Monday, November 10, 2008

More on the Georgia Senate Runoff

Late last week, we examined the Georgia counties where both Saxby Chambliss and Jim Martin did well in Tuesday's Senate race election. That investigation provided an idea as to where each of the candidates may be focusing between now and the December 2 runoff and where either may be vulnerable. But let's look at some of the other numbers from last Tuesday's vote and see if there are any other clues to whether Martin actually has a chance or if Chambliss is something of a foregone conclusion.

2008 Presidential Vote: Georgia

52.3% 47.0% 0.7% 100.0%
*Source: Georgia Secretary of State

One thing that I was interested in seeing was the amount of ballot roll-off* from the presidential race to the Senate race. In other words, who voted for president and then just skipped voting for the Senate race and/or all the other down-ballot races? The idea here is that if there were a significant number of Obama voters that didn't vote for Martin, then he 1) is already starting with a smaller base or 2) has some other potential voters to mobilize for the runoff election.

2008 Senate Vote: Georgia

49.8% 46.8% 3.4% 100.0%
*Source: Georgia Secretary of State

What we see is that overall there were approximately 169,000 fewer votes cast in the Senate race than in the presidential race. But how we get to that 169,000 figure is an interesting sidenote. There were about 182,000 fewer Chambliss voters than McCain voters and 86,000 fewer Martin voters than Obama supporters. In the best case scenario for Martin, if he was able to mobilize all those Obama voters behind him, the former state Senator representative would gain almost 100,000 votes on Chambliss. That would bring the margin between the two candidates down to less than 25,000 votes. That's certainly closer, but not close enough to make Chambliss really sweat it out.

There are three wildcards here, though. First, we know that turnout is likely to be far lower on December 2 than it was on November 4. The above is a better (not best) case scenario for Jim Martin. But we know that all those voters aren't going to come back to the polls for the runoff. That means that attempting to handicap that turnout will help us to better understand how competitive the runoff will actually be. The other two wildcards will help us there.

What about those Libertarians? Native son and Libertarian presidential candidate, Bob Barr, wasn't able to sway all that many voters over into his corner. Barr was seen as a major potential factor in bringing Georgia into play for Obama, but after polling well over the summer, Barr's support wavered in the polls down the stretch. In fact, the Libertarian's senate candidate, Allen Buckley, polled much better on November 4, totalling about 100,000 more votes than Barr did in the presidential race. That can be chalked up to strategic voting. Buckley, slim though his chances were, had a better chance of winning that Senate seat than Barr did of winning the presidency.

For comparison's sake, I looked at North Carolina as well. Here's a state that had a competitive Senate race as well and has a pretty good base of Libertarian support. The outcome was very similar in the Tar Heel state. McCain and Obama crowded Bob Barr out in the presidential race, but Libertarian, Chris Cole did comparatively better in the Dole-Hagan Senate race. The roll-off in North Carolina was about a third of what it was in Georgia, but the Libertarians did around 100,000 votes better on the Senate level in both states' Senate races.

Well, what does all that mean? For starters, there are likely a sizable number of people who voted for McCain and Buckley who Chambliss could target in some way. I would imagine some reprise of the McCain campaign's socialism, big government, big spending liberal arguments could persuade some of those voters to participate in the runoff and vote for Chambliss.

The other wildcard will affect the turnout Jim Martin is likely to expect. And we talked about this one at the close of the previous post on this race. Very simply, how involved will President-elect Barack Obama be. Are 58-59 seats in the Senate better than 57, or does it even matter since 60 seats are basically off the table for the Democrats. If the 44th president feels like getting involved, Georgia is a place where there's the most potential for impact. Counting and recounting in Alaska and Minnesota, respectively, aren't arenas where Obama can make all that much difference. In a campaign, a newly elected president could make a difference.

And as we saw, there were a fair number of Obama voters who dropped off after that vote and didn't cast a vote in the Senate race. If the primary/general election campaign infrastructure that Obama had in place can be recharged to some degree, Jim Martin figures to be the beneficiary in the December 2 runoff.

