During the fall of 2000 and 2002 I had several conversations with Rob Daves, the man who ran the Minnesota Poll at the Minneapolis Star Tribune. The conversations raised questions of methodology and accuracy of the poll; Daves defended both. I was also a party to email exchanges between Daves and Ben Whitney, the head of Norm Coleman’s successful 2002 Senate campaign.
The Minnesota Poll consistently showed Coleman trailing by substantial margins in the fall of 2002. Based on the Coleman campaign’s internal polling and other factors, Whitney thought Daves had it wrong. We published Whitney’s critiques of the Minnesota poll on Power Line and I laid out the case against the Star Tribune’s Minnesota Poll immediately following the November 2002 election in “The trouble with the Star Trib poll.”
Rob Daves left the Star Tribune in one of the cutbacks at the paper over the past few years. Now the paper contracts its polling to Princeton Survey Resarch Associates International. Thus Princeton Survey Research is the source of the poll data showing Norm Coleman 10 points behind Al Franken in today’s Star Tribune story by Kevin Duchschere.
Kevin Duchschere is a terrific reporter. I have found Duchschere to be the soul of fairness in my dealings with him over the years, but his story today is mystifying. It provides no data that would allow readers to take a look inside the poll’s data, even though the story acknowledges that the poll conflicts (by 20 points!) with the Survey USA poll on the Coleman-Franken race released yesterday. The Star Tribune’s sidebar on how the poll was conducted is likewise lacking in such data. In place of internal data that might allow readers to evaluate the poll themselves, Duchschere gives us the assurance of Star Tribune editor Nancy Barnes:
â€œThis is an independent pollster who is respected across the country,â€ Star Tribune editor Nancy Barnes said. â€œItâ€™s the same pollster who found the presidential race in Minnesota to be a dead heat last month. All polls have a margin of error, but on the whole we trust that these results have merit.â€™â€™
Doing a little big of digging, we have obtained the internals of the September 15 Princeton Survey Research poll (in which Coleman led by four points) and today’s Princeton Survey Research poll. The September 15 poll had a party identification breakdown between Democrats and Republicans of 34-31. In today’s poll the party identification breakdown between Democrats and Republicans is 42-26. (Michael Brodkorb has posted a Coleman campaign memo with the Princeton Survey Research internals here.) The shift in the breakdown of the sample tracks the difference in the results reported and raises the question whether Minnesotans have shifted party identification by 13 points in the past three weeks or whether the current poll sample is invalid. At the least, I think readers should be provided the information that raises the question, particularly in light of the history of the Star Tribune’s Minnesota Poll.
To comment on this post, go here.