You may recall that right before the general election in November, I put together benchmarks that selected statewide candidates would need to hit, on a county-by-county basis, in order to get over 50% in closely-fought states. I wanted to go back and see how well this measure worked; Georgia seemed like an apt place to start, not just because it was very close at both the presidential and senatorial levels, but also because a bit of troubleshooting is in order to see what happened with the steep dropoff in the senate runoff. Let’s start with the original table, which contains the 2008 benchmarks (and the 2004 Kerry/Bush numbers, on which they were based:
County | % of 2004 statewide vote |
What we need to break 50% statewide |
2004 Pres. |
---|---|---|---|
Statewide | 100.0 | 50/49 | 41/58 |
Fulton | 10.2 | 68/31 | 59/40 |
Cobb | 8.5 | 46/53 | 37/62 |
DeKalb | 8.4 | 82/18 | 73/27 |
Gwinnett | 7.4 | 42/57 | 33/66 |
Chatham | 2.8 | 59/41 | 50/50 |
Clayton | 2.4 | 79/20 | 70/29 |
Cherokee | 2.2 | 29/70 | 20/79 |
Richmond | 2.1 | 66/34 | 57/43 |
Henry | 1.9 | 42/58 | 33/67 |
Muscogee | 1.9 | 60/39 | 51/48 |
Bibb | 1.7 | 60/40 | 51/49 |
Forsyth | 1.7 | 25/74 | 16/83 |
Fayette | 1.6 | 37/62 | 28/71 |
Hall | 1.5 | 30/69 | 21/78 |
Columbia | 1.4 | 33/66 | 24/75 |
Houston | 1.4 | 42/57 | 33/66 |
Coweta | 1.3 | 34/65 | 25/74 |
Douglas | 1.3 | 47/52 | 38/61 |
Paulding | 1.2 | 32/67 | 23/76 |
Clarke | 1.1 | 67/31 | 58/40 |
Carroll | 1.1 | 38/61 | 29/70 |
Dougherty | 1.0 | 68/32 | 59/41 |
Now let’s take a look at the 2008 numbers, including both the senate general election and runoff. (I’ve also included the white percentage of each county, as a means of seeing if a higher non-white electorate meant a higher drop-off in the runoff. But, as an indication of how polarized Georgia is, notice how well the white percentage in each county correlates with the Republican share of the vote in that county.)
County | % of 2008 general statewide vote |
2008 Pres. | 2008 Senate General |
% of 2008 runoff statewide vote |
2008 Senate Runoff |
2007 white % |
---|---|---|---|---|---|---|
Statewide | 100.0 | 47/52 | 47/50 | 100.0 | 43/57 | 62.0 |
Fulton | 10.3 | 67/32 | 63/33 | 10.4 | 60/40 | 48.5 |
Cobb | 8.0 | 48/54 | 42/53 | 8.6 | 36/64 | 63.4 |
DeKalb | 8.2 | 79/20 | 76/21 | 8.7 | 74/26 | 33.9 |
Gwinnett | 7.4 | 44/54 | 43/53 | 7.8 | 36/64 | 61.0 |
Chatham | 2.8 | 57/42 | 55/42 | 2.7 | 52/48 | 54.9 |
Clayton | 2.5 | 83/17 | 81/17 | 2.4 | 80/20 | 24.8 |
Cherokee | 2.4 | 24/75 | 24/71 | 2.4 | 18/82 | 83.9 |
Richmond | 2.0 | 66/34 | 64/34 | 1.9 | 59/41 | 43.4 |
Henry | 2.2 | 46/53 | 46/51 | 2.2 | 42/58 | 61.5 |
Muscogee | 1.9 | 60/40 | 60/38 | 1.6 | 57/43 | 47.6 |
Bibb | 1.7 | 59/41 | 57/42 | 1.7 | 53/47 | 46.2 |
Forsyth | 1.9 | 20/78 | 20/75 | 2.0 | 15/85 | 80.4 |
Fayette | 1.5 | 34/65 | 34/62 | 1.8 | 31/69 | 73.5 |
Hall | 1.5 | 24/75 | 26/68 | 1.6 | 20/80 | 86.9 |
Columbia | 1.4 | 28/71 | 28/69 | 1.4 | 24/76 | 78.8 |
Houston | 1.4 | 40/60 | 40/57 | 1.4 | 35/65 | 66.3 |
Coweta | 1.4 | 29/70 | 30/66 | 1.4 | 25/75 | 77.4 |
Douglas | 1.4 | 51/49 | 50/46 | 1.4 | 44/56 | 62.2 |
Paulding | 1.4 | 30/69 | 32/64 | 1.3 | 26/74 | 81.8 |
Clarke | 1.2 | 65/34 | 61/35 | 1.1 | 62/38 | 67.5 |
Carroll | 1.1 | 33/66 | 35/60 | 1.1 | 30/70 | n/a |
Dougherty | 1.0 | 67/32 | 64/34 | 1.0 | 64/36 | 33.3 |
Let’s start with how Obama and Martin (in the general) fared against the benchmarks that I set for them. On the whole, their actual percentages seemed to lag the benchmarks by about 2-3%, which is apt, as they both finished around 47%, 3% shy of a majority. There were only a few counties where they exceeded their benchmarks, and these are also the counties that are undergoing the most demographic change (in a way that’s favorable to the Democrats). Cobb and Gwinnett Counties are two of the four large counties in the Atlanta metro area, and are traditionally very conservative (they were Newt Gingrich’s turf back when he was in the House). But with Obama getting up to 48% in Cobb County and 44% in Gwinnett, they’re approaching swing county territory. (Cobb County is seeing growth in middle-class African-Americans and Gwinnett County is becoming an entry point for many Latino and Asian immigrants.)
