Blue Mass Group

Reality-based commentary on politics.

New poll! Patrick still ahead; Gabs and Reilly tied for 2nd

By

Here’s the latest from Survey USA and Channel 4.

429 likely Democratic primary voters.  MOE: +/- 4.8%.  Confidence: 95%.  More on interpreting the MOE.  Data collected 8/19-21.  Results from 8/3 poll in (parens).

Patrick: 34% (35%)

Gabrieli: 30% (30%)

Reilly: 30% (27%)

Undecided: 6% (8%)

You can see all previous Survey USA polls, sorted however you want, at this excellent page on the Survey USA site.  Here, in graphical form, is where these polls have been over the last few months:

My quick take: good news for Tom Reilly, who is right back in it.  No one should count this guy out too early.  Not-so-good news for Chris Gabrieli, who is unchanged despite a gajillion dollars spent in the time frame between the two polls.  A push for Patrick, whose ads had barely gone on the air when this poll was taken.

Please share widely!
fb-share-icon
Tweet
0

Comments

  3. Channel 7News and Suffolk University came out their own poll today as well.

    <

    p>
    Gabrieli 32%
    Patrick 24%
    Reilly 20%
    Undecided 24%
    MOE +/- 4% at 95% confidence.

    <

    p>
    I take all these polls with a grain of salt, although I do think the Undecided number is a lot more accurate than the Survey USA poll.

  4. I don’t really see Reilly getting past the baseline of 33% at the start of polling (at least in the opinion polls).  I see him picking up the folks who were with him and strayed away, but unless something happens to significantly change the dynamics, that seems to be the ceiling.

    <

    p>
    That said, the question is the depth and commitment to come to the polls.  I think Deval’s core and Reilly’s seniors are the most likely to actually show up at the polls.  I wonder how much of Gab’s support comes from people who respond positively but don’t actually have the drive to wander down to the polling place, or who write the primary date onto the calendar and make it a point to vote before going off to work.

    <

    p>
    The other question – are there any other hot local primary races that could generate turnout?  I know there are some races out in the Berkshires, where a strong turnout should help Patrick, but is there anything else people think will generate passions and turnout?

    • There are two ends — supporters going to the polls and campaigns pulling them out.  I’ve gotten flamed for saying that Reilly will have a good GOTV thanks to union support (those who think unions can’t do gotv need to look at some  data from ’00 and ’04 in PA or MI).  Money translates poorly into GOTV — see Lieberman 06 —  and I think more and more that will be Gabs’ weakness.

      <

      p>
      Deval’s people are committed and loud, but can they do GOTV? 

      • I agree that unions can help enormously with GOTV, but I don’t see Reilly having that much real activist union support, outside of SEIU 1199, NAGE, and the corrections officers unions.  Remember that Tom Birmingham had incredible union support in 2002 – I believe that translated into higher actual votes than the pre-election polls showed, but not enough to overcome O’Brien’s lead and stronger, more widespread grassroots organization. 

      • Ummmm, Deval’s people are led  by John Walsh and Nancy Stolberg – if they can’t do GOTV, no one can.  They’re true professionals, and the kind of people you want to go into battle with.  I’ve worked with both of them, and they are the real thing.

        <

        p>
        There are also capable people with Reilly and Gabrieli.  Reilly will be concentrating on his surrogates to help turn out, along with robo calls to certain subgroups, like elderly voters.  Gabrieli will use his resources to poll every likely voter before 9/17, and he’ll use paid phone banks and robo calls to get the ID’d voters to the polls.

        <

        p>
        Patrick will be relying on his hard core supporters to make sure they turn out their ID’d voters.  My sense is that Patrick’s ID’s will be much more real than the other candidates, because they are coming from one-on-one voter contact, made by people like me, but the GOTV methodology is a new one.  I think it’s a lot like the Republican efforts in 2004, which were not like traditional campaigns – much more under the radar – but much more effective than expected.

        <

        p>
        Unless the dynamics change, like from the final debates, it will all come down to who can ID and turn out their supporters.

        • That Deval probably has the most and best “hard” IDs.  That only counts for so much.  I was in on a campaign for a state legislator going for a sixth term.  Though popular in his district, with a strong organization, we only had hard ID’s of about 1/5 of the people who voted for him.  After 10 years on the ground!

          <

          p>
          Hard IDs are an indication of campaign strategy as much as anything….

