When Is A Conspiracy Not A Conspiracy?

It's easy for people to take a few facts and jump to conclusions.   On election day, a man came into our polling place and exited the voting booth and asked why there wasn't an Assembly race on the ballot.  We hadn't notice that and at first were concerned.  But then we realized not every Assembly seat is up for reelection this year.

But he said that his wife, who has the same address, voted early at Loussac and she'd voted for an Assembly candidate.  We couldn't explain what happened at Loussac, but we checked and found out that he lives in Patrick Flynn's district and Flynn wasn't up for reelection.

But in checking things, I found a link on the Municipal Elections webpage that got me to all the different Sample Ballots (there were about 48 different ballots to take care of all the Local Road Service Area elections) and a list of each polling place which said which ballot was to be used at each polling place.

And as I looked at the sample ballots I saw the candidates for the Assembly races and School Board races.  As you can see, the School Board races are all city wide seats, so they all show up on every ballot.  Some of the Assembly races had only one candidate. (For the sake of space I left out JOHNSON, Jennifer)

ASSEMBLY - DISTRICT 5 - SEAT H  HONEMAN, Paul Write-in SCHOOL BOARD - SEAT A SMITH, Don DAVIS, Bettye Write-in SCHOOL BOARD - SEATB NEES, David W. CROFT, Eric CORNWELL-GEORGE, Stephanie Write-in

ASSEMBLY - DISTRICT 3 - SEAT D HALL, Ernie Write-in SCHOOL BOARD - SEAT A SMITH, Don DAVIS, Bettye Write-in SCHOOL BOARD - SEATB NEES, David W. CROFT, Eric CORNWELL-GEORGE, Stephanie Write-in

ASSEMBLY- DISTRICT 2 - SEAT A  MULCAHY, Pete DEMBOSKI, Amy LUPO, SCHOOL BOARD - SEAT A SMITH, Don DAVIS, Bettye Write-in SCHOOL BOARD - SEATB NEES, David W. CROFT, Eric CORNWELL-GEORGE, Stephanie Write-in

ASSEMBLY - DISTRICT 4 -SEAT F CLARY,Andy TRAINI,Dick Write-in SCHOOL BOARD - SEAT A SMITH, Don DAVIS, Bettye Write-in SCHOOL BOARD - SEATB NEES, David W. CROFT, Eric CORNWELL-GEORGE, Stephanie  Write-in

HTML Tables

As I looked at the contested races, the candidate order on the ballot  seemed to favor Mayor Sullivan's candidates.

Don Smith was first in his race with Bettye Davis.
Nees was first in his race against Croft.
Mulcahy was first in his race.  (All seemed to be Conservatives and I wasn't sure who was endorsed by the Mayor, but Mulcahy had been appointed to the Planning and Zoning Commission by the Mayor so that seemed a safe bet.)
Clary was first in his race with Traini.

Whoa! I thought.  I knew that the first position on the ballot gets an advantage at the polls, because a certain number of people, if they aren't familiar with the candidates in the race, will just vote for the first one.  The best way to deal with that is to rotate the order on different ballots.  Then each candidate is first on an equal number of ballots.
Not only were the ballots not rotated, but the Mayor's preferred candidate seemed to be on top in each race.

This had conspiracy written all over it.   But first, some of the research on positional advantage on ballots.


From Northwestern University's Kellogg School: 

"Specialists in the mechanics of voting have long recognized that the order in which candidates’ names appear on a ballot influences voters’ decisions. Typically, candidates listed at the top of a ballot earn a greater share of the vote than they would receive in any other position, regardless of their policies and personalities. Now research on voting patterns in local state elections coauthored by a Kellogg School researcher has taken the issue a stage further. It concludes that the first listing on the ballot also increases a candidate’s chances of actually winning office—by almost five percentage points."

Stanford Professor Jon A. Krosnick describes the positional effect and how they've demonstrated it:

"How do we know this? Well, consider this: In California’s 80 Assembly districts, candidate name order is randomly assigned. In 1996, Bill Clinton’s vote tally was 4 percentage points higher in the Assembly districts where he was listed first than in the ones where he was listed last — a difference that persisted even after we took into account pre-existing Democratic registration levels in the districts.
In 2000, George W. Bush’s vote tally was 9 percentage points higher in the districts where he was listed first than in the districts where he was listed last — again, persisting with registration taken into account."

He adds this note which suggests the magnitude of the impact:

"In Florida, for instance, candidates from the governor’s party get top billing, which is why in 2000 and 2004 George W. Bush was listed first on every ballot. (His brother, Jeb, was governor.) "

Other states, he adds, order their ballots in different ways.  Some require rotated positions, some require the previous winning party to be listed first, Minnesota requires the party with the least votes in the previous election to be on top.  Some do alphabetical by party, some alphabetical by candidate's last name. Some random. 


