Slog

News & Arts

Line Out

Music & the City
at Night

« The World Series of Baseball | The Morning News »

Thursday, November 5, 2009

Election Fun with Numbers: Voter Turnout

Posted by Jesse Vernon on Thu, Nov 5, 2009 at 7:00 AM

Curious about patterns in voter turnout and piqued by Eli's questions, I started digging around King County's and Washington State's stats from past elections. Other than some shoddy webmanship (y'all gotta fix this shit), nothing particularly surprising emerged—though it looks like initiatives and propositions tend to rally Seattle voters more than mayoral elections do. I'm no electoral clairvoyant, and I'm sure people like this have a much more complex calculus for predicting elections. But I was frustrated with the lack of data about the specifics of turnout. Where do late voters fall on the political spectrum? Do votes even get counted in the order they are received, or in reverse, or is there a sorting hat? (You can check if your ballot has been received here.) So I took what was out there and did some number wrangling. For sensitive mathophobes, I've hidden the calculations (and more sexy, sexy graphs) below the jump.

1999 never happened

Let's start with Seattle. There were 377,085 ballots issued for this election, and as of 8:00 p.m. last night, 164,113 (43.52%) were received. But that doesn't mean that the other 212,972 are yet to trickle in. The average voter turnout for the past two mayoral election years is 48.39%. If this number is a reliable indicator,* we should be waiting on about 18,358 ballots. As Dom pointed out last night, Mike McGinn is only leading by 462 votes.

What about King County, where the battle for executive is ostensibly over? King County has this dry .pdf of turnout results from elections past, but a graph is so much prettier. The average voter turnout in King County for a non-presidential-election year (calculated over the past 11 years) is 51.01%. As of 8:00 p.m. last night, 438,557 ballots were received (of 1,084,590 issued). So again, if we pretend that turnout is our only indicator, we are still waiting on around 114,692 ballots. Dow Constantine is winning by 45,399.

One caveat about the Seattle and King County stats: These calculations are not completely accurate because the latest available update for specific vote tallies was 4:20 p.m., whereas the total-votes-counted numbers are as of four hours later.

Washington State did not believe in the internet before 2004

Washington State did not believe in the internet before 2004

Now to peek outside of our bubble at the yellow-bellied rest of Washington State. Average voter turnout in Washington in a non-prez-election year? 56.47% (calculated 2005 through 2007—the state's data before then is hidden in an archaic web maze; I'm not that eager). We have 3,583,278 registered voters, and as of 6:30 p.m. last night, those tired machines had counted 1,170,658 votes. ::mathmath:: There should be 852,819 ballots still teetering in. As of that same time, Referendum 71 was passing by 37,866 votes. And the evil Eynitiative 1033 was losing by 137,421.

*These numbers are not meant to be actual predictions—rather, I'm hoping they'll be interesting barometers to compare to the actual numbers when know 'em. It will give us an idea of how steady voter turnout is (or if the Obama surge had any staying power), and maybe we can find some telling patterns and get better at answering these pressing questions. Well, except what God wants. Sorry, Eli.

If you find any fuzzy math or can direct me to people who have more stats about local voter-turnout patterns, kindly let me know.

Comments (28) RSS

Oldest First Unregistered On Registered On Add a comment

Great reporting Jesse. You won't find anything like this on the Suburban Times or SeattlePI.com.

Posted by I Got Nuthin' on November 5, 2009 at 7:03 AM

I like how almost all Google searches bring up the old domain name and thus that charming "this domain has been retired" shit instead of, say, bus schedules.

Posted by Fnarf

on November 5, 2009 at 7:04 AM

I'm pretty sure that there should be more than 137,421 ballots left to come in in washington state. first, it's only 20k more than the number left in washington. second, it's exactly the same as the amount of votes that 1033 is currently losing by. might want to check that. i have no idea about local voting patterns, however.

Posted by Cook on November 5, 2009 at 7:05 AM

@3, There are probably not that many more to come in the mail. If it was mailed on election day before the late evening it probably was delivered yesterday. Some will come today but after that its really only the out of state voters that are going to be left.

Posted by giffy on November 5, 2009 at 7:26 AM

@3, fixed that number. It should actually be 852,819. Thanks!

Posted by Jesse Vernon on November 5, 2009 at 7:28 AM

i actually don't care about this at all, but i just want to say, @4, OH SNAP. ok i'll stop being a douchebag now.

also, it seems to me that if king county is only missing 110k ballots, the state missing 850k is wayyy too much. especially since there are no congressional races. i'm betting that the state will have pretty low turnout (comparatively) and king county (and possibly a few other western counties) will be higher than a normal off year. just a thought. that being said, go mcginn!

Posted by Cook on November 5, 2009 at 7:34 AM

Just checked my ballot status. It's in, but not counted. Mailed it on Tuesday.

Posted by PTrig on November 5, 2009 at 7:48 AM

As to the question of the order in which they're counted, I dropped my ballot in a drop-box early on election day and according to the status site it has not been counted. Assuming it's not been lost, I find some comfort in the fact that McGinn's current total is shy by at least my vote

Posted by onemorefor71 on November 5, 2009 at 7:49 AM

3D bar graphs can not possibly be categorized as sexy.

Posted by josh

on November 5, 2009 at 7:50 AM

@9: Ah, so you're not a girth man, you prefer the length of a line graph, eh?