  5. From Suffolk’s announcement of its new poll results:

    <

    p>

    Fueled by more than $5 million in paid television advertisements and direct mail, venture capitalist Chris Gabrieli has taken control of the Democratic Primary and is poised for a decisive victory in November, according to a 7NEWS/Suffolk University poll released today. …
    “Chris Gabrieli is clearly the person to beat in both the primary and the final election,” said David Paleologos, Director of the Suffolk University Political Research Center. “His widespread strength cuts across all demographics including gender, party, age, and geography.”

    • The Suffolk University poll was conducted August 17 thru August 21, 2006. The margin of error is +/- 4.00% at a 95% level of confidence. All 600 Massachusetts respondents indicated they were registered voters and had voted in the 2002 gubernatorial election.

      Suffolk University is scheduled to release 376 pages of cross-tabulation data, marginals, and charts on its website (www.suffolk.edu) on Tuesday, August 22, 2006.

  6. So, I decided to average the two polls conducted over the weekend, and guessed the Suffolk survey size from the average sample size they’ve used before (246, a little less than half the whole poll)

    <

    p>
    DP: 30.52%
    TR: 26.54%
    CG: 30.77%
    Und: 12.16%
    MOE: 3.85%

    <

    p>
    If anything could be call a “statistical dead heat”, I think that would be it. Looks about 41/42% chance for Patrick to a 44/45% chance for Gabrieli.

    • That tells us very little because the questions and methodologies are so different in the two polls.  It’s OK to do for fun, but don’t think you’ve modeled the feelings of the electorate in any theoretically robust way.

      • We have dueling methodologies that have produced rather different results — one has Gabs up by 8, the other has him down by 4. Other than the Gabs cheerleaders telling us that the Suffolk poll was better four years ago, we don’t really have any basis for judging which pollster has crafted the better question, let alone how well they have constructed their samples to approximate who will actually turn out in four weeks. I’m a Patrick supporter, and perhaps biased in his favor, but I don’t think the net results of these two polls, plus the Healey one that’s been leaked, justifies the conclusion that “Gabrieli is pulling away.”

        <

        p>
        And that’s before you get to the issue of the polls’ timing. When were these interviews conducted? Gabs has been on the air for a month or more, Patrick started four days ago. Once Patrick has more time on the air behind him, this is likely to help his numbers — clearly his ads are most of what has pushed up Chris’ numbers.

        <

        p>
        Gabs has bought himself a place at the Primary final table. This is not a surprise. But it’s very early for his supporters to be popping the champagne corks.

          • Yes, but how much is Patrick on the airwaves.  I’ve seen Gabrieli’s ads and Reilly’s ad, but I haven’t seen anything from Patrick except what I’ve seen on his website.  Is he able to effectively match Gabrieli with airtime?

      • I don’t think that alone is a very good model of the feelings of the electorate in an at all robust way, but I do think it gives a better and more accurate impression than the two polls individually. It increases the sample size to create smaller error margin and should average out biases caused by the alternate methodologies. To correct my previous numbers now we know the Suffolk sample size:

        <

        p>
        DP 29.70%
        TR 25.31%
        CG 30.95%
        U 14.16%
        MOE 3.54%

        <

        p>
        Now, personaly I find averaging polls of different methodolgies from similar times seems the best way to increase accuracy, based on my experiences with statistics as a physicist. I noticed from your bio on yr site that you have a degree in comp and poli sci, so you probably no more than me about numerical poli sci theory than I do, but maybe if you know a good reference on the topic, let me know.

        <

        p>
        My personal amalgamation of all polls so far, weighted by exp(-t/14) where t is the number of days between the poll and election day:

        <

        p>
        DP 30.39%
        TR 25.47%
        CG 30.20%
        U 13.94%
        MOE 8.88%

        <

        p>
        It’s a tight race. We’ll see how things change once Deval’s on TV more.

      • I added the raw numbers of the two polls together, and just treated it as one large sample. Helps average out biases, such as those that Sco mentioned.

    • …i think he has the gist right.  dead heat for DP and Gabs. 

      <

      p>
      this is close to best scenario DP could have hoped for.  he can’t eclipse 38% plus 2% bonus for GOTV b/c he’s (often perceived as) leftmost out of 3. 

      <

      p>
      therefore, he needs reilly to stay +20% and the TV time for Big Dig seems to have done that. 

Archives