Who’s Winning the Absentee War?

by: Grebner

Mon Oct 13, 2008 at 01:46:43 AM EDT

Before we fight the final battle of Election Day, the two sides are skirmishing over absentee voters.  As of October 10, over 680,000 Michigan voters had applied for ballots, and 80,000 of those ballots had been marked and received back by the local Clerks.

Both the Democrats and Republicans have spent large amounts of money and effort to encourage their partisans to cast early ballots.  I don’t have access to either Party’s list of people identified as favorable, so I can’t directly compare what fraction each side has motivated.  But my firm, PPC, has a model which assigns party coding to each voter, as well as coding for their likelihood of using an absentee ballot, based on their previous individual history of using absentee ballots.

The comparison is “descriptive” rather than rigorous research, since the data isn’t perfect and it isn’t possible to adequately control for various confounding influences.  All that said, I was prepared for more bad news (see my previous analysis of the 11/2006 GOTV, which seems to show the MDP was outhustled, even as its candidates won). 

I’m pleased to say we seem to be doing just fine on the absentee voter front.

The following table takes quite a bit of explanation. 

• First, like the drunk who searches for his lost keys only under the streetlight because of the good light, this analysis is limited to voters in municipalities which use the Secretary of State’s centralized absentee ballot processing program – about 90% of all ballots in the state. 
• Second, I excluded municipalities that vote heavily Republican or Democratic, because including them creates severe statistical problems.  (If Detroit is included, for example, are we getting a reading on the MDP, or on City Clerk Janice Winfrey?)  So we’re considering here only townships or cities which are between 37% and 63% Democratic.
• Third, I’m considering only people over the age of 60, who under Michigan law have an absolute right to use an absentee ballot.  Including younger people might shed light on how well each political party is doing among college students, travellers, and so on, but it’s difficult to tell the difference between lack of success and lack of legal eligibility.
• Fourth, the column I call “PAST-AVS APPLIED” is my projection of the probability that a given voter would use an absentee ballot.  This statistic is based on age (older voters become increasingly dependent on AV ballots), previous use of absentee ballots, living in a municipality which makes absentee voting easy, living with a spouse who already uses an absentee ballot, having their name on a list to whom applications are automatically mailed, and a few other factors.  As you can see by comparing my estimate to the percentage who had already been sent ballots by October 10, the model works pretty well.
• Fifth, the columns labeled DEM, SPLITTER, and GOP represent the the number of voters I code as (respectively) 75%+, 25-75%, and less than 25% likely to vote for a typical Democratic candidate.  This estimate is based on a huge amount of data my firm has collected over the years, as analyzed by an extremely complicated statistical model.  (It now includes the party choice for 80% of the people who voted in our abortive Jan. 15 presidential primary.)  Thus there are 80569 voters aged 60+ who I think are 75%+ Dems but who are roughly 0% likely to use an absentee ballot.  (The range is actually 0% to 5%. The next row, labeled “10%” represents the range from 6% to 15%.)
• Finally, the three columns labeled “APPLIED” show the percentage of the group which are shown as having actually applied for absentee ballots on the Secretary of State site, as of October 10.

PAST-AV DEMS APPLIED SPLITTER APPLIED GOP APPLIED

0           80569   3.5        106223    2.4        91309   2.9

10           4275   9.4          11527    6.4         2573    8.8

20           7280   4.6            5990    4.2       14410    2.5

30           9293   7.8            9017    4.6       15778    4.7

40         10582  11.3          11364    6.4       16433    6.8

50         12260  14.6          13763    9.1       18908    9.3

60         12127  21.0          14614  12.4       15946    14.7

70         12909  36.7          12956  24.8       15418    32.9

80         23101  59.2          15694  48.6       26880    58.7

90         36782  82.5          19062  72.6       34765    84.6

As I scan the table, three things jump out.  First, the Dems are certainly holding their own against the MRP’s aggressive absentee program.

Second, the people I have coded as “ticket-splitters” don’t seem to be on anybody’s radar, so nobody is hustling AV applications for them.  Left to their own devices, they’re noticeably less likely to have gotten an absentee ballot than their neighbors who are identified with either party.  (The fact that people in the same band have similar histories of using absentee ballots indicates this both parties must be doing something right.  Nobody is stopping the ticket-splitters from obtaining ballots just as they have previously; but nobody is badgering them either.)

Third, it looks as if the Dems are actually BEATING the Republicans among mid-likelihood absentee voters; that is people I estimate are 30% to 65% likely to use an absentee ballot.  This suggests the Republicans have focused their absentee efforts more narrowly on “the usual suspects”, while the Dems seem to be successfully targeting many “infrequent” absentee voters.