Posted by Baconcat on November 5, 2009 at 8:05 AM

@10, I'd like to get my stubby little snout deep into your pie chart.

Posted by baconpussy on November 5, 2009 at 8:15 AM

@11: As long as you promise to take your heart pills and diabeetus meds. I'll wear a nametag, in case you forget my name.

Do you need some kind of ramp or special walker for this sort of thing?

Posted by Baconcat on November 5, 2009 at 8:23 AM

My husband and I checked our ballots - not received yet apparently, or not scanned into their system yet. Mailed them on Tuesday, hope they didn't get lost. This whole voting by mail thing freaks me out a bit. I miss voting in person.

Posted by caitlin s on November 5, 2009 at 8:54 AM

I think the late ballots should skew uberliberal/McGinn, because lazy stoners like me tend to procrastinate. And be uberliberal.

Posted by Violet_DaGrinder

on November 5, 2009 at 9:04 AM

Sent my ballot Monday from university . . . in Montréal. My ballot status shows it hasn't been delivered by USPS yet.

(yeah, there are a few of us way out there).

Funny enough was the parked car I saw on the way to the mailbox which had WA plates. That alone is extremely rare to see in Québec, but the owner of this mid-80s Mercedes was clearly flouting something illegal: the rear plate was series X, while the front plate was a much older series M. I was all, "Let me guess, you tried to avoid paying for the monorail tabs when that was still on the front burner by re-registering your car in Twisp instead of Woodinville."

Posted by The Waking Hours on November 5, 2009 at 9:28 AM

My ballot and my wife's were mailed from the same mailbox at the same time on the day of the election. Mine's been received, hers hasn't. I'm guessing there's a large pile of ballots that haven't been scanned as received yet.

Posted by Matt the Engineer on November 5, 2009 at 9:44 AM

Jesse, the Secretary of State's web site has a tab on the election results that shows ballots counted, estimated turnout and estimated remaining ballots left to count.

http://vote.wa.gov/Elections/WEI/VoterTu…

Posted by Smartypants on November 5, 2009 at 10:07 AM

@17 The numbers on that site do figure into my calculations. But the percentage they give is not estimated voter turnout; it's voter turnout so far. Also, "estimated ballots on hand to be processed" does not include ballots not yet received.

Posted by Jesse Vernon on November 5, 2009 at 10:26 AM

Seattle is forecasting turnout in this election of 57%

Posted by David Miller on November 5, 2009 at 10:27 AM

Jesse,

Even though King County Elections has received 164,113 Seattle ballots, they have only processed 109,874 of them. That means that they are still sitting on 54,239 ballots in addition to any that are still trickling in in the mail.

http://your.kingcounty.gov/elections/200…

Posted by passionate_jus on November 5, 2009 at 10:29 AM

@12: It probably wouldn't hurt to get some of those dog stairs for the bedside so I can get up there and do my bidniss.

And how did you know I'm a huge Wilford Brimley fan? DIABEETUS!!!

Posted by baconpussy on November 5, 2009 at 10:32 AM

Jesse - I applaud your work, but unless I'm reading you wrong, your turnout forecasting could stand some refinement. Simply lumping all elections as presidential/non-presidential skews things badly. Turnout in any given year tracks most closely to the same point previously in election cycles, eg 2009 with 2005, 2008 with 2004, 2007 with 2003. At the very least, seperating on- and off-year elections would give you much closer results. It's standard practice in the campaign industry to judge turnout based on prior turnout in on- and off-years, not "pres vs non-pres". Take the last four points you have data for on a voter (including primaries) in the given cycle and use that to judge if they'll vote again. Usually 3/4+ (that is, someone who has voted in at least three of the last four elections - for this general, that would have been 2009 primary, 2005 p&g, and 2001 general) can be counted on to vote again. So I'm afraid your turnout predictions will be high as you're including even year turnouts in your off-year predictions.

Also, @9,@10,@11 as a sequence made my morning.

Posted by Juris on November 5, 2009 at 10:45 AM

A quick note: "ballots received" is not the same as "ballots counted". So if you're trying to get an idea of how many more votes we're waiting to appear on the scoreboard, take a look at the "ballots cast" on the results page.

Posted by meggo on November 5, 2009 at 11:03 AM

The only comparison to this year would be a non-presidential year with no US Senators or Congressmembers on the ballot and at least one contentious ballot issue - and a mayoral race that was close. Where people mailed in their ballots.

In other words, we have no comparison.

Posted by Will in Seattle

on November 5, 2009 at 11:08 AM

I'd say the biggest issue with trying to predict an outcome based on past turnout is our switch to all-mail ballots. How many people who used to be too lazy to go to a polling place in an off-year election were still too lazy to sit down in their living room & fill out the ballot? Or will all-mail actually improve "turnout"?

We have almost no information regarding past "late voters" unless you want to start with the hypothesis that the people who mailed in their ballots late Tuesday are also the ones who went to the polling place late in prior elections. I don't think it works that way, though.

Posted by serotonein on November 5, 2009 at 11:15 AM

You could extrapolate out a pretty good guess at expected returns by looking at the number of permanent absentees in 2005 and their rate of return.

Posted by Juris on November 5, 2009 at 11:29 AM

My Ballot Status tracker says 'Ballot mailed', then 'Ballot received', then 'In the bed of Tim Eyman's truck', then 'Buried outside Sequim'. Is that bad?

Posted by sidereal on November 5, 2009 at 2:57 PM