Wait – Romney only leads by 6?

So – Matt Towery apparently slipped in the shower and hit his head and put out a “poll” showing Romney leading by about a bazillion points, and also that only 5% of Georgians think Obama will do a better job on the economy, and that no one wants to go to dinner with the President.  Best comment I saw was that maybe just a lot of people that responded weren’t hungry.  Now on the heels of this Fox 5 poll (and to anyone from Fox5 reading this – how can you release that garbage with your name on it?), internet pollster YouGov is out with a new poll of Georgia – one of what seems like a multi-state monthly poll that checked in on almost half of the states.

YouGov has a national panel of respondents that they randomly select and match to a stratified sample that complete their polls online.  It’s new-fangled technology to be certain, but they have a decent track record.  They weight their survey using the 2006/2007 American Community Survey results and use 2008 exit polling and turnout as a guide.

They show Romney with a 6 point lead – 50/44, but looking under the hood at their crosstabs, they also show African Americans making up just shy of 26% of the electorate.  I personally think people get a little carried away “reweighting” every poll that comes out until it suits their fancy – maybe fewer African American respondents was naturally and scientifically balanced by interviewing more white Democrats than normal – you never know and that’s why it’s dangerous to reweight to your heart’s desire and also kind of the point of the margin of error.

But I will say that in 2008, African Americans made up 30% of the actual voters (roughly the same % as registered voters), and even in 2010 – a down year for Democrats they accounted for 28% of all voters.  For 25% of the electorate you’d have to go all the way back to 2004 – when the decidedly very white John Kerry was the standard bearer.  So…if the electorate is really 4-5% more African American than YouGov’s sample and you were to recklessly reweight this survey, you’d come up with pretty much a 47-47 tie.  My firm conducted an IVR survey for Better Georgia before the conventions in August and we had among likely voters a 49-46 Romney lead, with a tie game among all registered voters.

So pick your poison – giant Romney lead (IA/Fox5), modest Romney lead (YouGov), tiny Romney lead (20/20 Insight for Better Georgia) or jump ball (YouGov re-weight/20-20 RV).  One thing is for sure – at the Presidential level this state isn’t as boringly predictable as it used to be – and could be a legit swing state contender in 2016.  Maybe let’s not get rid of that electoral college afterall!


    • wicker says:


      There is a difference between “some” and “many.” In any event, Obama’s problems with black voters are nothing compared with Romney’s problems with white ones; problems which are due to Obama’s success in convincing the electorate that Romney is another George W. Bush.

      • caroline says:

        Actually I think Romney has done a good job of convincing people that he’s the second coming George W. Bush with his policy stances.

        • wicker says:

          Which makes me wonder why the GOP nominated a moderate like Romney to begin with. Someone with a staunch conservative record can move towards the center while still retaining the base, like Reagan did in making George H. W. Bush his VP. But if a moderate makes a move towards the center, he loses the base without picking up enough independent voters to make a difference. So, were Romney to come out and say that deficit/national debt reduction is a bigger need right now than (the Bush agenda of) tax cuts and wars in the Middle East, sure he’d pick up some independent voters, but the base would stay home on election day.

          It will take a brilliant play by Romney to get out of the position that he is in. I am curious to see if he has that sort of brilliance in him.

          • caroline says:

            Well, first off all what was his opposition in the primary? Newt? ROTFLMAO! Santorum? Puhleeze. 9-9-9 Herman Cain? He was a governor which generally does better than someone in congress. So that probably was a plus on his side.

            I think the crux of the problem is that’s what popular with the GOP base is not popular with the rest of the country in general. The GOP does really well with white evangelical voters who only make up about 20 to 25% of the voters but are 2/3 of the republican primary voters as I understand it. So a small tent produces a lot of problems is long story short.

      • Three Jack says:

        wicker, you nailed it. Romney has major problems with white/fiscal conservative voters. He is out trying to shore up minority support that he has no chance of getting while ignoring the very people who he might persuade. Perfect example was his 60 Minutes interview last night where he went all-in for means testing so-called entitlements. Might as well just run as a dem, at least then he could run on his principles.

  1. saltycracker says:

    Chris – since you brought up polls, spreads, national attention and colleges:
    ‘Noles 49, Tigers 37 in a classic game.

  2. caroline says:

    I don’t doubt that Romney is going to carry GA but the poll that has Obama with 35% seems really off kilter. I mean didn’t John Kerry get about 42% of the vote back in 2004?

