Summary: Comparing turnout in the 2008 runoff and 2010 midterms, most of the "runoff drop-off" (58%) was attributable to decreased early voting opportunities during 2008's runoff, which will not occur to the same extent this time.
---
DailyKos Race Ratings assumes that
Since Democrats have fared very poorly in the rare runoff elections of previous years (see the cases of Jim Martin in 2008, or Wyche Fowler in 1992), our model treats finishing over 50 percent as make-or-break for Democratic candidates Michelle Nunn and Jason Carter.In 1992, Fowler (D) lost the runoff by just 1.5%, even gaining a marginally higher percentage in the runoff. That election does not support the idea of Democrats fairing poorly in runoffs. Discounting 1992, leaves the 2008 runoff.
Taniel's post in May reasons why avoiding a runoff is considered "make-or-break", although the author never puts it in such stark terms. Moving from a 2-point spread to a 14-point spread was quite a drubbing.
But that was a Presidential year, and Taniel's post acknowledged that turnout in a Senatorial runoff might best be compared to turnout in a midterm year:
In 2008...the runoff's electorate was 17 percent smaller than than that of the 2010 midterms. (It was, however, comparable to that of the 2006 midterms, which had no competitive statewide race.)Alternatively, 40% of the voting eligible population voted in 2010 versus 36% of the VEP in 2008's runoff. Below, I'll return to a comparison of voting in the 2008 runoff and the 2010 midterms, concluding that most of the drop-off (58%) was attributable to decreased early voting opportunities during 2008's runoff, which will not occur to the same extent this time.
---
THERE'S NO DATA SHOWING A "RUNOFF DROP-OFF" AMONG MOTIVATED MIDTERM VOTERS
There's also the assumption that youth and African Americans would have lower runoff turnout, despite the GA Secretary of State not publishing demographic data for 2008's runoff election. There's no data to support the hypothesis that "runoff drop-off" would occur this year among the subset of more motivated midterm voters.
Compared to Presidential years, there is a midterm drop-off among youth (partially, perhaps mostly, a function of lower voter registration among youth). But I found no data showing a "runoff drop-off" among midterm youth voters.
In 2008, demographic data by race was only published (and tabulated) for the early voting period. Those numbers were used by 538 to project a disproportionate drop-off in African American participation in the runoff overall.
In the 2008 general election, early voting occurred over a 45 day period and accounted for 53% of votes, and 34% of early voters were African American (Pew, "A Survey of Georgia Voters in the 2008 General Election"PDF). But in the 2008 runoff, with wide variability in availability, early voting accounted for 23% of all votes (491,540 of 2,137,956) and 23% of early voters were African American. There were only 3 days of early voting in some key Democratic-friendly counties.
The unequal number early voting days obviously tilted the final early voting numbers and shaped a key narrative. This narrative could have even influenced real individual decisions about whether or not to vote on election day by creating a sense of lower community turnout.
The AJC recently published that, in 2008's runoff, African American participation was "down only two points from the general election," concluding that "race may be a smaller factor than many think." This 2% is less than the 3-6% guesstimated at 538. (Several days ago I contacted the author of the AJC article--through their online contact form--but have not received an answer about the source for this 2% figure). Anyways, there's no data to conclude that this "runoff drop-off" would also affect the subset of more motivated midterm voters.
----
Returning to the comparison between 2008's runoff and the 2010 midterms, here's a look at early voting and election day voting as a percentage of the voting eligible population.
| 2008 (VEP: 6,281,872) | 2010 (VEP: 6,464,406) | ||||||
| Early Votes | Election Day Votes | Early Votes | Election Day Votes | ||||
| # | % of VEP | # | % of VEP | # | % of VEP | # | % of VEP |
| 491,540 | 7.8% | 1,646,416 | 26.2% | 774,690 | 12.0% | 1,847,837 | 28.6% |
Therefore, some 64% of the runoff drop-off was due to decreased early voting, comparing 2008's runoff to the 2010 midterm. Further, it can be said that the election day drop-off in the 2008 runoff compared to the 2010 midterm was 8.3% ([.286-.262] / .286) versus a 34.7% drop in early voting ([.11984-.07824] / .11984). If you assume that early voting in the 2008 runoff "should" have dropped by the 8.3% drop seen on election day--a scenario with equal early voting availability--then fully 58% of the overall runoff drop-off was attributable to decreased availability of early voting, between the 2008 runoff and 2010 midterms ([261,275*{1-.083}] / 410,519).
----
In the 2014 (actually, the 2015) runoff for US Senate, the early voting period has apparently been set for December 15th to January 2nd (See: (1& 2:PDF). It's unclear if early voting would be closed for each of the official holidays on the 25th, 26th and 1st, but I'd make that assumption. Note that for the general election, early voting starts on Columbus Day (an official state holiday). Anyways, I count 12 days (Dec 15-19 [5 days], 22-24 [3 days], 29-31 [3 days] & Jan 2) ignoring potential weekend voting, which would be determined by individual counties.
Also in 2012, under the newly enacted 21-day period, early voting accounted for 35.2% of all votes (1,885,558 of 5,360,701, and 34% of early voters were African American.
---
THE GUBERNATORIAL RACE
The current argument for why GA's US Senate runoff is underrated does not apply to the Gubernatorial and statewide runoffs--currently scheduled for December 2, 2014. Here early voting days will be curtailed potentially as much as in 2008. The above cited PDF from Dawson County indicates the early vote period as "November 17 – 26 (or best effort)", which would be the same as 2008's runoff.
But Carter's chances could also be underrated. It has been suggested that "with the U.S. Senate race stripped away, a gubernatorial runoff would become less ideological. The Republican incumbent would become the primary focus". Further, the AJC also reported last week that Democrats may file suit to push the statewide runoff back to the same date as the federal election. TBD after election day.
In 2008, it took more than a week for the SoS to determine there would be a runoff election, decreasing early voting:
county results were in the process of being certified and a decision was expected from Handel as early as Wednesday, Nov. 12. [snip] Thanksgiving holidays combined with the Secretary of State's delay in certifying the senate race has shortened the normal early voting period---
- Source
The "Thanksgiving holidays" include Friday, which in Georgia is the official state holiday for Robert E Lee's Birthday. From 1986 through 2011, this state holiday was observed on the "correct" date (Jan 19) and originally revived to push back against the introduction of MLK Day (Jan 20). Of course, the correct date for Lee's birthday is in history books. Of note, four other states still have a holiday for Lee on Jan 19th: AL, AR, FL & MS.