Blue Mass Group

Reality-based commentary on politics.

Pollin’ pollin’ pollin’: It’s a horse race…

By

New poll out: CBS4/SurveyUSA, 700

voters:

  • Healey 35%, Patrick 30%, Mihos 20%

  • Reilly 36%, Healey 31%, Mihos 22%

Link from CBS4/Jon Keller. Mihos says he’s “stunned”. Keller calls it a “boomlet” for Mihos.

Whatever. It’s March. Eight more months of Christy’s Henny Youngman

act. Patrick’s in the hunt. Reilly’s in with the pack. Call it even

between Patrick, Reilly and Healey; and I’d do a “put” on Mihos’ stock.

Please share widely!
fb-share-icon
Tweet
0

Comments

  1. You’re right to call that essentially even. Laymen won’t agree…but they don’t know much about political polling.

    <

    p>
    In any event, all Patrick should worry about now is polling above Reilly. Once he defeats Reilly, Healy can be taken out pretty easily, especially if Mihos is still in the race. Healy is no Romney, that’s for sure. People are sick of Romney and her ties to him are going to hurt.

    <

    p>
    Plus, I’ll be out and about for Patrick… and I’m a political genius, so he’s bound to win! mwahaha

  2. Interesting breakdown though. Look at the blacks for Patrick in the general: 46%, to Reilly’s 54%. Also, Deval’s Cape Wind may have hurt him in the Cape region. Interesting Age breakdown as well…

    • You can’t draw any conclusions from the size of those  subsamples.  There were 35 total Blacks polled.  The cape region is larger (140 responses) but remember that region is more Republican — particularly Plymouth County — than elsewhere in the state.

  3. sco is absolutely right about those subsamples.  They are waaaaay too small, even the Cape region (or any region).  You’re talking a margin of error (at 95% confidence level) somewhere on the order of 8.3%-10.9%.  CBS4 should be embarrased to have released a poll this flawed and they should fire their pollster.

    <

    p>
    Here’s the biggest problem with this poll:
    “Filtering: 800 Massachusetts adults were interviewed 3/3/06 – 3/5/06. Of them, 700 were registered voters.”

    <

    p>
    One eighth of the survey respondents aren’t even registered voters.  Why in hell would you even include them in the poll?  Even worse, look at the survey instrument.  …No screen for likely voters.  Even the 700 who were included could be bad voters, voters who never vote in gubernatorial elections, etc.  This poll is just too flawed to be taken seriously.  Sorry CBS.

    • Your criticism of the poll is slightly off base.  While I agree with you that a screen for likely voters is useful (though they have their own problems and biases), non-registered voters were not included in the poll results.

      <

      p>
      And to be fair, the subsamples of most polls are going to have large margins of error.  Those results should not take away from the expected accuracy of the poll’s top-level numbers.

  4. OK:

    <

    p>
    Yes, it’s just one poll.
    Yes, it, like all polls, is flawed in some ways.
    Yes, we need more polling to determine the new trend.
    Yes, Mihos just got in.
    Yes, when Healey announces her running mate, there will be another shakeup.

    <

    p>
    BUT

    <

    p>
    If, HYPOTHETICALLY, every poll from now till September shows Reilly beating Healey by 5 points and Patrick losing to Healey by 5 points, then it is remarkably ignorant to call it “even” between Reilly and Patrick.  To call it even is as ridiculous as after the 2004 New Hampshire primary when Joe Lieberman came in a distant 5th but called it a three-way tie for third.

    <

    p>
    NOTE: I am undecided in the gubernatorial race still – I might decide to support Patrick and I might decide to support Reilly – and if Gabrieli gets in, I might decide to support him.
    NOTE: Whoever wins the Dem primary, I will support and volunteer for against Healey (and Mihos).  If it’s Reilly, great; if it’s Patrick, great; if it’s Gabrieli, great.

    <

    p>
    I recognize that many of the bloggers and many of the posters are already Patrick supporters.  Good.  Great.  But just be objective about poll numbers.

    <

    p>
    When one Dem is beating the GOoPer by 5, and one Dem is losing to the GOoPer by 5 – it’s not “even.”  The poll may be flawed, and we still need more data before there’s a trend; but, in the vacuum of the one poll, it is certainly not “even.”  Take those rosy glasses off and be more objective.

    • If, HYPOTHETICALLY, every poll from now till September shows Reilly beating Healey by 5 points and Patrick losing to Healey by 5 points, then it is remarkably ignorant to call it “even” between Reilly and Patrick.

      <

      p>
      I don’t think anyone here would do that, or disagree with you that it would be ridiculous.  But that’s just a fantasy right now.  What we have is this one poll, and yes, it does make things look pretty even.  For all that you can tell from a poll taken before most people have paid any attention to the race – which is not much.

    • was that given that there are eight whole months between now and November, I could see any of Reilly, Patrick, or Healey winning. I don’t see any lead to be prohibitive, and indeed perhaps no lead would be prohibitive at this early date. So yes, ro restate the obvious, Reilly’s ahead of Healey, and Healey’s ahead of Patrick. That’s what the poll literally says, but literally speaking, a March poll means exactly squat.

      <

      p>
      I also suspect that this may well be as high as Mihos gets. That’s a hunch — I don’t claim to be psychic.

  5. sco – yes the subsample of any poll will be small.  That’s why you shouldn’t look at subsamples at all.  You need to look at the crosstabs.  I was simply saying that your criticism of those would say “look at how Patrick is faring amongst black voters” is accurate.

    <

    p>
    But, I still stand by my criticism of this polls sampling methodology.  Why bother including nonvoters at all then?  Why call them and complete a survey and not include their responses?  Why not just dump their responses?  In my mind, it calls the methodology at least somewhat into question, and therefore raises concerns about other issues – was the geographic quotas appropriately considered?  Was their an accurate gender and party balance (the crosstabs reveal appropriate gender splits, but the party splits are way off, oversampling Republicans*)?  And it still doesn’t allay my concerns that not all registered voters are likely voters.  The purpose of polls of election contests is to mirror an election day scenario.  If you are polling people who will not be voting on election day, you cannot do that!

    <

    p>
    Even the opening to their news article is flawed:

    <

    p>
    “Our exclusive Fast Track from Survey USA surveyed 700 Massachusetts residents over the weekend, on the heels of Christy Mihos entering the race as an independent. And for now, a surprisingly-large slice of the electorate is giving Mihos the green light.”

    <

    p>
    The respondents are “residents” in one sentence and part of the “electorate” in the next.  But not all residents are voters.

    <

    p>
    Another problem – the name order in the ballot test questions do not match the name order that would be on an actual ballot in Massachusetts (which are done alphabetically).  A small issue, but an issue nonetheless, and one that runs contrary to standard industry practices.

    <

    p>
    On a side note, I concur that to say it’s “dead even” now is a bit premature.  No one is up on TV yet.  Mihos still has the chance to crack a few more of his “jokes”.  We’ve had a couple of polls that might provide us with some data points to look at a trend, but I wouldn’t consider this one one of them.  Just too many problems with it in my mind.  I hope one of the campaigns or someone in the press will ask the pollster to defend his methodology so we can hear some answers.

    <

    p>
    *Party splits in poll sample:
    GOP DEM UNEN
    20% 39% 39%
    Party splits in Massachusetts:
    GOP  DEM  UNEN
    13%  38%  49%

Archives