A poll taken by Indiana State University and the Evansville Courier Press shows Democrat Brad Ellsworth besting incumbent John Hostettler by 15 points. The poll was of 603 registered voters; the MoE is +/- 4%. The NRCC has legitimate qualms about the scientific credibility of the poll, though, which are spelled out below.
"The conventional wisdom is that a high percentage of undecided is bad news for the incumbent," he said.
University of Evansville political scientist Robert L. Dion said he senses that Hostettler is behind at this point, but not 15 points behind.
The undecided voters probably are "disappointed conservatives" hesitating to commit to Hostettler again because they are unhappy about the conduct of the Iraq war or the economy, Dion said. ... Steiger's telephone interviewers asked self-described undecided voters if they could check back with them within two weeks after an Oct. 4 debate in Vincennes, he said, but 42 percent said no.
"I consider that a refusal to say who they're for," he said. "We'll call back the remaining 58 percent. That will be really interesting, to see how the undecideds are breaking."
Ed Patru, a spokesman for the National Republican Congressional Committee, questioned the poll's legitimacy.
"The methodology of the poll is flawed," Patru said in a written statement.
"The poll tests registered voters instead of likely voters, and it was conducted over several weeks, so it's pretty difficult to claim that the survey provides an accurate snapshot of voter sentiment. ... Dion said a weakness of ISU's poll is that it was conducted over a period of 15 days - a period, he said, that coincides with an uptick of support for Republicans nationally.
"Fifteen days, that's a long time," he said. "People who study public opinion say a good poll is a snapshot of a given moment in time. But over the last couple of weeks gas prices have been plummeting, and there's been the fifth anniversary of (the Sept. 11, 2001, terror attacks)."
