The two most recent polls on the MN guv race — one from Rasmussen Reports and one from the Strib — portray the race very differently, and both throw a couple of curves, including one Rasmussen curve that the Tom Horner campaign finds particularly alarming (and, they would argue, inaccurate and unfair).
On a headline basis, Rasmussen shows Dayton and Emmer virtually tied (statistically insignificant lead for Emmer) and appears to show Horner taking a sharp drop, compared with other recent polls. The Strib poll shows Dayton with a big, statistically significant lead over Emmer. The Strib also confirms other recent polls that showed Horner rising over the past month and now at 18 percent.
I frequently warn of overreliance on polls (“crack for political junkies”). But:
a) We political junkies can’t stop inhaling them and
b) The particular dynamics of this race increase the always-present possibility that the polls will become a cause of the very effect they purport to measure.
IP nominee Horner’s chance of become a serious contender depend on him demonstrating that he has a real chance to win. The only way he can do it is through his showing in neutral polls.
Let’s be blunt. For several reasons, many DFLers are not thrilled with their party’s nomination of Mark Dayton. But most DFLers and lefties of all stripes are horrified at the prospect of Tom “cut taxes for business owners and programs for the poor” Emmer as governor.
The same is true in reverse. Many Republicans and old-fashioned pre-Tea Party-style conservatives are less than enthusiastic about Repub nominee Tom Emmer, but most will overcome their reservations if that’s the only way to keep Mark “Tax the Rich” Dayton out of the governor’s office.
Pretty much every poll has shown that that neither Dayton nor Emmer has locked down their party base. It makes sense that some Dems who aren’t committed to Dayton, and Repubs who aren’t committed to Emmer are flirting with the idea of voting for Horner.
This “flirting with the idea” bit is both Horner’s opportunity and his problem. If every Minnesotan who was ideologically to the right of Dayton and the left of Emmer were to vote for Tom “I’ll raise taxes less than Dayton and cut spending less than Emmer” Horner, he could pull into serious contention.
But enormous barbed-wire barriers stand between Horner’s current position and that outcome. He is still a relative unknown, does not have the kind of war chest that can buy name ID through massive advertising (nor does he have well-funded “independent” groups buying ads on his behalf as Horner and Dayton both do). The Independence Party lacks the massive computerized membership rolls and party workers to staff phone banks and provide get-out-the-vote operations that the Dems and Repubs have. Although he apparently made it a close question, Horner was unable to prevent the Chamber of Commerce from lining up behind Emmer. That was a blow to Horner.
But one — perhaps the biggest — Horner problem is that many DFLers — even many who might prefer him as governor — won’t vote for Horner if they think it might lead to Gov. Emmer. And the same for Republicans and the specter of Gov. Dayton.
Wasted vote syndrome
It’s sometimes called the “wasted vote syndrome.” Voting for your second choice to prevent your nightmare candidate from winning is also called “strategic voting.” There’s no way to handicap Horner’s chances without considering that factor, which, although it sends IPers to the moon with frustration, is a serious recurring problem for them.
The only cure is for Horner to convince people that he seriously can win, that a vote for him won’t be wasted. And the only way to do that is to do better and better in public polls.
Recently, that has gone fairly well for Horner. After scoring in the high single digits or low teens during July and August, he rose in three straight polls from 9 percent in a SurveyUSA poll in early August to 10 in an Aug. 12 Rasmussen to 13 in a Humphrey Institute poll of late August and then, in a SurveyUSA poll taken Sept. 12-14, Horner scored 18 percent, doubling where he started with the same pollster a month earlier.
Then on Friday, Rasmussen reported its just-completed poll this way:
“The latest Rasmussen Reports telephone survey of Likely Minnesota Voters shows Emmer earning 42% support to Dayton’s 41% when leaners are included. Independence Party candidate Tom Horner is a distant third with nine percent (9%) of the vote. Six percent (6%) like some other candidate in the race, and two percent (2%) are undecided.”
“This is the first survey of the governor’s race to include leaners. Leaners are those who initially indicate no preference for either of the candidates but answer a follow-up question and say they are leaning towards a particular candidate. Rasmussen Reports now considers results with leaners the primary indicator of the race.”
