Wednesday, November 19, 2008

Georgia Senate Runoff: Early Voting (Day 2)

Yesterday, we got our first glimpse at the early voting statistics for the Georgia Senate runoff. We also raised the question over the potential impact the staggered starts to early voting in the 159 counties across Georgia might have for the fortunes of the two candidates involved. First, let's incorporate the new data from yesterday and then I'll make some further comments on the staggered early voting starts.

[Note: I should add that early voting began in Fulton County on Tuesday and not on Wednesday as I said yesterday. In other words, there was only a one day lag for one of the most Martin-friendly counties in the Peach state. There were conflicting reports on the Fulton situation, and I opted for the AJC report, thinking it would be the most reliable. I was wrong. The Secretary of State still has a list of when early voting will take place and in which counties. But I'll have more on that in a moment.]

So, what's new?
  • With Fulton added to the mix on Tuesday, the African American proportion of early voters increased, but only by a couple of percentage points. But at just under a quarter of the early votes, the African American proportion still lags almost ten points behind where it did over the course of the entire general election early voting period.
  • Once the second day numbers are added, we also see that the female percentage of early voters increases by a modest amount. Like the African American vote, though, the percentage of women voters lags behind the overall proportion during the general election early voting.
  • Also be sure to note that the spreadsheet now has tabs at the bottom for additional statistics on each of the individual days of early voting thus far.
The overarching message? Well, for starters, we may not want to read too much into the early voting numbers in a runoff election. [But you know I'll read into them, don't you?] Both of the demographic breakdowns above don't particularly favor Jim Martin's in his quest to reverse the general electorate's candidate preference ordering in the runoff. But let's look at the when each of Georgia's 159 counties actually started (or will start) early voting and see if the late starting counties favor one candidate over the other. First, how did the general election look on the county level? The typical red for Republicans, blue for Democrats color scheme applies.

[Click Map to Enlarge]

Now, let's remove the counties that started early voting on Monday. [I would say on or before Monday, but the two counties -- Gordon and Richmond -- that began early voting last Friday are shaded in dark gray below. However, neither is included in the analysis that follows. The same is true of the two counties -- Towns and Treutlen -- on which the the secretary of state's office still has no early voting information. They are shaded in pink on the map below. Finally, the two counties in light bluish gray -- Madison and McIntosh -- are counties where there is no early voting. Advance voting will start on Monday in both areas. Both Madison and McIntosh are included in our analysis.]

[Click Map to Enlarge]

What can be gleaned from that map? At first glance, not much. Just eyeballing the red and blue counties -- those counties that started early voting after Monday -- doesn't really turn up any noticeable trends. Once we combine the information from both maps, though, we start to get a clearer picture of whether one of the candidates benefits from the staggered starts to early voting. On November 4, Saxby Chambliss won 121 of Georgia's 159 counties. Of those 121, 45 counties started their early voting operations after November 17. Jim Martin, on the other hand, won 38 counties on election day and 18 of those had late starts to early voting.

Yeah, but Georgia isn't under the county unit system anymore, is it? So the county counting is irrelevant. That's true, so let's look at the proportion of the November 4 vote total from each of those collections of counties. I've removed Allen Buckley's vote total, and as a result, I'm just looking at the share of the two-party vote each candidate's late-start early voting counties comprises. [Yeah, I agree. "Late-start early voting counties" is confusing.] Basically, the result is a wash. The 45 Chambliss counties that began (or will begin) early voting after Monday made up 12.3% of the two-party vote on November 4, while the 18 Martin counties (including Fulton) that got off to a late start in early voting accounted for 11.77% of the total two-party vote during the general election. If anything, then, Chambliss is at an ever so slight disadvantage due to the staggered start to early voting in counties across the Peach state. Again though, it isn't by much. And with the proportion of early African American and female voters down from where they were during the general election's early voting period, it may not matter anyway. That drop likely outweighs any disadvantage Chambliss may be getting because counties he won on November 4 are off to a late start in early voting.

Let me note again today that I'll be updating this data daily and that I'm still trying to track down the day-by-day data for the general election's early voting period. That information will help to a more informed projection of the runoff vote. Also, tomorrow I'll add in a map to reflect the different start dates for each county's early voting windows.

Recent Posts:
Georgia Senate Runoff: Early Voting (Day 1)

Hillary Clinton vs. John McCain

The Ads: Georgia Senate Runoff--UPDATED


Unknown said...

It seems to me that there are a couple of common motives to vote before an election:

--It's hard for you to vote on election day for some specific reason

--You anticipate crazy lines and snafus on election day

I'd guess that the second group has a disproportionate number of African-Americans, in part because precincts where they are the majority tend to be less well staffed and have older equipment, and in part because of a distrust of the process.

But for a run-off, there's less of a chance of really long lines on election day, so the second reason becomes less important.

That could be an explanation for the drop in African-American participation before election day.

Jack said...


There's no simple way to figure out the percentage of the voting population that was African American in the general election in the counties that have already started voting in the runoff, right? That way we could see the precise dropoff.


I hope you're right.

Anonymous said...

I've got some good leads to check out in the morning, so hopefully I'll be able to track that data down. If not county-by-county, the day-by-day totals will help us to draw some (admittedly imperfect) comparisons between runoff early voting and general election early voting.