{Original analysis below the fold}
Last year, leaked memos from Michelle Nunn's campaign estimated
we will need 1,378,001 votes to win. [snip] Democrats with strong campaigns have generally won 1.2 million votes.
We are aiming to enlarge the electorate by approximately 200,000 votes from drop-off and inactive voters.
[snip]
In addition to enlarging the electorate, we aim to persuade another 100 thousand white voters, reaching 30 percent white support, which our polling shows is feasible, while discouraging the kind of Republican surge we saw in 2010 so the total vote in the U.S. Senate race is approximately 2.7 million
Monday, during a campaign appearance in Georgia, Michelle Obama
said:
“If just 50 Democratic voters per precinct who didn’t vote in 2010 get out and vote this November -- just 50 per precinct -- then Michelle Nunn and Jason Carter will win.”
Given
2,727 precincts in Georgia, the campaign is talking about
roughly 136,350 voters that didn't turnout in 2010. Our candidates seem to see a clear path to victory.
Below the jump ropes, I'll explore the pool of unreliable voters.
In Georgia, for example, the [DSCC] estimates that there are 572,000 unregistered African-American voters, and that there are more than 600,000 likely supporters of Michelle Nunn, the Democratic Senate candidate there, who voted in 2012 but not in 2010. The goal, then, is to register the African-American voters, and to target the likely Nunn voters to show up at the polls during a midterm election. - NYT
The drop-off midterm voters are often single women, younger and lower income, but what about Georgia-specific projections? Of course, Georgia Victory 2014 has very detailed data about which individuals and groups to target and in which regions, etc. But, like most campaign volunteers, I'm not in the in. So I performed an analysis of official election data.
In the 2010 midterms, 52% of registered voters in Georgia actually voted (2,622,527 / 5,032,354); while in the Presidential year of 2012, the figure was 73% (3,908,500/5,360,701). A large 21% drop-off overall.
As you can see in the below table, the drop-off is much worse among younger registered voters.
Estimated midterm drop-off in Democratic votes by age group for Georgia. |
AGE |
2010, voting among registered (a) |
2012, voting among registered (a) |
% Midterm Drop-off (b) |
Number registered in 2012 (c) |
Number of unreliable voters (d) |
Estimated Dem support (e) |
Estimated Dem drop-off (f) |
18-24 |
18.5% |
50.3% |
31.8% |
629,970 |
200,295 |
65% |
130,192 |
25-29 |
28.6% |
56.7% |
28.1% |
457,155 |
128,280 |
65% |
83,382 |
30-34 |
39.2% |
65.7% |
26.5% |
456,007 |
120,895 |
60% |
72,537 |
35-39 |
47.0% |
71.7% |
24.8% |
450,029 |
111,392 |
55% |
61,266 |
40-44 |
52.7% |
75.8% |
23.1% |
519,985 |
120,096 |
55% |
61,266 |
45-49 |
59.3% |
80.0% |
20.8% |
526,033 |
109,304 |
50% |
54,652 |
50-54 |
59.3% |
82.0% |
22.7% |
476,792 |
108,407 |
45% |
48,783 |
55-59 |
66.1% |
84.3% |
18.2% |
413,659 |
75,490 |
45% |
33,970 |
60-64 |
71.2% |
84.3% |
13.2% |
413,659 |
54,487 |
45% |
24,519 |
65+ |
70.7% |
82.2% |
11.5% |
908,540 |
104,365 |
30 |
31,310 |
|
|
|
|
|
|
|
606,664 |
(a) Official statistics for the % of registered voters that actually voted.
(b) The additive difference between the percentages.
(c) Number of registered voters in 2012 for these age groups. (I did not tabulate current voter registration data for this column, which has been updated through July 2014 but is disaggregated by county).
(d) The total pool of "unreliable voters" is obtained by multiplying (b*c) the percent "drop-off" by the number of registered voters (This number includes people across the political spectrum).
(e) This is a ballpark estimate for Michelle Nunn's level of support among this age group (loosely based on the cross-tabs from the recent YouGov poll)*.
(f) This is the final number of unreliable Democratic voters obtained by multiplying (d*f) the total "unreliable voters" by the guesstimated level of Democratic support.
|
For simplicity, this analysis ignores how people belong to cohorts which move into older age groups with time. Nonetheless, this worked out nicely, and ended up being in line with the "more than 600,000" quoted above from The NYT. No post hoc manipulation of numbers necessary ;-P
I performed a similar analysis for African Americans by gender, where Nunn's support was guesstimated at 95% among women and 80% among men.
Estimated midterm drop-off in Democratic votes among African Americans by gender, Georgia. |
Group |
2010, voting among registered |
2012, voting among registered |
% Drop-off |
Number Registered in 2012 |
Number of unreliable voters |
Estimated Dem support |
Estimated Dem drop-off |
AA Women |
54.8% |
77.3% |
22.5% |
949,498 |
213,588 |
95% |
202,908 |
AA Men |
44.2% |
65.8% |
21.6% |
660,486 |
142,569 |
80% |
114,055 |
I also broke up African Americans by gender and by age group (not shown). Those results mirrored the above data according to age group and reflected the lower participation of African American men.
--------------------
The end message is that, yes, there are 140,000 or even 200,000 votes to be found among "midterm drop-off voters", inactive voters and especially newly registered voters. What percent of eligible voters are actually registered in Georgia, according to age groups, gender and race groups?
To register to vote visit www.MyGAVote.com and to check your registration status visit http://www.mvp.sos.ga.gov/
Volunteer to help GOTV with Michelle Nunn, Jason Carter and Georgia Victory 2014.
* P.S. I found it very difficult to guesstimate a level of support for Michelle Nunn according to age group in Georgia. If anyone wants to provide advice, I'm all ears.
For example, Landmark Communications polls "active" voters (basically registered voters). The only cross-tabs available from them are for their July 25th release, and have Nunn losing by 7 among 18-39 year olds (n=148), winning by 12 among 40-64 year olds (n=420), and losing by 7 among 65 and older. There's also the YouGov poll with a larger sample size, but it uses a likely voter screen. It finds more realistic margins (ie Democratic advantages) in the youngest groups: Nunn up 25 among 18-29 year olds (n=86, weighted n=211), Nunn up 16 among 30-44 year olds (n=241, weighted n=474) and Perdue leading age 45 and older.
---
UPDATE I:
Adding the below image of my guesstimates for all racial groups:
Not shown in the above image: Compared to other groups, a much lower proportion of eligible Hispanics/Latinos are registered to vote. Among African Americans as a whole, voter registration actually (slightly) exceeds the proportion among whites in Georgia.
UPDATE II: I corrected an error in the number of registered African Americans in 2012. The "estimated Dem drop-off" (also called "lost Dem votes") increased by ~20,000 due to this.
FYI, the official statistics come from the SOS page on turnout by demographics (2010: http://sos.ga.gov/... and 2012: PDF.