In 2000, George Bush lost Minnesota narrowly to Al Gore by 60,000 votes out of the nearly 2.5 million cast. The Minneapolis Star Tribune’s Minnesota Poll, however, reported that Bush was trailing Gore in the state by a huge margin on the Sunday before the election. In retrospect, the Star Tribune’s obviously erroneous Minnesota Poll might itself have helped Gore secure his margin of victory over Bush in Minnesota in 2000.
Since the 2000 campaign Minnesota has frequently been cited as a battleground state both by the Bush campaign team and by impartial electoral analysts. Is it? Today’s Minneapolis Star Tribune reports the results of its current Minnesota Poll on the Bush-Kerry race in “It’s Kerry over Bush by 12 points,” 50-38 percent. The results reported in this particular poll are not inconsistent with our sense of the status of the race in Minnesota today, but putting to one side the timing of the poll — the week Richard Clarke testified before the 9/11 Commission– the poll sample does in fact raise questions about the reliability of the results.
The Star Tribune itself notes the possible impact that the timing of the poll had on its results: “The poll found that support moved dramatically during the four days of interviewing, perhaps reflecting a toll on Bush from Clarke’s media interviews and testimony before the Sept. 11 commission. On Sunday and Monday, Kerry’s lead was four points. But it ballooned on Tuesday and Wednesday, boosting him to the 12-point lead. That shows that there’s a potential for large shifts in support as the campaigns ramp up their efforts closer to the election.”
But the sample from which the poll results are derived seems obviously out of whack: “[Democrats] and independents who lean to the DFL represent 53 percent of the likely voters in the state, while Republicans and independents who lean toward the GOP represent 39 percent.” That is a sample that is hard to square with our knowledge of the Minnesota electorate and the outcome of the races for state office in 2002. In 2002, the relatively unknown Republican candidate for governor (Tim Pawlenty) garnered 45 percent of the vote against the relatively well-known mainstream Democratic candidate (Roger Moe), who finished second in a three-man race with 36 percent of the vote.
Former Democratic Congressman Tim Penny ran as the candidate of Jesse Ventura’s Independence Party and won 16 percent of the vote, thus complicating a reading of the outcome. (The senate race had so many twists and turns that I think it is even less useful as a barometer of the electorate’s current party affiliation.) Republican candidates won two of the three remaining statewide office in the 2002 election, with incumbent Attorney General Mike Hatch surviving as the lone statewide Democratic officeholder.
We wrote a lot on this site about the Minnesota Poll during the 2002 race and summarized our qualms with the poll’s reliability in the immediate aftermath of the 2002 election in “The trouble with the Star Trib poll.” Until today, I thought that the poll had corrected its obvious sampling problems, but today’s story raises a serious question in my mind about whether that is in fact the case.
If you are unfamiliar with the methodology of the Minnesota Poll, you won’t want to miss the sidebar “How this poll was conducted” that always accompanies stories reporting the poll results. Here is the description of how the pollster adjusts the raw survey results, the traditional highlight of the sidebar:
The sample first was weighted to take into [account] unequal probabilities of selection from sampling: Weighting accounts for the number of telephones going into the household, and household size. It then was weighted for age, gender, and education to take care of minor fluctuations in the sample, and align it with the findings of the 2000 Census of the adult population. It is assumed to be representative of all Minnesota households with telephones, within the margin of sampling error. It is not weighted on characteristics for which there are no good population parameters, such as party affiliation or religion.
All that having been said, however, we believe that President Bush in fact trails Kerry in Minnesota today by an amount consisitent with the results in the poll results reported in the Star Tribune story.
DEACON asks: Trunk, which results do you mean, the 12 point gap or the 4 point gap of a few days earlier? Ordinarily I’d be the last person to question your take on the status of any race in Minnesota, but if you mean Bush is roughly 12 points behind in Minnesota, I’m dubious. First, I don’t see how a poll that uses a distorted sample pool is likely accurately to reflect the magntude of Kerry’s lead. Second, I’ve been following poll results from other states as reported by Real Clear Politics. They show Bush up by 6 percent in neighboring Wisconsin and trailing only narrowly in Michigan. I’ve also seen polls that say the race is close in Pennsylvania. And, nationally, the race seems to be a toss up, as it was in 2000, when (as you point out) Bush lost narrowly in Minnesota. So if you are saying that Bush is roughly 4 points behind, that’s plausible. But I’ve got to believe that he’s not, and has never been, behind by double figures.
BIG TRUNK responds: Based on information from confidential sources to which Rocket Man has previousluy alluded, we have believed that President Bush was recently behind in Minnesota by a large margin. Today’s Star Tribune Minnesota Poll, with a 12-point spread inflated by the poll’s traditionally Democrat-biased sample, and with the obvious issue of timing, is the first information other than the March Rasmussen poll that I have seen suggesting that President Bush is within shouting distance of Kerry in Minnesota.