The big deal to Horner’s spokester Matt Lewis was that this description of the poll’s finding made it look as if Horner had dropped from 18 to 9 percent.
Horner at this point is something like the proverbial shark, which must move forward and can never swim backward. Polls that show Horner at 18 percent will not be enough to persuade anybody-but-Emmer and anybody-but-Dayton voters to risk a wasted vote on Horner. He needs to move across 20, then 25, and pretty soon. The election is five weeks from tomorrow. But if he drops back, his descent will accelerate as fewer strategic voters feel they can afford to risk him.
In fact, Horner did not drop to 9 percent. If Rasmussen had taken and reported this poll with the same methodology he used in his last MN guv poll, he would have corroborated the SurveyUSA poll (and the Strib Poll that came out Sunday) by reporting that Horner is the first choice of about 18 percent of likely Minnesota voters.
Rasmussen did say that in his writeup of the poll, just after announcing that his poll found the race to be Emmer 42, Dayton 41, Horner 9, when Rasmussen reported that:
“Excluding leaners, Emmer edges Dayton 36% to 34%, and Horner chalks up 18% support. Horner’s loss of support when leaners are added highlights the tendency in most races for supporters of third-party candidates to gravitate to one of the major party nominees as Election Day approaches. Last month, with leaners absent from the totals, Dayton held a 45% to 36% lead over Emmer… Horner… earned 10% support at the time.”
Can you follow that? I doubt it. I stared at it a long time, talked to some poll-savvy sources and no one could really figure out what Rasmussen was doing. I eventually got the full explanation from Scott Rasmussen himself, but that explanation is not even consistent with the way Rasmussen wrote up the poll.
But here’s the missing element that Rasmussen hinted at in the writeup of its new poll. As explained to me by Rasmussen himself, before Labor Day, his company just asks people whom they support. After Labor Day, they switch to a different survey that attempts to locate what they call “leaners” and attempts to figure out where the leaners will end up.
It goes like this (and bear in mind, Rasmussen employs the robo-dial, automated interview technology, so the likely voter is responding by pushing buttons on his/her phone to choices from a recorded voice):
First it asks you whom you support. (That one came out Emmer 36; Dayton 34, Horner 18. The other 12 percent either favored someone else or were not sure whom they favored.)
In a typical “leaner” poll, those whose first response was that they were not sure whom they supported might be asked if they were “leaning” toward any candidate. If they were, they might be lumped in with the supporters of that candidate or they reported separately as “leaners” toward that candidate. Most pollsters believe that “leaners” tend to end up supporting the candidate to whom they are leaning.
But if Rasmussen was using the traditional definition of a leaner — if only the 12 percent of voters who said “someone else” or “not sure” were offered a second chance to commit to a candidate — it would have been statistically impossible for Horner’s support to fall from 18 to 9 percent across those two questions. You can take my word for it or fool around with the numbers yourself. It just doesn’t work.
But that’s not what Rasmussen did. Respondents who have already indicated a preference for Dayton, Emmer or Horner — without calling themselves undecided — are then asked in a separate question whether they are certain they will vote for the candidate they named. In this poll, 54% of Emmer supporters said they were certain, 59% of Dayton supporters and 53% of Horner supporters said so.
Everyone who didn’t say they were certain was treated as a “leaner.” That’s a much bigger portion — almost half — of the sample.
The voice robot then asks those who were not certain of their candidate a couple of follow-up questions.
Unlike most pollsters, Rasmussen doesn’t release its actual questionnaires and was unwilling to tell me exactly how these questions were worded. It would be nice to know. Rasmussen is often criticized within the profession for this lack of transparency. But the follow-up questions ask the less-than-certain voters to imagine that it is time to make their final choice and say for whom they will vote.
This is the version of the poll that Rasmussen describes as “including leaners.” If Rasmussen is describing his questions accurately to me, it is inaccurate for the Rasmussen writeup of the poll to say that “leaners are those who initially indicate no preference for either of the candidates.” In fact, most of those who are being scored as “leaners” did indicate a preference on the earlier question.
