Absentmindedness
I've spent a few hours today analyzing absentee voter behavior over the past three elections. I've uncovered some interesting trends.
First off, election night totals include all the absentee ballots that were mailed in. In recent elections, they have comprised roughly 20% of each candidate's vote total on election night.
What is interesting are two factors I've found to be remarkably consistent. The first is that candidates that most would label "liberal" ultimately derive about 33% of their final vote total from absentees, whereas those most would label as "conservative" typically derive around 45% from absentees. The second factor is that the absentee ballots liberals receive are roughly 50% mailed in, with the remaining 50% submitted on election day. Conservative candidates receive roughly 35% of their absentee ballots by mail, with the balance turned in at the precinct polling stations.
The upshot is that while all candidates get a bump from the late absentees, conservatives have historically enjoyed a bigger bump.
Applying the math, I have calculated what the vote totals would have to be in order to sustain that trend. According to my projections, the final ranking for the Chico City Council will be Flynn, Sorensen, Gruendl, Herbert, Nickell, Dailey.
In the Supervisor race, the factors are less clear. In the last election Maureen Kirk stood for, her absentee total was only 25% of her final total. If her election night total was 20% absentee, her bump would be around 436 votes, or a total of 6089.
The last time Steve Bertagna stood for office, his absentee votes were 44.2 percent of his total. If he got a 20% mail-in absentee on Tuesday, and the trend remains valid, his bump should be over 2700. This could be problematic, inasmuch as there are only 2662 uncounted absentee ballots in the Third District. It's more likely that the county absentee voting trends are different from city precincts, but there's still a chance that Bertagna could ultimately prevail. He and Kirk are only about 250 votes apart, and with 2662 in play, there's a lot of room for movement. Unfortunately, there's no useful absentee data available on the clerk-recorder's website from the primary last June.
But I digress. If recent trends are sustainable, the ultimate outcome in the city council race could see Sorensen not only leapfrogging Nickell, but Gruendl as well. If this happens, Herbert comes within about 500 votes of third place. If his unexpectedly poor showing on election night was a result of a higher than usual late absentee vote, it could be very close indeed.
Wishful thinking? Perhaps. The old sayijng about lies, damned lies, and statistics may obtain here. But look on the bright side; if I'm totally off base, you can slag me silly when the official Statement of Vote is published. Check back after Turkey Day and we'll see.
Comments
Surly you did this voting pattern analyses before the election? Why would you need anything but the 2006 Primary statement of the vote from the clerk-recorder’s web site? You are the pac-man, you have the Clerk’s updated voter file on your computer. It has all of the voting history info in it, including voting methods. How else could you have targeted your mail?
There is all kinds of useful pac info on that file for example of the residents at : One is a democrat the other declines to state, one has voted at the polls every election since the 2003 recall and the other hasn’t bothered to vote. In targeting mail a conservative pac might not want to bother mailing to this address.
Posted by: T F Green | November 10, 2006 10:26 AM
Surly you did this voting pattern analyses before the election?
No, I didn't. I think I was pretty clear that I did that yesterday. Absentee voting patterns did not become meaningful until Wednesday morning.
And don't call me Surly.
Posted by: Alan Chamberlain | November 10, 2006 10:57 AM
I made some similar calculations last night from the same data, and I have to agree with you. Sorensen has a chance as does Bertagna.
Posted by: Anthony Watts | November 10, 2006 02:15 PM
Alan,
RE: your conclusion: "The second factor is that the absentee ballots liberals receive are roughly 50% mailed in, with the remaining 50% submitted on election day. Conservative candidates receive roughly 35% of their absentee ballots by mail, with the balance turned in at the precinct polling stations."
What data are you looking at that makes you think this is the case?
The numbers (source):
Democrats in Butte turned in 10,456 absentee ballots pre-election day.
Republicans in Butte turned in 14,495 absentee ballots pre-election day.
Unprocessed ballots left to count in Butte County (source): 13,838
Maybe I’m missing something, but are you suggesting that in Butte County on Election Day, liberals should be able to count on 10,000 new ballots and conservatives should be able to count on nearly 29,000 new ballots?
Or are you suggesting that of the 13,838 unprocessed ballots, 65% should be conservative leaning? or what?
Conventional wisdom in California is that the dems lose the absentee votes, sometimes badly, and then make up for this deficit in the votes cast on election day in the voting booth. Republicans traditionally have had a lopsided advantage in the “8:01’s� (called that because they are the first votes counted and put up on the board right after the polls close at 8:00 p.m. and they are exclusively the absentee votes.)
On November 7, the early county returns throughout the State showed that Democrats were almost even w/ the Republicans in the number of returned absentee ballots (41% to 43%). In Butte County, the numbers weren’t so close. Only 36% (10,456 ballots) of the total returned absentee ballots (29,272) were from Democrats and nearly 50% (or 14,495) were from Republicans.
