New AJC Poll Released

The AJC has released their exclusive poll conducted by Abt SRBI that shows Governor Nathan Deal with a slight lead of 46-41 over Carter, with 5% going to Andrew Hunt. It also shows Perdue with a lead of 44-42, with 6% going to Amanda Swafford. Perdue’s lead is within the margin of error of 3.6%, though.

The full crosstabs are available, but a quick synopsis shows:

  • 1,170 Georgians polled via cell phones and landlines from October 16-23
  • African-Americans – 29 percent
  • Females – A tick under 50 percent


  1. So here’s my problem with this poll. I’m going to assume that their black likely voters are more female than male. In actuality the gender numbers for blacks are nearly 2:1.

    So – let’s assume that out of the 29% you have something like 19 black women, 10 black men.
    In order to get to MAJORITY male (something that has never happened before, btw) you would need your non-black gender split to be something like 41 men – 30 women. In other words, your non-black/white sample would be something like 58% male.

    Now, in years past midterm elections (even in 2010) for the white vote were about 50-50 men and women. So, you take the single best demographic for Republicans, white men, and you have them over-represented heavily and you still can’t get the Republicans above 44 and 46.

      • Thanks – my point is still valid at 55%. The problem is them using some sort of intensity screen for determining who a likely voter is. So when they talk to the white guy who hates Obama so much that he can’t wait to tell you about it, everyone of those guys ends up in the sample. When they talk to the suburban white mom who has other things on her mind (but will vote and is much more likely to vote for Nunn/Carter than the aforementioned white guy) she doesn’t make it through the screen at all/as often.

        When you normally poll for Olive Garden or whatever, an intensity screen (how likely are you to eat at Olive Garden, do you remember the last time you ate at Olive Garden, do you know where your neighborhood Olive Garden is) makes a lot of sense. You’re trying to take a population of 100% and really narrow down to the 5% that are most likely to be customers.

        When you talk about voting, something that 50% of all registered voters will end up doing, it’s going to be much more random as far as who makes it and who doesn’t. And while the aformentioned Obama voter is going to be in the electorate, you can sure of that, there will be plenty of people who aren’t as passionate who will also show up. A screen that filters out only the most likeliest of voters doesn’t account for the many voters who will happen to vote but might not. But there are hundreds of thousands of these people – and crucially – they are different in their outlook and opinions then the guy who makes it through the screen.

        We’ve already established that the electorate has NEVER been majority male, this isn’t even up for debate the way the black vote is. You can see how this could lead to serious problems in the poll. Just ask President Romney.

    • Georgia has never in modern history (20 years) seen a general election electorate with the female vote being lower than 54% of the total vote.

      Females continue to give large percentage of their vote to the Democratic candidates, while men continue to give large percentages of their vote to Republicans.

      Therefore, when you weight the female vote to a proper 54%, the poll numbers fall.

      I don’t understand why so many polling firms are just not getting this right. It’s not … hard to understand.

      • I don’t even think you can do a simple reweighting, because they are using some sort of intensity scale that is skewing who gets in, which is obvious for the fact that it overloads males.

        So if you take like all white males and white females, it’s cutting off way too high (I would guess) on the conservative/Republican side (or else how would so many white men make it in?).

        So their likely voter screen clearly has a conservative bias, so even when you reweight for men and women, the base male and female numbers don’t include enough Democratic voters.

      • Bobloblaw says:

        I agree too male. But if females in the sample have an extraordinary support for the Dems, it balances out. It isn’t just the % of each gender but the % each candidate gets from each gender.

  2. Daniel N. Adams says:

    Since on election day the “undecideds” won’t have a percentage, here is my prognostication on both Gubner’s and Senitahr’s race… R 51% D 44% L 5% MoE 2.5% -(no Common Core maths used).

    • Bobloblaw says:

      I think that is too optimistic. I’ve been in the runoff camp. Now I think Nunn has a 35% chance of winning outright. I’ll have to compare her numbers to Saxby’s in 2008 when he got 49.6%. I still think Govs is 95% runoff.

  3. Salmo says:

    I’m still having a hard time seeing the libertarians pulling numbers that high, and those votes will break at least 2:1 to the GOP if they go elsewhere. 2-3% is the more likely ceiling.

    • Will Durant says:

      The Governor’s race will see quite a few former ruby red Republicans I know protesting with a vote for the Libertarian. Carter is just coming across as too green (in political experience, not environmental). Deal is just coming across as too orange as in the new black.

      • Bobloblaw says:

        I’ve wondered how well the libertarians will do without boortz around on a daily basis. Actually historians will pin point the moment GA went to the toilet as when Herman Cain took over for Boortz.

          • Trey A. says:

            My idiot brother early voted and went Libertarian on both top ticket races–not because he’s a libertarian. In fact, he thinks Libertarians are pretty awful, but he knows they’re not going to win either race and he says he’s expressing his displeasure at both national parties.

            So there you have one Libertarian voter who is not an “extremist.” Surely he’s not the only one.

Comments are closed.