Pollster bragging rights: KSTP-TV, Minnesota GOP

MPR’s Bob Collins beat me to it, but since I’ve written lots about media polls this cycle, might as well finish the thought.

Let’s not begin without acknowledging a major caveat: polls are taken a few days before Election Day, the electorate can shift in the remaining days, and getting the number exactly right is often a matter of luck.

Still, the undisputed polling champ was KSTP and its pollster, SurveyUSA, whose final poll pegged Mark Dayton with a 1-point win. As of this writing (the ungodly hour of 4 a.m.), Dayton leads by 0.4 points.

Public Policy Polling, which surveyed only once, late into last week, gave Dayton a 3-point margin, which is pretty darn good. By the way, SUSA’s final included cell-phone-only voters and PPP’s didn’t (though it had a huge sample size).

The poll that will have tomatoes thrown at it and crude jokes made at its expense is the MPR/Humphrey Institute’s, which showed Dayton with a 12-point lead (16 percent if you counted leaners, according to HHH Prof. Larry Jacobs).

The GOP immediately called B.S. on that, and they were right. 

Jacobs and his crew have some ‘splainin’ to do, perhaps in a command performance before MPR CEO Bill Kling and news boss Chris Worthington. It’s one thing to consistently err on the side of Democrats, but that huge a gap can’t be explained by shifting sentiment.

The Star Tribune and Rasmussen played it safe, eschewing final-week polls for temperature-taking two weeks out. The Strib once again pegged the DFLer too high (+7 Dayton), while Rasmussen had it +3 Dayton. Nate Silver of FiveThirtyEight.com noted that nationally, Rasmussen overestimated the GOP in 75 percent of its U.S. polls this year … but not locally.

St. Cloud State was also a D-tilting outsider, rating it Dayton +10 a couple of weeks out. I guess this only fuels conservative complaints about liberal bias on campus.

A couple of other near-feathers in SUSA’s cap: they forecast First District DFL Congressman Tim Walz could win by 9; he’s leading by 5. They also had veteran Eighth District DFL Congressman Jim Oberstar up by a mere 1, confirming a GOP internal poll that shocked some in the political world. Republican Chip Cravaack has a point-and-a-half lead, but even though SUSA got the winner wrong, they got the drift right.

I have to stop here to doff my cap to the Minnesota GOP: Not only was their Cravaack claim credible, and their MPR criticism justified, but the internal poll they released in response to the HHH numbers — showing Emmer tied — may have been the best survey of all, since the governor’s margin is likely to end up inside 0.5 percent. Kudos to their number-cruncher, Public Opinion Strategies.

It’s a shame no media outlet took a whack at a legislative poll, since the likely GOP takeover was the biggest surprise of the night. With so many races, they’re tricky to do, and probably pretty expensive too. But even though everyone picked the GOP to gain seats, I didn’t hear anyone talk switch. On a night when the pollsters snuffed most of the surprises, the numerical blackout made that shock much more startling.

Comments (11)

  1. Submitted by Joseph Skar on 11/03/2010 - 05:51 am.

    What a total failure in political polling. Thanks for ripping the SUSA polls for the last few months on this blog.

  2. Submitted by Brian Simon on 11/03/2010 - 08:10 am.

    Good post, particularly that last point about the complete absence of legislature polling.

  3. Submitted by David Brauer on 11/03/2010 - 08:35 am.

    Joseph – you have a selective memory. If you read the previous (last pre-election) post on SUSA, you’ll see no “ripping,” though there is a historical analysis of how they’ve fared in Mn.

    (Let’s not forget that, even though they have nailed the last two governor’s races, SUSA missed the Franken win by 5 and were 7 off on the Obama number. They’re not HHH, but they’re not infallible.)

    I’ve written repeatedly how highly rated SUSA is by an analysts such as the New York Times, and provided sober analysis of their methods, such as:

    http://www.minnpost.com/braublog/2010/09/21/21633/kstp_pollster_explains_excluding_cell-phone-only_voters

    Your side had a good night last night, but don’t milk it.

  4. Submitted by Thomas Swift on 11/03/2010 - 08:40 am.

    The HHH/MPR team play an important role in Minnesota politics. Their wacky polling leads Democrats to believe they have comfortable margins, which keep the leftist turnout low.

    That means Hennepin county is forced to manufacture more car trunk ballots, which leads to a couple of amusing weeks of non-plussed denials from the Democrat party.

    When I retired last night, Henn cty was reporting about 200k more votes than it has registered voters. They said it was a software glitch, and I was confident they’d have the numbers fine tuned by morning…I was right.

  5. Submitted by Joseph Skar on 11/03/2010 - 08:49 am.

    David – How many posts did you have ripping the methodology, come on. We need cellphones… landlines only? Moral of the story Larry Jacobs should be fired, 12 points what a joke.

  6. Submitted by David Brauer on 11/03/2010 - 08:57 am.

    Joe – bad argument. SUSA added cellphones to their final, very-accurate poll, so they obviously felt compelled to change their methodology due to *legitimate* methodological criticism. HHH didn’t have them.

    I guess if you’re going to conflate criticism and questioning for ripping, then I did. But I suspect most readers understand the difference.

    By the way, the other methodological analysis I did was on likely voter models. I didn’t criticize SUSA there, but explained their particular flavor.

    (By the way, I included SUSA president Jay Leve’s criticism of HHH’s likely voter method, which was far sharper than anything I wrote about SUSA.)

  7. Submitted by David Brauer on 11/03/2010 - 09:04 am.

    Tom – it’s hard to refute your speculation about HHH and turnout, since we don’t know how turnout was affected, but a boy can dream.

    And of course, the “car-trunk ballot” myth doesn’t become truer upon repetition. Only the bitter-enders (“rightists”) still cling to that one.

    (If you’re new to the conversation, see: http://www.minnpost.com/davidbrauer/2008/11/12/4565/minneapolis_election_director_speaks_ballots_in_my_car_story_false)

    That said, I doubt anyone will listen to the siren song of the HHH poll from here on out, feel free to shout from the rooftops on that one, as long as you don’t make stuff up.

    I definitely think GOP stalwarts can do some poll-bashing this morning, but as we see, winners often overdo it. That frequently leads to becoming losers.

  8. Submitted by John Edwards on 11/03/2010 - 11:43 am.

    The Star-Tribune and HHH/MPR were not the only liberal polling organizations that overstated Dayton’s total. The left’s polling gold standard, New York Times’ fivethirtyeight blogger Nate Silver, had it Dayton 45, Emmer 39, and Horner 14.

  9. Submitted by Stan Daniels on 11/03/2010 - 06:46 pm.

    Nice wrap-up David. Considering you are an admitted lefty, I thought you did a nice job keeping your analysis balanced.

  10. Submitted by David Koski on 11/04/2010 - 01:19 am.

    It is possible that most of the polls were correct and vote tampering was what altered the outcome. To throw out respected polls and go with an obviously biased KSTP is rather cowardly. Tony Sutton is taking the “high ground” and accusing the Democrats of tampering. Regardless of who blames first, do some research, many know that elections are fixed with software manipulation, the evidence is out there. Kudos to Public Opinion Strategies? With an acronym, POS, what an unfunny joke on the populace.

  11. Submitted by Kevin Slator on 11/11/2010 - 05:48 pm.

    He’s a wonk and not a pollster, but I heard Blois Olson say a couple of times before the election the winner would be “Dayton in a recount.” How much better could a prediction be?

Leave a Reply