MPR news, December 17, 2010
Suggested weighting future polls to account for geographic variations. Jacobs also recommended other ways to improve the poll.
A few weeks ago, Minnesota Public Radio News and the University of Minnesota's Humphrey Institute announced a review of polling methodology and said an outside firm would audit the analysis.
The action was taken after an October poll showed Democrat Mark Dayton with a 12 percentage point lead over Republican Tom Emmer in the race for governor. Dayton was eventually elected, but the margin was much closer - less than a half a percentage point.
The review raised several questions about the poll. One key issue is that the people in the 612 area code responded to pollsters at a higher rate than people in the rest of the state. The 612 area code consists mostly of Minneapolis, a heavily DFL area. It appears the pollsters didn't adjust the results to take that into effect.
Larry Jacobs, who conducted the poll for the U of M's Humphrey Institute and MPR News, suggested weighting future polls to account for geographic variations. Jacobs also recommended other ways to improve the poll. They include changing the order of questions to try to determine which way undecided voters are leaning and conducting the survey closer to Election Day.
Frank Newport, Editor in Chief for Gallup, reviewed the recommendations and said it's common practice to weight a poll by region. He also suggested the U of M review how it screens for likely voters.
Chris Worthington, managing director of News at Minnesota Public Radio News, called the review a necessary step and said he looks forward to more conversation about the recommendations.