Today, the Columbus Dispatch released the results of their mail-in poll in the Governor’s and Senate races. The Dispatch poll is the only poll outfit that conducts its polling entirely by mail. It’s lead for some rather unpredictable results, and as such, has been generally regarded as the least accurate measure of public opinion in Ohio.
The Dispatch claims that their poll has a margin of error of 2%. I don’t see it.
Of the Dispatch respondents, only 54% said they voted for Strickland in 2006, even though Strickland got 61% of the general election vote—a seven-point undersampling. The Dispatch’s respondents consisted only 45% of people who voted for Obama in 2008, even though Obama carried 51.5%, another nearly seven-point undersampling. In fact, a majority of respondents said they voted for McCain. Despite the fact that the Democrats have an over million voter registration advantage over Republicans, the majority of the Dispatch’s sample was registered Republicans.
Despite that and a disproportionate high sample size of central Ohioans, the Dispatch poll still shows the Attorney General and Secretary of State races essentially tied (however, that could be regional bias that would favor Franklin County Maryellen O’Shaughnessy. And the Treasurer’s race would appear to be within striking distance. But again, that could be the result of the regional bias of the poll since Boyce is from central Ohio.
In other words, this poll is, like the prior Dispatch polls, is completely useless. It predicted that several of the RON amendments in 2005 would pass—they all went down in defeat.
There was a reason the Dispatch had used the Ohio Poll by the University of Cincinnati earlier this year. The question is why would they go back to their widely discredited joke of a mail-in respondent poll.