Kennedy in trouble? Wetterling out?

The political wags are having a field day today with news that Patty Wetterling has scheduled a news conference to discuss her “future plans.” Usually, that’s code for “I’m dropping out.” If true, it’s lousy timing. A Zogby poll out this afternoon might suggest she’s the frontrunner in the race, except few people seem to believe that.

Still Rep. Mark Kennedy’s performance in these polls is not inspiring.

The Zogby poll shows Patty Wetterling over Kennedy now 50.8-to-39.9 percent. Amy Klobuchar has a 48.6-to-42.8 lead over Kennedy. That tends to mirror a Rasmussen poll from a month ago.

Even Ford Bell, who’s gotten very little traction — and very little attention in this race, is close to Kennedy — 43.3-to-42.8 percent.

If a virtual unknown is knocking off Kennedy, it could mean two things. (a) Those surveyed really don’t like Kennedy or (b) The poll might be bogus. It’s unlikely that Ford Bell’s message has resonated to the point where it’s elevated him to a competitive state, since few people have heard his message.

Some of those dismissing the poll have suggested it’s an online poll and, thus, has little scientific value. But, it’s not. Here’s the Zogby methodology. Zogby does provide online polling for Web sites. But this particular survey is not an online poll. Is the survey accurate? Who knows? As with most surveys, take them at face value and look for trends. If a bunch of polls start to say the same thing, then maybe it’s time to sit up and pay attention.

Minnesota GOP boss Ron Carey put out a release that said:

“On Election Day in 2004, John Zogby confidently predicted that John Kerry would defeat President George W. Bush and get at least 311 votes in the Electoral College. In 2002, Zogby had Walter Mondale defeating Norm Coleman by six points in his final poll.

The thing with polls is they don’t predict anything. They only tell you what a bunch of people who were surveyed happened to think on the day they were surveyed. Polling firms strive to have those surveyed be representative of the larger group (that is: us), but even then it only represents what the larger group thinks on a particular date they were asked. Nothing more.

(Update 1/20)- I checked on the Zogby poll in the ’04 elections and the last one appears to have been taken on behalf of Reuters on 10/29-31 2004 and at that time showed Bush/Cheney with a 1% lead over Kerry. A Zogby poll two weeks before that showed Kerry narrowing his lead, but still trailing Bush by 2%. A further reading of Carey’s release shows that Carey isn’t disputing Zogby’s polling, just John Zogby’s opinion. The methodology of Zogby’s polling in the final weeks appears to be fairly solid, actually. (End update)

Kennedy’s campaign said today that with the news that’s been about for the past six months, and Kennedy being a Republican, being behind in the polls now isn’t “the worst case scenario.”

But what of Wetterling? If she were to drop out, a run in the 6th District is the obvious solution since the DFL is putting up very little fight there, with only Elwyn Tinklenberg carrying the DFL banner. And since she had a respectable showing in the district in her race against Kennedy in 2004, one would think she’d be in a strong position against the GOP.

In the gubernatorial race, Mike Hatch has a slight lead over Tim Pawlenty. But Pawlenty has slight leads over challengers Steve Kelley and Becky Lourey.

Meanwhile, Rasmussen Reports is out with a poll today in the race for governor of Pennsylvania showing Republican Lynn Swann, the former receiver for the Pittsburgh Steelers, narrowly leading Democratic Governor Ed Rendell 45 percent to 43 percent.

  • Bob Collins

    I’ve gotten a couple of e-mails and voicemails on the polling methodology vs. prediction comment above. And there is a parallel discussion taking place on the MPR Forum. So here’s more on that:

    It doesn’t matter whether the marketplace or people or whomever take Zogby seriously or not. That’s entirely up to them and has nothing to do with methodology.

    Let’s face it: people believe whatever polls give them whatever results they want, and tend to dismiss those that have rather distasteful results.

    I get that. I think that’s fine. People can and should be able to believe whatever they want to believe.


    There’s all sorts of GOOD reasons in the various poll methodologies to disagree with the polls and that, it seems to me, is the only intelligent way to ascertain what polls to believe and what polls not to believe.

    For example, some polls survey “likely voters”. Others survey “registered voters.” Those aren’t the same thing. Some polls weight the number of Democrats vs. the number of Republicans based on likely voting patterns (I believe this is part of the Strib methodology). That is also a practice that can be debated and should be factored into what polls to believe.

    If somebody were to say to me, “I don’t believe that poll because that company has been wrong before,” that gets me curious. It doesn’t get me convinced. Were they wrong? Why were they wrong? What is their methodology.

    Zogby may well be wrong in this particular poll. As I said, I don’t know. The Ford Bell results certainly raise a red flag. On the other hand, the Klobuchar-Kennedy results are pretty consistent with the Rasmussen results.

    So if they are wrong, if people think they’re wrong, I think that’s fine. But so far nobody has dived into the methodology to analyze where the weakness is.

    I would welcome that, and you know, this comments section here actually works. (Hint!)

    The other point gets to the “proving right or wrong.” Well, how? How can you prove whether a poll taken on the 16th of January is right or wrong?

    Now Zogby did two polls before the November election in 2004 (the last two polls they did) showing Bush with a lead (but within the margin of error). Were they wrong? Based on what? The November election? That’s wrong too.

    A gentleman from the Republican Party in Minnesota kindly sent me a copy of a New York Times article from November 7, 2004 that he said (according to my voicemail) shows that Zogby had a poll showing a Kerry win on election day when he made his prediction. The article he e-mailed (and I very much appreciate that, by the way) me said this:

    John Zogby. In what he described as a ”bungee jump,” the pollster flatly predicted in May that Mr. Kerry would win the election. He went on maintaining through the summer that the race was ”Kerry’s to lose.” Then, even though his own election-eve poll showed Mr. Bush a point ahead, Mr. Zogby took an even bigger plunge on Election Day.

    Late Tuesday afternoon he predicted that Mr. Kerry would win Florida, Ohio, Iowa and New Mexico (none of which he did) and get at least 311 votes in the Electoral College, while Mr. Bush was assured of only 213. (Mr. Zogby modestly declined to forecast the remaining 14 electoral votes).

    ”I did something I shouldn’t have,” Mr. Zogby cheerfully confessed on Thursday. ”I am a better pollster than predictor.”

    He said that his Election Day prediction was inspired not by the faulty exit polls but mainly by his own polling among young voters and field reports of high turnout among the young.

    ”I don’t know that anyone was hospitalized over my prediction,” he said. ”If there are any orphans that are out there, from the bottom of my heart I apologize. We’ll try to start up a fund.”

    The only problem with that article is that it actually doesn’t show there was a poll that supported Zogby’s stupid prediction. In fact, it showed there was a poll that said the opposite. Zogby, in fact, acknowledges he ignored his firm’s polling in favor of some personal inteviews or something (note: The Times article made reference to “faulty exit polls,” but didn’t indicate what “faulty exit polls” there were).

    As I cited earlier the Zogby exit poll on the record showed Bush ahead by .4% on election day.

    So, yeah, people can believe what they want to believe, but it also calls up the old adage about not having the right to change the facts.

    Folks are certainly free to disagree with poll results. It means more to me, though, if we can pull apart the methodology.

    Can we do that?