The results were close to the same in a CNN/TIME/Opinion Research Group poll released last Wednesday that showed Strickland with 48 percent support to 47 percent support for Kasich.
However, a Quinnipiac University poll released on Tuesday showed Kasich ahead, 51-41 percent. And an earlier Quinnipiac poll had Kasich ahead by as much as 17 points.
What gives?
The first thing to remember when discussing polls is that they are based on probability, not certainty, said political scientist John Green, director of the Bliss Institute of Applied Politics at the University of Akron.
Polls use a scientific, randomly drawn sample aimed at representing the outcome of what would happen if everyone in a group — such as all likely voters in a state — were questioned.
Even if the same organization did the same poll on the same day, the results wouldn’t always be the same, said Green.
The range of possible results often is referred to as a poll’s margin of error. Pollsters often say the results would fall within the poll’s margin of error 95 percent of the time.
In the new Dayton Daily News/Ohio Newspaper poll, the margin of error is plus or minus 3.3 percentage points in 95 out of 100 cases.
The margin of error is just one possible source of inaccuracy in polls.
Others include lack of proper training by interviewers, flawed wording of questions and the order in which questions are asked, writes Sheldon R. Gawiser and G. Evans Witt in their book, “20 Questions a Journalists Should Ask About Poll Results.” Also, while most polls on the governor’s race refer to “likely voters,” there’s no “generally accepted way to define likely voters,” Green said.
“Different people do it in different ways,” he said. “Pollsters almost get into fist fights over how to measure likely voters.”
For the Dayton Daily News/Ohio Newspaper Poll, likely voters were determined by people who said they were most interested in the election and who said they were going to vote, said Eric Rademacher, co-director of the University of Cincinnati’s Institute for Public Research which did the poll.
When polls are taken also may have an effect on the results, said Rademacher.
The new poll on the governor’s race was taken Oct. 14-18, near the time of President Barack Obama’s rally at Ohio State University that drew 35,000 people and was aimed at increasing Democratic enthusiasm.
Experts also caution that polls shouldn’t be used to predict an election’s outcome. Rather, they provide a snapshot of what voters were thinking when a poll was taken, said Green.
Polls also are more valuable in measuring the movement in opinion from one survey to the next than they are in providing the specific percentages of support for candidates, Green added.
About the Author