  3. SouthernCamelot says:

    I feel many are missing a much larger picture — that is, with Romney continuing to sink, and especially if he sinks in the debates (which I think most of us know is a likely scenario based on the fact he hasn’t been able to win on issues since training for POTUS since 2007) he could go down to embarrassing levels. I foresee a 39%-40% poll for Romney after the first debate, from a credible pollster coming out soon for Ohio, at which point “freefall” like hasn’t been seen in a generation in national politics will happen and Republicans will not know how to react to. The cat will be out of the bag. Republicans will then start the viscous cycle of giving up (out of extreme anger at the unlikable, even by the base, Mitt Romney) while Democratic enthusiasm goes up as it does with excitement (think of the Clinton speech… Democrats and Dem. leaning independents will change polls and vote totals based on a level of excitement in the air, that’s their nature). At this point certain states like North Carolina and even the Democratic base of Georgia will feel like they can compete again. Keep in mind Barack Obama picked up 47+% of the vote in 2008 compared to McCain’s 52%+. Romney is already polling at 50 and if he goes below 50 in Georgia, it spells real trouble. Georgia Democrats currently have little OFA resources available as they’ve been diverted to swing states. Some could come here to help with momentum by mid-October. The entire state candidates for the Democratic party will finally find some unity because behind the force of the upcoming “tidal wave” I’m predicting because it will allow their more vulnerable candidates to keep their seats, reverse the negative trend, and also possibly legitimize certain districts and build a blueprint for a very important 2014 election where winning the Governorship is not an unrealistic possibility by any means (and I dare any Deal supporter to challenge me on that… Independents and the Democratic base combined with a good candidate can make him the next surprise one term governor).

  4. Doug Deal says:

    That is absolutely NOT the point of margin of error. Margin of error is the smallest possible error based on a perfectly random sample of a population. It does not include systematic error and things introduced based on the gut feeling of the pollster.

    • Well to get really in the weeds, margin of error is half of the confidence interval for a confidence level of 95% – so if the margin of error is reported as +/- 4%, the maximum confidence interval at 95% for the poll would be 8% – meaning if the poll reported that some position was held by 50% of the population, there’s a cumulative 95% chance that the real value in the population as a whole is between 46 and 54 – but also that it’s most likely to actually be 50 and least likely to be 46 or 54 but still inside that 95% range.

      What I’m trying to say about re-weighting based on race or party identification or anything else can be illustrated by this example. Let’s say that 47% of Georgians are going to vote for Barack Obama, that 38% of them will call themselves Democrats when they do, and that 30% of them will be African American. These are roughly the results of the 2008 election, by the way.

      A perfectly valid poll of this electorate could show that 47% will vote for Obama, 41% will call themselves Democrats and 27% of them will be African American. That’s because all three observations (Obama vote, Democratic %, African American race) will be within the poll’s margin of error – assuming it’s 3% or higher.

      So there’s the danger with assuming that the sub-samples are perfectly observed (like what % Obama gets from Democrats and non-Democrats, or what % Obama gets from African Americans and non-African Americans) and just reweighting. In the case of the above, a Republican may say the Democratic share of the electorate is too high, so I’m going to drop that 41 to 38 and Obama will only get 45% of the vote – yes now his party ID might match what he believes the true # to be, but his Obama number will be wrong – though still within the margin of error.

      And a Democrat may look at that poll and say the African American # is too low, I’m going to raise it to 30 instead and now Obama will get 49. Again, the African American # may not be matching what he believes the true # to be, but again his Obama number will be wrong – though still within the margin of error.

      Now, I personally believe the YouGov poll is somewhat of an outlier – as they have observed the % of the electorate that will be African American (25.5%) as too low and outside the margin of error for what I believe the actual # will be (30-31%). So based on that, I’m suggesting that you add a bit to the Obama % and subtract a bit from Romney. But in general, if you’re looking to re-weight a poll based on something like African American % or party ID (the former is popular with Democrats, the latter is popular with Republicans this year) first ask yourself if the observed # that you think is wrong is within the poll’s margin of error – and if it is – you’re really playing with fire if you start re-weighting at that point.

      • To illustrate the dangers of re-weighting based on observed variables in a poll, take three theoretical polls of the 2008 general election, all with a margin of error of +/- 4%, here is what they observed for Obama %, Democratic % and African American %.

