After my last diary about sampling in the national polls with respect to party ID, I decided to look at some state polls for the same effect. Sadly, my pool is very limited in most states with respect to gaining access to internals, and I am forced to rely on either partisan polls, or some that have not had the greatest track record in the past, like SurveyUSA and ARG.
The sidebar on Drudge as of this typing show some tightening in the daily trackers tomorrow, but factoring in today’s numbers from Hotline and Rassmussen, Obama’s national average lead from 10 polls between 9/29 and today from which I gained access to the internals stands at 6%. The Democrats sample advantage in those polls over Republicans is 7.9%. I’m not suggesting they should be even, but closer to past figures of a 3-4% advantage for the Dems would be more representative, I feel, of the populous. Even Gallup put out numbers today which downplay expectations for the young voter and minority turnout for Obama, making it more reflective of past voting populations.
I selected 9/29 as a point because it is in the middle of the bailout mess, which I think changed the whole nature of this race. I hesitate to go too far back in state polls as I’d run into convention bounces and such, which may put more noise into the data. I call this writing ”preliminary” because I want to check out more than the 2000, 2004, and 2006 exit polls, and I fully anticipate to continue to add more polls to these figures as election day approaches. But I did want to pass along some of the anomalies I’m finding already.
In Florida, 4 polls show Obama has an average lead of 2.5%, but the Democrats have a sample representation advantage over the GOP of 3.8%. This is significant because in the 2006 and 2004 elections respectively, exit polls show a 3 and 4 point advantage to the GOP. The 2004 exit poll also showed a 2.7% lead for Kerry in a state he lost by 5 points, despite the 4 point GOP advantage. So we begin to see the noise we know all so well with exit polls reflected in these figures, but the 2008 polls are still reflecting a 7-8 point swing to Democrats in Florida if we are to believe these polls are an accurate reflection of these electorates.
In North Carolina, 4 polls average to a 1.3% lead for Obama, and a 10.8% advantage for the Democrats. In 2004, the GOP had a 1% turnout advantage in the exit polls, which showed a 6% lead for Bush (he won by 12%). In 2000, the exit polls show a 12.6% turnout advantage for the Dems, but also showed Gore with a 3.3% lead, in a state he lost by 13%. Perhaps pollsters are looking to the 2000 election exit polls as a turnout guide for their polls? Either way, the change between 2004 and today represents almost a 12% point shift towards Democrats in a state with a competitive Senate and Governor’s race. While those other races may help pull more Dems to the polls, a shift that large smells fishy to me, as I have great familiarity with NC’s voter population. I will have to look back at past elections to see if the exit polls look more like the 2000 electorate (which was still very flawed) or the 2004 turnout. I expect the Dems to have an advantage, but not in double digits.
Ohio: 5 polls average to a 4% Obama lead, and 5.4% Dem sample advantage. In 2006, the Dems, riding a wave of corruption from state Republicans, did see a turnout advantage of 3% in exit polls. So perhaps that will follow suit this year? In 2004, the GOP had the advantage, however, of 5%. In 2000, the advantage was 4% for the Dems. That exit poll showed Bush ahead by 1% in a state he won by 4%. So I’m not sure what to make of Ohio, particularly with the voter registration scandal that is going on now. I’ll have to punt on it until more polls become available. These ID shifts between elections, however, may suggest that movement of 7-8 points is possible in some states, depending on GOTV efforts (if you believe the exit polls).
Virginia: 4 Polls show an Obama lead, and Democratic advantage at an identical 6.8%. In 2006, the GOP advantage in exit polls was 3%, and in 2004 it was 4%. So again we are seeing approximately a 10% shift in the electorate towards the Democrats if the polls are accurate. The 2000 exit polls again showed a Dem advantage of 4.2%, and Gore ahead by 0.6% over Bush – but the real result was a Bush 7% win, so again there is a noise problem, but they are the only thing I’ve found so far close to the current poll representation. Like in Florida, the 2004 exits showed a 5.9% Kerry lead in a state he lost by 8%, and despite the GOP turnout advantage.
So again, take this with a bit of a grain of salt for now. I have several other battleground states in the chart I’ve created, but they all have 2 or 1 poll with available internals in my 9/29-now time frame. I will update as more polls become available.
Overall, if the numbers hold as they are going forward with the inclusion of more polling data, I have great reservations about the 7-10%+ shifts in the partisan ID of the electorate as illustrated by these polls. Like I said before, I’ll buy some shift towards the Dems, but for the population to shift this much would be a realignment of almost historical proportions, which I am just not seeing in today’s public given that such a change would be an acceptance of Democratic policies which trend towards socialist. Current Congressional approval ratings, and the bailout backlash simply do not suggest acceptance for such a change. I also have a lot of concern about comparing these figures to past exit polls, given their inherent reporting problems, but they are the best thing I can think of to use for past elections.
Also, if any of you have access to premium poll data, or can recommend some sites with state polls, I’d love to know, as this has really caught my attention.