And what about those McCain voters? Chambliss could activate some of those, right? Sure, I just think it is less likely than Obama voters who rolled off turning out for the runoff given the enthusiasm gap that was present before the election and the actual cumulative results from November 4. In any event, we aren't talking about a ton of voters here, but in a close race, those voters at the margins could prove consequential.

In the end, we can add a few more parts to the equation. We already knew about the likely turnout drop between the general election and the runoff, but now we can factor in -- at least in our thinking if not statistically in some way -- the possible influence those Libertarian and Obama voters from the original vote.

*One other cause for this ballot roll-off might have been the negative tone of the Senate campaign over the final weeks. There is some research to suggest that this has happened before. It could also be argued that the presidential race was also negative, but it isn't a stretch to say that it was outpaced in negativity by the Chambliss-Martin Senate race. Also, though Georgia got some last minute attention from the presidential candidates, neither campaign was focused too heavily on the Peach state despite the shrinking polling gap during the final week. That wasn't the case in the Senate race.

Recent Posts:
Omaha to Obama

A Slideshow Chronology of the Electoral College on Election Night

Frontloading and The Rules in 2008: The Maps


Unknown said...

I don't understand this:

'There were about 182,000 fewer Chambliss voters than McCain voters and 86,000 fewer Martin voters than Obama supporters. If both Chambliss and Martin were able to mobilize all their parties' presidential candidate voters behind them, Martin would gain almost 100,000 votes on Chambliss, bringing the margin between the two candidates down to less than 15,000 votes."

If they both mobilize all their parties' presidential votes, doesn't that mean the spread would widen by 100,000 votes, making it a Chambliss blow-out?

Anonymous said...

Good catch Scott. That Chambliss bit wasn't supposed to be in there.

Anonymous said...

The simple answer is yes, it would, Scott. I was playing around with several scenarios and mangled them in the post.

An administrative note: I couldn't strike that material properly without completely destroying the readability of that sentence. It has been corrected but the quotation Scott included in his comment was in the original post.

Thanks Scott.

Jack said...


Your fix, unfortunately, added another error. You now refer to Martin as "the former state senator." Martin was a member of the state House, not the Senate.

Anonymous said...

I'm on a roll today. I'm going to chalk this post up to post-election withdrawals. lieu of deleting it altogether.

FHQ's diligent contributors always have my back though.

Thanks y'all.

Tomorrow will be better. To mark the one week anniversary of the 2008 election, I'll finally have a wrap-up post up; complete with map and electoral college spectrum. And I have another interesting post that I think we can all have some fun with. I've got a few things I need to check on that one though, so I don't repeat this post's delightful miscues.

Anonymous said...

Hi - I just stumbled into your blog; very useful, cogent analysis. Thanks. Hopefully the work of maintaining a blog will strengthen your dissertation and your career.

Two comments:

[1] your earlier post on the Chamblis/Martin runoff does useful county-levle analysis, but doesn't discuss just how depopulate some Ga. counties are: Atlanta ex-urb counties have more than 100x more voters than some of the rural SE counties - what happens there won't matter.

[2] In this post, you don't discuss cross-over votes: the ticket-splitting voters who vote R for Pres and D for Senate and vice-versa. Yes, as a practical matter there probably aren't that many of them. But in a close runoff the behavior of ticket-splitters could be determinative, esp. if there are asymmetries in how they break (so that they don't just cancel each other out).

Did Ga. exit polls reveal a significant # of ticket-splitters?

Anonymous said...

Hey Hexagonal.

Thanks for stopping by and thanks for the kind words.

I think you're absolutely right on the points you've made. Those counties aren't where the votes are for the most part. Henry County to the southeast of Atlanta is the exception. This election isn't like the electoral college and we certainly aren't dealing with the county unit system that dominated Georgia in the time prior to Baker v. Carr and Westberry v. Sanders. And as I said, turnout is the key to the runoff. That's where those wildcards come into play.

This post and its predecessor were jumping off points more or less. I'll hopefully have time to augment them some between now and December 2. I'll look into your exit polling question. I don't know right off hand.