Through a glass, darkly, I can glimpse what Rasmussen hopes to accomplish with this unusual method — especially in three-way races. As the Rasmussen writeup notes, there has been a tendency for third-party and independent candidates who make a strong showing at some point to fade as Election Day approaches. Experience has suggested to him that the “leaner” approach picks up the uncertainty that supporters of third-party and minor candidates are feeling. He didn’t say this precisely, but it’s almost as if in the few minutes they spend being questioned by a recorded voice, voters may go through a compressed version of the process that causes some voters who flirt with a third-party candidate to chicken out and switch to their second choice for fear of seeing the candidate they most dread win the election.
(The 1998 victory of Jesse Ventura in Minnesota should always be noted as a significant if somewhat rare exception to the tendency of third-partiers to fade. Polls did show him surging at the end of the campaign, but he was never ahead in any poll until he won on Election Day.)
As a matter of company-wide policy in all states, Rasmussen treats the complicated “leaner” version of the poll as the best indicator of the state of play after Labor Day. Although Rasmussen is criticized within the polling industry for failure to disclose his questionnaires, he does at least describe the two ways that he is measuring the race and the results he gets both ways. But by indicating the post-Labor Day “leaner included” model is the preferred method, most media references to his poll will adopt that one.
Rasmussen told me Friday that he sees some indications that Minnesota is a special case. Specifically, he said, the fact that Horner had risen from 10 to 18 percent on what Rasmussen calls the “first preference” question, and the fact that the portion of Horner supporters who say they are certain they will vote for him as almost equal to the portion of Emmer and Dayton supporters who say so, are both unusual for a third-party candidate. Rasmussen has decided to look closely at the next three-way poll in Minnesota and may — if those unusual factors continue for Horner — decide to emphasize the non “leaner” version of the numbers. We’ll see.
But Lewis of the Horner Campaign is correct to argue that the latest Rasmussen Poll does not show Horner fading. Apples to apples, it shows that he has risen over the past month. Lewis does agree that if Horner cannot show continued growth in his support, the traditional wasted vote argument will hurt him down the stretch. “We have to acknowledge that the wasted vote syndrome can be a real syndrome,” Lewis said.
As far as the main thing that the polls are designed to measure, both the leaner and the non-leaner versions showed Emmer barely leading Dayton, certainly within the margin for error. That’s good news for Emmer, who has trailed Dayton in nine out of 14 of the polls in the race going back to March (that includes three of the four previous Rasmussen polls).
On an apples-to-apples basis — using the last two Rasmussen polls and the non-leaner method — the race changed from a Dayton 9-point lead in August to an Emmer 2-point lead in mid-September.
The good news for Emmer was offset two days later by the release of a new Strib poll showing a big, statistically significant 39-30 percent Dayton lead over Emmer (with Horner at 18). This one had a bigger sample (949 likely voters, compared with 500 for Rasmussen) and was taken by the longer established method of actual human-to-human interviews.
The Humphrey Institute’s Larry Jacobs, whose own poll three weeks earlier had shown Emmer and Dayton tied at 34 percent, noted that the Strib was the only poll in Minnesota that was reaching Minnesotans on cell phones. This is expensive, and increasingly important, and more and more younger Americans use cellphones and have no landlines, Jacobs said.
But Jacobs was also concerned that the inclusion of cell-phoners might have overstated support for Dayton and understated support for Emmer because cell-phoners are disproportionately found among the young. And 18-34-year olds are the age group most favorable to Dayton (in the Strib poll, he carries younger voters by 42-24 percent).
Somehow, the Strib poll ended up categorizing 77 percent of its interviewees as likely voters. It would be very surprising if any state — even Minnesota — had a 77 percent turnout in a midterm election. (Jacobs’ own poll projected roughly a 60 percent turnout. Minnesota led the nation in the 2006 midterm with a 60 percent turnout.)
I should note, although I do not subscribe to them, that there are long-standing Republican suspicions that the Strib’s polling is biased against Repubs; and Dems often make the same argument, in reverse, that Rasmussen’s polls are suspiciously favorable to Repubs.