I looked up the state law:

"(6) The names of the candidates for each office shall be set out in the same order on ballots printed for use in each house district. The director shall randomly determine the order of the names of the candidates for state representative for each house district. The director shall rotate the order of placement of the names of candidates for governor, lieutenant governor, United States senator, United States representative, and state senator on the ballot for each house district."

But this gets complicated.  There was a Supreme Court decision in 1998 where a candidate sued the Division of  Elections because he was disadvantaged by having a lower place on the ballot. [If the link doesn't work, start here at the Supreme Court site, link Alaska Case Law Service, then click "By Party Name" and write in "Sonneman"] The case said that the State had switched from rotation to random order with a 1998 amendment.  From the Court's decision:

The amendment was recommended by the Lieutenant Governor's Election Policy Transition Team. Its report stated that the amendment would save “between $150,000 and $250,000 per election cycle.” However, the actual cost of ballot rotation in the 1994 primary and general elections was $64,024. The amendment was also intended to eliminate the confusion of voters who relied on single-order sample ballots and were confused when they found a different rotation of candidates' names on their actual ballots. The team also concluded that “[r]esearch indicates that the order of candidates' names on American ballots does not significantly influence voters.”

Sonneman lost his case.  The Court decided that since the order was random, everyone had an equal chance for the coveted first spot.  I couldn't tell if it had been changed back since and that was why it said rotational in the statute I found.  I was going to see if I could call up Sonneman to see if he knew, but I got his obituary.  I don't have the app that lets me call the departed. 

So I called the Municipal Clerk's office to find out how it was done in Anchorage.  She was ready for that call.

Later, I videoed Deputy Clerk for Elections Amanda Moser explaining how the order is determined so you can listen to hear and/or read below.




Names are placed on the ballot in random order.  They have a written procedure and, in fact, the Clerk, Barb Jones, and her staff, and the Municipal Ombudsman were there as witnesses.  Here's  the procedure:

Procedure for Letter Drawing
Anchorage Municipal Code 28.40.010  Form
C.    The names of all candidates for the same office shall be on one ballot with spaces for write-ins equal to the number of offices to be filled. For each municipal election, the clerk shall determine the random alphabetical order in which the candidates' last names are placed on the ballot, regardless of the office sought, by conducting a chance selection of each letter of the alphabet. The sequence in which letters of the alphabet are drawn shall be the sequence of letters utilized in establishing the order in which the candidates' last names appear on the ballot.

1. Ensure that all 26 letters are present
2. One person will draw a letter from provided container.
3. A second person will read aloud the letter drawn.
4. A third person will record the letter drawn.
5. The fourth person is an observer.
4. Continue until all the letters are drawn.
5. After all letters are drawn the Clerk and other observers will sign sheet provided.
6. The Deputy Clerk will post on the Municipal Website.

And here's a copy of the list they made when they did the drawing.  Note, the date was January 24, 2013.  That's before people filed to run for office.

So, for each race, once they had candidate names, they went through this list.  Any Q's?  No?  M's?  etc.  In one case they had two candidates running for the same School Board seat whose names began with C - Croft and Cornwell-George.  Which should come first?  O comes before R in the alphabet, but they had to use this chart instead of the alphabet.  If you check the list, R is number 20 and O is number 24.  So Croft came first.

And when I looked further into this, I found out that in Eagle River, Demoboski, not Mulcahy was the Mayor's favorite.  And on the ballots with Ernie Hall's race, there were actually two Assembly races because Harriot Drummond had resigned to take her seat in the State House.  In that race, when I checked, Tim Steele's name was before the Mayor's candidate Cheryl Frasca.


So, you ask, if nothing was wrong, why write this post?  A reasonable question.  Here are some reasons:
1.  Don't jump to conclusions. It's always good to be reminded that one should do one's homework and get all the facts before jumping to conclusions, especially negative conclusions.  It reminds us that we see what we are looking for instead of what's actually there.  This is a good example of that and finding out there was no conspiracy, even though, at first glance, my evidence pointed in that direction.  

2.  We should write about good things as well as bad.  When the media only report things that go wrong, we get an unbalanced sense of how the world is.  The Clerk's office had thought through how they were going to do this, wrote up a procedure, and did the order randomly before they even knew who the candidates were.  And when a blogger called them up to check on what they did, they were prepared for me.  Their foresight on this should be recognized.

3.  I had all this information.  I didn't want it to go to waste.  A lame reason, but I'm trying to be honest here.


Final Thoughts

If the bump in votes due to position on the ballot is as big as the research says, then that's a pretty good argument for rotating the names.  But I think the research also needs to tell us if there is a population threshold when it rotating the ballots makes sense.  In my polling place we used less than 20% of the ballots.  Even if the names had been rotated, would we have used enough ballots to get to a different name order?  Should different parts of town get a different order for the School Board races - since they show up on all the ballots?

And I still have to find out what the current state law is - random or rotational?