By Dan Kennedy • The press, politics, technology, culture and other passions

Polling the Senate race

Good luck making sense out of polls about the Massachusetts Senate race.

Following Democratic candidate Martha Coakley’s even-bigger-than-expected victory in the Dec. 8 primary, most political observers had assumed she would cruise in the final. That assumption has been looking questionable since last week, when a Rasmussen poll showed Coakley with just a nine-point margin over her Republican challenger, Scott Brown.

Then, last night, Public Policy Polling released the results of a survey showing Brown actually leading Coakley by a margin of 48 percent to 47 percent. Let the tea party begin!

A few hours later, the Boston Globe published a story about its own poll, in which Coakley is maintaining a comfortable 15-point lead.

So what’s going on here? Who knows?

Frankly, I would start by throwing out the Public Policy Polling survey — it’s a robocall. (“If Scott Brown, press 1. If Martha Coakley, press 2.”) Would you hang on the line? I wouldn’t.

I’ll also point out that the Globe’s poll was conducted by the University of New Hampshire Survey Center, a highly respected operation. I’m no polling expert, but I do know that Rasmussen gets mixed reviews.

Also, as best as I can tell from diving into the fine print, it looks like the Globe/UNH poll was the only one of the three in which respondents were specifically asked about the third candidate in the race, libertarian independent Joe Kennedy, who receives a not-insignificant 5 percent. Indeed, given the vagaries of polling, that alone could explain the difference between Rasmussen’s nine-point margin and the Globe’s 15-point spread.

What’s making everyone hypercautious is that we have absolutely no idea who’s going to turn out in the Jan. 19 special election. And what if there’s a blizzard?

My guess, though, is that Coakley’s right where you’d expect her to be with a little more than a week to go.

Discover more from Media Nation

Subscribe to get the latest posts to your email.


Tomorrow’s corrections today


Anti-Coakley (and anti-Brown) push-polling reported


  1. Harrybosch

    Reminds me of those heady days back in ’94 when folks seriously thought (me included) that Romney had a shot.

    Brown won’t crack 40%.

  2. Peter Porcupine

    DK – you are correct, she is exactly where I would expect her to be, given her inept and arrogant campaign style, the ignorance of Mass. Dems of valid concerns of voters about the DC horsetrading on health care, and simmering resentment by Capuano voters who would rather suffer a GOP for two years than a Coakley for decades.

  3. paul

    It looks like what we have is a tricky electorate to forecast in a special election. Brown’s supporters appear to be a combination of a small but highly motivated group of voters who are supporting Brown as the way to kill “Obamacare” once and for all plus a larger (but probably less motivated) group of moderate independents who are tired of sending nothing but Democrats to Washington. Meanwhile, Coakley has a palpable enthusiasm gap among her supporters, even though they seem to be a clear majority.

    The thing is that this looks a lot like the Democratic primary — the polls were showing Coakley leading but the enthusiasm was with her opponents. But, when the votes were actually counted, Coakley won handily. Are we seeing a similar dynamic here? And will the fact that the race appears to be closer than expectations help to light a fire under the Coakley GOTV effort?

  4. Michael Pahre

    Nate Silver describes Rasmussen’s polling in the 2009-2010 election cycle thus far as having a national “house effect” leaning towards Republicans at 4-5 percentage points.

    Silver is the only prominent analyst using any kind of sophisticated mathematical tools, so I would take seriously his result.

    Note that Rasmussen’s house effect in the current election cycle doesn’t mean that Rasmussen is necessarily wrong; everyone else might turn out to be wrong. It only means that Rasmussen is consistently leaning more Republican in its polls by 4-5 percentage points. The accuracy of its polls only get determined by a check against the electorate in an actual election, which will happen on January 19th here in Massachusetts.

    • Dan Kennedy

      @Michael: Rasmussen gave Coakley a nine-point lead. Let’s say Rasmussen’s methodology overstated Brown’s support by four points. It’s also clear that Brown was awarded another three or four points by Rasmussen’s decision not to ask respondents about Kennedy. (“Other candidates” gets 1 percent.) You can see where this is going.

  5. LFNeilson

    I realize that polling has become an accepted part of politics, but I still don’t like the practice. I prefer to evaluate a candidate on philosophy, positions, accomplishments and voting record. Does a poll tend to skew the voters’ choice? I think so. Are they accurate? I think not. Polling calls rank one peg above vinyl siding pitches in my book. They always put me in a funk.