        Obama – 47, Dem – 38, AA – 30

        #1 – O 47, D 41, A 27
        #2 – O 45, D 38, A 30
        #3 – O 49, D 35, A 31

        All three of these polls have observations that are within the margin of error of what the actual were – yet if you re-weight some of these observations to “match exactly” what the electorate will look like – forgetting that margin of error bakes in uncertainty in other questions that isn’t strictly related to any one observation, you could go astray. In the case of #1 – both potential re-weight corrects will make the Obama observation less accurate since it happened to be dead on while the other two were not – but within the margin of error. #2 is an example (all of these are theoretical) of a survey where the Democratic % and the African American % could both be dead on but the Obama # could be too low (yet within the margin of error). #3 – the Democratic # is “too low” so if you weighted that up you would increase the error that already exists on Obama’s % possibly even outside the margin of error for the poll, while the AA # is too high, so this is a case where you’d probably get a more accurate Obama # by re-weighting just the AA # but then your D number would be either on the edge or outside the margin of error.

        The YouGov poll uses a stratified sample – which in theory at least should produce more accurate results because they’re weighting in advance and then finding appropriate numbers of respondents for all of their demographic groups – and in practice firms that employ this type of method (Gerber) are popular with candidates because they typically do good work. However, my beef with YouGov is that they’re using old data when it comes to Georgia, which is quickly trending more African American and less white – and particularly at the voting booth.

        • SallyForth says:

          Whew! Chris, you said a mouthful! I love it when you get out there in the weeds, you numbers wonk you. 🙂

  5. Harry says:

    The thing that gets me – on the national level – every election the media says that the GOP candidate makes all these “gaffes” but what stupid things the Democrat does are never reported. And we know that B.H. Obama has had many stupid, embarrassing, and misleading statements.

    The debates should help at least partially remove the media filter.

  6. Harry says:

    Romney – Ryan main talking point to young people:

    Many more people are now making a near-min-wage income. Prices are jumping tremendously for food, for gas, for whatever. The irony is that money is becoming more and more WORTHLESS yet there are few new jobs that pay much more than min wage…if at all!!

    We have price inflation, high unemployment, low wages and low savings rates.

  7. SmyrnaModerate says:

    Given McCain won Georgia by only 5.2%, Romney would be ecstatic to carry the state by 6%. Four more years of demographic changes offsets decreased enthusiasm on the democratic side. Romney’s margin of victory here will probably be 4.5-5%.

    • It’s a good point but to be honest – I don’t see that much decreased enthusiasm on my side if anything whatever amount that exists is offset by decreased enthusiasm about Romney among Republicans – which may be your point!

    • SallyForth says:

      I think we all have to be realistic about the ’08 race – after all, it WAS McCain (the guy that GWB knifed in the back in 2000 and still became his lapdog) and his albatross Palin that the Republicans trotted out. Pretty much anybody with a “D” could have beat that duo.

      This time Governor Romney brings governance experience as CEO of Massachusetts to the table, international accomplishment and responsibility with the ’02 Olympics when our country was reeling from Sept. 11, 2001, and good experience in the business world (although the only thing we hear in the media is accusing him of everything that ever went wrong with Bain Capital, never any of the good things accomplished such as saving AMC Entertainment, Burger King, Burlington Coat Factory, Clear Channel Communications, Domino’s Pizza, Aspen Education Group, DoubleClick, Dunkin’ Donuts, D&M Holdings, Guitar Center, Hospital Corporation of America (HCA), Sealy, The Sports Authority, Staples, Michael’s, Warner Music Group, The Weather Channel, et al.) Omigod, I caught Chris’s off-in-the-weeds disease! 🙂 But you get the point – like all of us, Romney isn’t perfect but he’s a decent well-qualified candidate.

      Today’s Mona Charen column, “Romney’s gaffes linger, but Obama’s are ignored” is on target. Even with all the international tumult and some of the Obama gaffes she pointed out, the national press has still devoted a week to the stupid Romney out-of-context video clip illicitly smuggled from a private dinner with contributors last spring. That is unethical and probably illegal to have done, but James Carter made a mighty fine buck off it and the gotcha media.

      Where is the media’s investigative journalism these days on things that matter, such as Charen calls for – an accurate account of Obama’s gutting of Clinton’s welfare reform, the tax increases on the middle class hidden in Obamacare, and the total absence of an Obama plan to revive the economy in a second term? They could do us voters some real public service by shedding light on those and other important things prior to the upcoming election. Oh yeah, they’re too busy knifing the Romneys, while gushing over Michelle’s green dress on “The View” yesterday, and how cool Barack was singing an Al Green snippet…..

      I like Obama, and by doing some personal digging, I also see a lot to like in Romney. We have two good men from which to choose this time, but you sure couldn’t tell that from the most lop-sided media coverage we’re seeing.

Comments are closed.