You can ask all the right questions in a political poll and still get a wrong answer. The results are based not only on how people answer the questions but on a pollster's educated guess about who'll vote and who won't.
That's one reason why, in 10 polls of the Texas governor's race taken since Sept. 1, Gov. Rick Perry's lead over his Democratic challenger, Bill White, has ranged from a low of 1 percent to a high of 14 percent. In four polls over just the last two weeks, Perry led 7 percent (Blum & Weprin, for the state's biggest newspapers), 14 percent (Public Strategies Inc., for Belo television stations), 5 percent (Shaw/Jasperson for the Texas Lyceum) and 11 percent (Rasmussen Reports). The latest University of Texas/Texas Tribune poll had the two candidates 6 points apart.
Some of those differences are the result of which respondents the pollsters decided to count. People who've voted before? People who say they'll vote? People who are registered to vote and say the election is important to them?
In 2008, when the original round of Obama-mania swept through the electorate, the number of new voters jumped dramatically on both sides of the partisan ledger. One question now is whether those same new voters will show up in November to vote for governor and other offices.
Many pollsters assume they won't. Even in 2008, when the motivation to turn out was at its height, a lot of those folks voted only for president and skipped other races on the ballot. In that year's Democratic primary, 2.9 million people chose between Hillary Clinton and Barack Obama. The next contest down on the ballot was for U.S. Senate — and 697,734 of the people who voted for president didn't make it that far.
"Those aren't likely voters," says Bryan Eppstein, a Republican pollster, political consultant and lobbyist based in Fort Worth. He thinks the most reliable indication that someone will vote in this election is a history of voting in earlier elections for governor, in 2006 and 2002.
Jeff Smith, an Austin-based pollster who works for Democrats, shares that view. "The stuff we do is based on the voter file," he says, referring to the official records of who voted and in what elections (but not how they voted, which remains legally secret). He throws out the new 2008 voters unless they've taken part in elections since then. "I don't think most of them will be back," Smith says, adding that the turnout this November will probably "revert to what we've seen in past gubernatorial elections."
In 2008 — a presidential election year — 45.6 percent of the state's voting age population, or 8.1 million voters, showed up for the general election. In 2006 — a gubernatorial year that featured a governor's race with four major candidates — 26.4 percent of the adults, or 4.4 million, showed up at the polls. The numbers were a bit higher in the 2002 gubernatorial race that featured Perry and Tony Sanchez: 29.4 percent of the state's adults, or 4.6 million, turned out for that election. If those numbers hold this time, somewhere around 5 million Texans will vote for the state's next governor.
Another thing the pollsters are talking about this year is voter intensity: who's more likely to cast a ballot. That gets right at whom they think they need to poll and how much weight they should assign to different sets of voters. "There are two keys to polling," Eppstein says. "Accurately capturing the opinions of the people you're interviewing and making sure the people you're interviewing are representative of the people who are voting. We look at people who have a history of voting in this type of election." he says. Eppstein thinks differences in various polls this year are easy to explain. "They're clearly interviewing different groups of people," he says.
Everybody agrees, now that it's over, that 2008 was a Democratic year, at least nationally. Texas Democrats did better than usual but still didn't win a single statewide office, a majority of the congressional delegation or either statehouse chamber. Still, Democrats were more motivated than Republicans, and it showed up in both the turnout and the results.
This year, the prevailing wisdom is that conservatives are more motivated than liberals. "It's easier to get people to vote against something. They have that going for them this time," Smith says of the Republicans.
Still, Smith says, he sticks with voter history over other factors, like intensity, when polling. People who'll take 10 minutes to give their political opinions on the phone tend to be people who are interested in politics and who'll participate in elections, he says. "That's why I do it on the basis of voter history and not on, 'Are you excited right now?'"
Eppstein says anger at the federal government might motivate one side, but the Democrats who came for the first time in 2008 weren't all that likely to come back this year anyway. "This is not a race between Republicans and Democrats," he says. "It's a race between conservatives and Obama, and a lot of Democrats who voted for Obama aren't motivated to vote this year."