  6. mike_b1

    PP, sounds like you are projecting — and not by a small amount.

    Nice try, though.

  7. Al

    I don’t see the Coakley campaign as arrogant. I only see it as passive and uninspiring, as would be expected when led by a career prosecutor practiced in speaking like the fine print at the bottom of an insurance policy – don’t get painted in a corner & always leave a legal out. Any expression of arrogance is only the voicing of frustration that Brown can’t seem to get at her. Now, as for Capuano voters staying home, and putting up with 2 years of a Republican, instead of her, I don’t think so. I voted for Capuano. Once the primary was over, Coakley got my support. I will be out and voting bright and early, regardless of any weather. I don’t want Brown, or any Republican representing me for even 1 day, let alone 2 years. I particularly don’t want someone representing me who brags about wanting to be the vote than stops the health reform.

  8. Tom G

    Fixed it for you:

    My guess, though, is that Coakley’s right where you’d expect her to be – having spent so much of her time since December 8 at home in Medford – with a little more than a week to go.

    Martha’s rope-a-dope success in the primary led her to decide to pursue a rope-a-dope strategy in the general, which may cost her and Democrats the seat.

    If Martha is a hard worker, if she’s a well-rounded liberal not a party line voter, if she’s a clever legislator and a strong debater and a leader, she has not put those characteristics on display in her campaign. It’s as if she thinks her resume guarantees a victory. It doesn’t.

    I’m going to write-in Khazei. He campaigned like he really wanted the office. He put his values and policies on the table for all to assess and vet. Martha’s largely a mystery and it’s her fault becuase of how she chose to campaign. We deserve better.

  9. I believe it will be much closer than most people expect

  10. Ben

    In spite of conventional wisdom, polls usually show Libertarians drawing votes equally from both parties. The Globe poll shows this. In fact, Kennedy gets a lot more support from voters who are favorable towards Coakley vs. voters who are favorable towards Brown. Of course, this poll identifies Kennedy as an “Independent”, so it may not be possible to draw conclusions from this.

    The Globe poll also shows Kennedy voters to be significantly less interested in the race. So in spite of my high hopes for him, 5% on election day would really surprise me at this point.

  11. Dunwich

    “Does a poll tend to skew the voters’ choice?”
    Does talk radio have an influence: do newspaper endorsements? If you like Brown, what would you care what a poll said (or an editorial board or talk host)?

    Are they accurate? I think not.

    Although there will always be anomalies, of course they’re accurate. Why would news organizations and candidates pay for inaccurate information?

    With all Coakley’s faults, she’s still has the gender factor combined with an AG’s perceived strong will and determination. That and voter registration.

    • Dan Kennedy

      @Dunwich: I would suggest that a voter inclined to support Kennedy will do so if he thinks Brown has no chance, but might vote for Brown if he thinks Coakley could be defeated.

      Turnout is also likely to be higher if there’s a perception that the race is close.

      So yes, polling matters.

  12. here in Boston suburbs, I just got “push polled” on the election.

    Oddly, they identified the candidates by party first “Republican Scott Brown” *, etc. and then asked if I supported either one (no mention of the faux Kennedy libertarian).

    When I said Coakley (in fact, I already voted absentee in case of bad weather), they asked if it would change my vote if I knew Coakley supported “tax payer funding of abortions”.

    Call came in around 8:40 Sunday night from DC number: 202 461-3440.

    Reverse lookup tells me it’s a landline in Westchester, DC and is unpublished.

    * odd because GOP label is pretty toxic here.

  13. Thomas

    I wonder how many of these people that spend their time blaming the Dems for everything imaginable in the comments section of every news website in this state actually vote?

    I doubt many of them do……….You look at these comments and you have hundreds of people speaking of their disgust of the Dems and just a few standing up for the them, and yet come election time it is the Dems winning by a landslide.

  14. ben

    Regarding the comment from another Ben above, Dan’s buddy Nate Silver makes the point that third party candidates draw much higher in polling so the Globe poll may not be doing the best job in projecting by including him.

  15. Ben

    To the other ben: You’re right, third party candidates generally do a lot better in polls than on election day. But making the poll look less like the actual ballot is hardly a good solution. At best, it’s only pretending that the poll doesn’t have a serious methodological defect.

Powered by WordPress & Theme by Anders Norén