More analysis over the flip…
Slightly further from the core of Atlanta are Clayton, Henry, and Douglas Counties, and these counties are being transformed even more rapidly by a rapid influx of African-American exurbanites. Clayton County’s white percentage, 24.8% in 2007, is down from 34.9% in 2000. Douglas County’s white percentage is 62.2%, down from 75.9 in 2000, and Henry County’s white percentage is 61.5%, down all the way from 80.1% in 2000… and that is matched by the double-digit swings in their voting patterns since 2004, and the way they exceeded their benchmarks (in fact, by 7% in Douglas County).
This is balanced by the mostly white and right-wing exurban counties at the northern fringes of the Atlanta area (Cherokee, Forsyth, and Hall Counties). Here, Obama and Martin trailed their benchmarks by the largest margins (by 5 or 6%).
The whitest counties (Hall, Coweta, Paulding) were the only counties where Martin (in the general) actually outperformed Obama, further suggestive of the racial polarization of the vote. By contrast, Martin tended to underperform Obama the most in heavily African-American counties (down 3% in DeKalb, 4% in Fulton, 3% in Dougherty). Interestingly, Martin also way underperformed Obama (by 4%) in Clarke County, not heavily black but home of Athens and the Univ. of Georgia. To me, this suggests that the underperformance has less to do with Obama/Chambliss ticket-splitting than with undervotes (i.e. casual or sporadic voters, probably disproportionately young and/or black, voting for Obama and not voting downballot). There were nearly 180,000 undervotes statewide between the two races (3.93 million total in the presidential, vs. 3.75 million in the senate race).
Now let’s turn to the dropoff in Martin’s performance between the general and the runoff. My initial assumption (and that of many other observers) was that Martin suffered for a lack of African-American turnout in the runoff, without the draw of Obama at the top of the ticket. That’s probably still true, but it’s a little more complicated than that. I’d expect the heavily black counties (DeKalb and Clayton) to have formed a smaller percentage of the statewide vote in the runoff than in the general, but that didn’t happen; in fact, DeKalb County’s share of the vote went up a lot, from 8.2% in the general to 8.7% in the runoff. The percentages of the vote didn’t change much, either. Martin only gave up 2% in DeKalb and 1% in Clayton, while the lone counties where Martin actually performed better in the runoff than the general were Dougherty (mostly-black Albany, downstate) and, again, Clarke (Athens/UGA).
Instead, the big dropoffs seemed to happen in the in the suburbs and exurbs, where Martin’s runoff numbers tended to revert back to very close to the 2004 Kerry/Bush numbers. For instance, out in wingnut land, Martin slipped from 20% to 15% in Forsyth County, 24% to 18% in Cherokee County, and 26% to 20% in Hall County. More alarmingly, the same rate of slippage happened in the more favorable suburban counties, like Cobb County (42% to 36%), Gwinnett County (43% to 36%), and Douglas County (50% to 44%). Interestingly, the percentages of the statewide vote in these counties, as with DeKalb County, went up too (8.0% to 8.6% in Cobb and 7.4% to 7.8% in Gwinnett), suggesting that the reliable Republicans who haven’t white-flighted it out of these counties yet continued to vote reliably in the runoff, while participation by other voters in these counties fell off dramatically.
To me, these numbers suggest some miscalculation at the organizing level… perhaps a focus on turning out every possible vote in reliable Democratic constituencies (DeKalb, Clayton, and Clarke Counties), while allowing other counties to slip through the cracks that people still aren’t used to thinking of as potentially Democratic counties (Cobb, Gwinnett, Douglas), as apparently many young and/or black infrequent voters in these rapidly-changing counties didn’t make it to the polls in the runoff. Not that these missing votes really mattered much in the end — Martin needed to not just match his general election numbers in the runoff but beat them by another 3% — but it’s food for thought on where to go trolling for those last few votes to try and get over the top in Georgia.