I began following the results of the Georgia election about about 2:00 am election night because I, like many here, expected the results to be closer than they were. I thought the low numbers were suspicious, and you can see my diary here.
However, as the days have moved on, I've done some research which to me suggests that the numbers in Georgia are not an anomaly, but rather conform fairly closely to what one would expect.
To be clear: SoS has not been nearly as forthcoming as we would like and there are votes left to be counted. But the body of evidence suggests neither a huge discrepancy from predictions/efforts nor a case of a fraudulent or stolen election. We should keep our eyes on it to be sure, but the data just don't shout 'Fraud!'
Follow me below the jump for my take on Georgia.
First, a little background: I am not an expert, but I do have experience in election fraud analysis. I've written several papers and studied several election (from presidential to congressional) looking for signs of fraud. Obviously it's easier and more robust with access to machine error logs and local reporting, but looking at the topline we can get a sense for whether or not we should look more deeply into the results.
How 'off' are the 2008 returns from Georgia?
Let's start by comparing presidential votes to get a sense of turnout.
- R 1,914,254 D 1,366,149
- R 2,019,740 D 1,803,920
3,280,403 ballots in 2004 and 3,823,660 in 2008 with some in Fulton county left to be counted (see below).
So, the Republican candidate had a 5.5% increase in total votes from 2004 - which suggests decent but not extreme turnout and likely reflects the growing population of Deomcrats.
The Democratic candidate, on the other hand, had a 32.04% increase in votes from 2004. That is absolutely HUGE, and while I don't have the time to compare the increase to other states, I do intend to.
Total votes were up 16.5%, and as we saw above the majority of the increase broke for Obama in a huge way.
And please consider that a 16% increase in turnout is ENORMOUS considering national turnout looks all but flat - perhaps up closer to 5% by the time all votes have been tallied.
Why should we expect better? Obama had a great ground game and was a groundbreaking candidate, and as a result current estimates peg him at getting 437,771 more votes in the state than Kerry did. That is an enormous increase - particularly considering Obama pulled out of Georgia for a time.
What about the early voting? 2.02 million early votes would be roughly 52.89% of the total votes cast. One again, limited time (class in 40 minutes) prohibits me from doing a more robust number crunch, but that's an epic result that falls within reason.
So what did pollster think would happen?
Pollster: M 49.5, O 46.6
Actual: M 52.4, O 46.8
Pollster: C: 47.4 M: 43.3
Actual: C: C 49.9, M 46.7
Looks reasonable to me. McCain did a bit better than expected as did Martin.
We all wanted Georgia, and we layed our hopes squarly on the pollsters being off. Well, it turns out they weren't - their attempts to correct for early voting seemed close to spot on as an aggregate, which is what we saw around the country - that's how Nate managed to do so well on the popular vote and on projecting states and senate seats.
Speaking of which, what did Nate think about Georgia walking into it?
Senate: Chambliss 88% of victory
Nate's final words:
Georgia is a little trickier. Although all polling continues to show Jim Martin a few points behind Saxby Chambliss, Martin can earn a run-off by holding Chambliss under an outright majority of Georgia's votes. I would assign a 50 percent chance to a Chambliss win outright, a 40 percent chance to a run-off, and a 10 percent chance to a Martin win outright.
Finally, the votes to be counted. Reports suggest there may be several thousand votes left to count for 2008. Georgia's Secretary of State website suggests 96% of precincts have reported, and several articles suggest a number of absentee or early votes left to count. See the AJC article here for more info on votes left to count.
Fulton County elections officials may have violated state law when they sent their weary poll workers home early Wednesday morning before they had finished counting 30,000 absentee ballots.
Fulton is the only county with numerous ballots outstanding.
Several thousand votes will incrementally increase turnout, but they won't change much beyond that - which is what people expected walking in and what appears to have happened on the ground.
I want this as bad as the rest of you - I was up until my eyes hurt Tuesday night/Weds morning trying to figure out what was going on. But after the moment I posted my diary (when the vote totals suggested flat turnout of ~3.3 million in '04 and '08) an additional 500,000 ballots were closed, and indeed the race tightened. But the 5-6 point Georgia pres race we're seeing right now and the slightly tighter senate race conform with pre-election polling, and still represent hefty turnout all around, especially amongst African Americans.
Georgia was a victory - we showed how much change we could make in just 4 years and how close we could get. In VA and NC those efforts put us past the goal post, in the deeper and redder GA that got us close but not over.
Elections can be dirty, and elections have been stolen. All overt signs DO NOT point to Georgia in 2008 being such a case.
Additonal info from comments and after posting:
people in GA reported all day on Tuesday that election day turnout was substantially down - a lot less than expected. You should also expect a large number of white kerry voters in rural GA just would not vote for Obama - not unlike what we see in the other parts of Appalachia