obama up 28. Among voters under age 50, Obama by 6. Among Voters 50+, Clinton by 17. Age Gap is 23 points. Clinton ahead in South Texas and West Texas. Obama and Clinton within the margin of sampling error in North Texas, East Texas, and Central Texas.

It's a positive development to see a pollster concede that it doesn't know exactly how things are going to pan out, offering not one but multiple potential turnout models. Indeed, we are in a period in which polling hasn't been the most reliable -- largely because, I believe, turnout has increased so greatly over the past, with both new voters and voters new to Democratic primaries casting ballots for Obama and Clinton. This isn't to say that there isn't value in these numbers. In fact, SurveyUSA has been fairly good at predicting the margin of victory in contests thus far. Nevertheless, with so much riding on just who shows up to the polls (not to mention the way delegates are apportioned -- note that Texas has a funky system that could lead to a split between the popular vote leader and the delegate leader), I'mn not sure that I'd bet the farm on any numbers right now.