So when the first results were announced (absentee only) and Kirk only trailed Bertagna by 174 votes in the 3rd SD (48.2% to 51.6%), even though only 36% of the 8:01’s in the 3rd were from Dems, we knew Kirk was in good shape.
And what about the DTS absentee voters? Are they considered liberal or conservative in your analysis? DTSers turned in nearly 10% of pre-election day ballots and clearly played a major role in the City Council and the Kirk-Bertagna race.
In my opinion, Dan’s post re: unprocessed ballots is actually much closer to the mark in explaining what will likely happen w/ the remaining 13k AB votes. In the 3rd, Bertagna will need about 57% of the unprocessed ballots to swing his way in order to win.
In the City Council races, the only way that Nickell holds on is if he wins a higher percentage of election day AB votes than he won from pre-election AB voters. If the election day AB trend is consistent w/ the pre-election ABs, then there is a good probability that Sorensen picks up 80-120 votes and overtakes Nickell.
Re: the question of who votes AB pre-election vs. election day: In my experience, AB voters that strongly identify with a party tend to vote early (by mail) and AB voters who are more independent or undecided are more likely to cast their AB ballot on election day. So, in my opinion, Bertagna may pick up 50-100 votes if the election day AB voters vote the same way that pre-election AB voters voted. But, I think the AB voters who cast ballots on election day are more likely independent or undecided (even if they affiliate w/ a party) and those votes were more likely to go democrat this cycle. So I think there is a reasonable chance that Kirk actually doesn’t lose vote and maybe even gains votes.
Re: your prediction that Sorensen overtakes Gruendl…Given that there are probably somewhere in the neighborhood of 5,000 unprocessed ballots left to be counted for the City of Chico (source: add sup districts 2 and 3, the adjust a little), there is no way Sorenson will make up a 1,200-2,000 vote differential to place second behind Flynn.
The total numbers (source):
Gruendl received 8,950 votes
Sorensen received 7,756 votes (roughly 1,200 votes behind)
City Council AB returns (source: the 8:01’s):
Gruendl received 3,039 votes (16.23%)
Sorensen received 3,289 votes (17.99%)
If you accept that there are roughly 5,000 unprocessed votes in the City of Chico, and the incredibly unlikely scenario that Gruendl gets 0 (zero) votes in this last batch of votes, then Sorensen would need votes from at least 24% of the unprocessed ballots in order to overtake Gruendl. If Gruendl gets ½ of what he got from pre-election day Abs, then Sorensen would need to garner 32%... If the trend for election day Abs is consistent w/ pre-election day Abs, and Gruendl gets about 800 votes from the last batch, then Sorensen would need 40%... So, that’s not going to happen.
The only real question is what happens to Nickell-Sorensen, which will be decided by about 40-80 votes.
Tovey
Posted by: Tovey | November 10, 2006 03:33 PM
The HOA should have hired Tovey when they had the chance. You know that he did apply for the job.
Posted by: T F Green | November 10, 2006 05:39 PM
Tovey--
I analyzed the results of the last three elections. In the 2004 elections, the election night results included mailed-in ballots, and all candidates received between 19 and 21% of their election night number from mailed-in ballots.
After everything was totaled up and the final results were posted, candidates generally considered to be liberal (Schwab, Holcombe, Merz, etc.) got 33% of their total from absentees, while candidates considered more conservative (Bertagna, Wahl, Davis, Francis, etc.) received about 45% of their total from absentee ballots.
By deducting the mailed in ballots from the election drop-off, I was able to calculate that liberals get about half of their absentees by mail, and half on the day of the election, whereas conservatives get about 35-40% of their absentees by mail, and the remainder at the precinct polling station.
It's remarkably consistent, and the trend is stable across three previous elections. In 2002 (worth considering carefully, inasmuch as it was a midterm, pick 3, with six names on the ballot) and 2000, total absentees were roughly 25% of the total, while ratios were low to mid 20s for liberals, low to mid 30s for conservatives.
In 2004, the number of absentees took a jump (39% of total), with the relative differentials amplified by the spike.
It may well be that absentees went up even more this year, although it looks like it's roughly the same (38.7 if we get another 14,488 absentee votes).
And that's about right. You say ~5000 votes, but that's wrong (as are your earlier numbers). It's ~5000 ballots, each of which is worth three votes in a pick 3 race. That would give us a total of 61,699, or a 41 % increase over 2002, which would be a direct result of an overall increase in absentee voting (the election night totals only jumped 7% over 2002).
Anyway, that's the methodology. I will post my final vote count predictions in a future blog, nearer to the reveal, and we'll see if the numbers pencil.
Posted by: Alan Chamberlain | November 10, 2006 05:51 PM
Alan,
Well, I suppose we'll see soon enough who's theory more closely reflects the ultimate outcome. :)
re: HOA: I did meet w/ a number of individuals who ended up forming HOA, but I never applied for a job. I think it's fair to say I got to know them a little better and they got to know me a little better. Neither party pursued the relationship.
Tovey
Posted by: Tovey | November 11, 2006 07:49 AM