Both polls had the same sample size – about 400 respondents – and claim the same margin of error – about 5 percent. But a difference that might explain the wildly diverging results found in the two polls lies in the sample from which respondents were drawn.
The Hayhurst campaign's poll questioned registered voters Wednesday and Thursday. The SurveyUSA poll, conducted for the Mike Downs Center, questioned respondents Oct. 21 through Sunday. It limited its sample to people who had already voted or who were likely voters.
The News-Sentinel and WANE-TV were partners in that poll, which found support of 57 percent for Stutzman, owner of Stutzman Farms Trucking and state senator for District 13, which includes Steuben, LaGrange and Noble counties, and 32 percent for Hayhurst, a pulmonologist. The two are trying to fill the vacancy left by Mark Souder, who resigned May 18 because of an affair with a staff member.
The Hayhurst campaign's poll was conducted and its results analyzed by Riggs Research Services of Indianapolis. James Riggs of that firm declined to discuss the methodology of the SurveyUSA poll.
Michael Wolf, an associate professor of political science at IPFW, is one of the analysts who scrutinized data from the SurveyUSA poll.
“The real difference is the sample frame – registered vs. likely voters,” Wolf said.
He said it's possible that the Hayhurst campaign has turned the tables on a Stutzman lead if it has mobilized large numbers of voters to become likely, instead of merely registered, voters.
Until Tuesday evening, Wolf pleads a truism of survey research.
“We always say, ‘All polls are a snapshot in time,'” Wolf said.