Spin Cycle

So I think we can all acknowledge that someone in my position would rather not have to write a post like this to begin with, but here it goes.  Survey USA got a lot of buzz yesterday releasing a poll that showed Deal up 11 points.  Is it really that bad for Democrats?  Well only if you believe a couple of things in this poll that I think are unlikely:

  • Men will make up 54% of the electorate this year.  Women have historically made up about 54% of the electorate.  Women are split, men favor Deal in the poll by 20%
  • Under 35’s will make up 22% of the electorate.  Um, not likely.  In 2006 they made up about 15% of the electorate.  This is Deal’s best age bracket (which also seems fishy) as he leads by 20% here in this poll
  • 57% of the electorate makes more than $50k.  Maybe they will.  The median household income for Georgia in 2008 was $51k in 2008 according to the census department.  That means that the median individual income is a lot lower as there are a lot of dual earner households, plus most younger and retired people definitely aren’t pulling in that much.  So, Survey USA sees a much wealthier electorate than the state as a whole.  Again, it’s not hard to imagine that voters who vote this year are probably more well off than the state as a whole – but 57% making more than the median household makes?  Deal has a 27 point advantage among these voters.

So that’s my basic case to fellow Democrats to take this poll with a grain of salt.  Three groups that heavily favor Deal that seem to be overrepresented in this survey.  Maybe it’s because they called on Friday and Saturday and a bunch of men who were pissed and angry that Georgia, Georgia Tech and Georgia State were losing answered the phone.  The other potential problem with this poll is that they’ve got the African American share of the electorate at only 23%.  It was 30% in 2008, now nobody I know thinks it will be that high this year, but in 2004 and 2006 it was greater than 24% each year and since then African American new registers outnumbers whites by about a 5:2 ratio.

I don’t generally traffic in conspiracy theories about cell phone only voters.  Research has shown that they don’t deviate too much from their peers who have landlines, and Survey USA has done some of that research and shown that when you include them they haven’t moved the poll results very much.  But, and here’s the big but for this state – African Americans are disproportionately likely to be in cell phone only households.  According to research they use nearly twice as many cell minutes as their white counterparts.  Barnes is doing about 60 points better with African Americans than whites, and I believe even though non-cell phone polls have only minor discrepancies in heavily white states with polls that include cell phones, in this special case where the demo that is cell phone only is so different from that that isn’t, you could be accounting for a couple of points change on each candidate here.

I think when you add it (and subtract it) all up, this poll really comes out to about 46 for Deal and 42 for Barnes.  In other words, it’s anyone’s ball game and no one at this point is assured of winning without a runoff.  Of course, on November 2nd we’ll get the only poll results that matter.  Until then I’d work my butt off no matter what side I was on.


  1. bowersville says:

    A friendly reminder previously posted in another thread.

    SurveyUSA conducted a poll on 7/7/10 – 7/8/10 for the July 20th primary. The results were:


    We all know how that turned out.

  2. You raise some good points Chris. All polls should be taken with a grain of salt since you never really know who is going to show up on election day (or in the 45 day voting marathon prior to election day).

    Where did you see that under 35’s were Deal’s best demographic? I thought it was over 65’s which support him at 53%.

    There are things in this poll which could indicate the race is tighter. I’m concerned no non-incumbent Republican is over 50% and I’m concerned the Libertarians are polling so well. However, I’m encouraged that Barnes has been blasting Deal for a month now and hasn’t really increased his support, nor harmed Deal all that much (if at all).

    • I was looking more at the margin of the lead for Deal. He’s doing well among over 65’s, but I think that’s to be expected. He’ll need more than 53% from these folks and he’s leading by 17% here.

      But under 35’s are a curious bunch, while Deal only has 49% of the vote, he leads Barnes by 19% here. And there’s NO undecided, they’ve got Monds at 21% getting everyone else.

      Just assume for a second that instead of Deal leading by 19% here and Monds getting a similar share that he and Roy are tied at 40%. That takes 2 points off Deal’s total and adds 2 to Roy’s right there making it 47-40. I don’t think that’s far fetched to assume and I think when you start at 47-40 and weight up the Black number a bit all of a sudden you’re at…45-42.

    • ACCmoderate says:

      Buzz, I don’t think you’re going to see the negative ads create forward momentum for Barnes’ campaign.

      The goal is to set up a narrative of Deal as corrupt, corrupt, corrupt. As the election gets closer, thats when you bust out the positive ads about education, taxes, etc.

      I’m surprised that as a Republican with high name-ID in a deep red state, Deal isn’t leading by more. Maybe Barnes’ ads have accomplished something after all.

  3. fishtail says:

    The SurveyUSA poll was pre-determined from the get-go. The pollster assumes that there will be a big turnout of white voters and that minority voters won’t bother to vote. With such a skewed approach you will get any Republican beating any Democrat…but maybe that’s what SurveyUSA wants as an outcome.

    • Then 11Alive and especially V-103 should really complain since they paid for the poll. I would think 11Alive would want an accurate poll and I doubt V-103 would want African-Americans under sampled intentionally.

  4. Andre says:

    Based on numbers from the Georgia Secretary of State Elections Division, the average percentage of black voters in this state’s electorate for the past 12 years is 24.2.

    The SurveyUSA model has the black electorate at 23%.

    The arguments that the SurveyUSA model has the black percentage of likely voters “way too low” just isn’t holding water in the face of twelve years worth of facts derived from numbers certified by mostly Democratic Secretaries of State.

        • Oh I totally acknowledge that. But ’02 was 22%, ’06 was 24%, I’d expect ’10 to be about 26%.

          While we are quibbling over relatively small numbers, remember that this is a voter group that Barnes will get close to 90% with, vs around 35% with everyone else. So each percentage points narrows the margin by like 1.3% (55% less for one and 55% more for the other). So a 3% difference turns a 11 point lead into a 7 point lead etc.

  5. joe says:

    “Maybe it’s because they called on Friday and Saturday and a bunch of men who were pissed and angry that Georgia, Georgia Tech and Georgia State were losing answered the phone.”

    So your analysis is that the state of Georgia football is perceived as being Barnes’ fault.

    I guess that makes as much sense as anything else.

  6. Andre says:

    There are some questions to be answered about this SurveyUSA poll; particularly how their model of likely voters has men outnumbering women 54% to 46%.

    Historical turnout numbers from the Secretary of State Elections Division shows that over the past 12 years, women have outpaced men at the ballot box on an average of 54.53% to 44.54%.

    This major error by SurveyUSA in its polling model puts the results on shaky ground.

Comments are closed.