The sample was 1,000 adults, including 32% Hispanics, out of which there were 478 likely voters (registered; voted in every, or almost every election; and were at least “somewhat interested” in the upcoming general election). The likely voters were evenly split between Republicans and Democrats. Ralph Nader and Bob Barr, the Libertarian candidate, were also included in the presidential poll. All candidates were described by their affiliation. The margin of error was 4.5%. The poll was conducted by Darren Shaw, associate professor of government at the University of Texas. Neither Obama nor Noriega had significant crossover appeal. Only 6% of Obama’s support came from Republicans, and only 3% of Noriega’s. The hard numbers in the Senate race were Cornyn 39%, Noriega 37%, which leaves a large undecided of 24%. Shaw told me that roughly half of the independents were undecided in the Senate race. The presidential poll seems credible to me. In fact, McCain by six points is exactly the number that a prominent McCain booster told me recently he thought that the actual outcome would be. Noriega by two is more problematic, but overall, these numbers, with the people who have made up their mind in both parties representing about three-fourths of the respondents and the rest being undecided, seem a lot more credible to me than the earlier polls that showed over 90% of the electorate committed (48%-44%). I had an interesting conversation with Professor Shaw after he went over the numbers. I asked him how confident he felt about the accuracy of polls these days. His answer was that polling is “increasingly problematic.” The problems are cell phones, caller ID, and unwillingness to participate. The typical response rate for a poll is about 15%. To drive the response rate up, you have to use multiple call-backs, which increases the cost of a poll. Because young people use cell phones rather than land lines, all polls tend to undersample younger voters. Young minority voters are the hardest group to poll, Shaw said. He believes that telephone polls will eventually be replaced by online polls. Already there are companies that are building large data bases of likely voters — a couple of million. One will supply people who don’t have a computer with a laptop in return for their participation. The obvious flaw here is that the sample is not random. I have been very critical of Zogby’s online polls for exactly that reason. Zogby sends e-mails asking if you want to sign up to be polled. I signed up, and I participate regularly. But I don’t think that this methodology is trustworthy.