NOTE: This page has been updated to reflect the final results released by the Alameda County Registrar of Voters. FairVote's final analysis of the election is available here.
November 10, 2010
FAQ: Ranked-Choice Voting and the Oakland Mayor’s Race
- Out of all Oakland voters who participated in the mayoral race, 99.8% cast a valid ballot.
- 72% of Oakland voters ranked three different candidates for mayor. Among those who did not rank three candidates, four in five still had their vote count in every round of counting.
- The "undervote" (people who skipped the race entirely) in 2010 was half of the undervote in the 2006 mayoral election.
- 27% more Oakland voters have participated in this year's mayoral election than the 2006 election.
- The final results are delayed due to ballot-counting, not RCV. Just as in the 2006 mayoral race and this year's statewide race for attorney general, all the absentee and provisional ballots must be tallied before you can be sure of winners of close contests.
- With RCV, the winner is the candidate who has a broad base and is preferred by the most voters in that election. If Jean Quan defeats Don Perata, it will be because more voters ranked Quan ahead of Perata on their ballots.
How did Oakland voters handle the new ranked-choice voting method?
Out of all Oakland voters who participated in the mayoral race, 99.8% cast a valid ballot. Furthermore, the great majority of voters ranked more than one candidate: 72% of Oakland voters ranked three different candidates, 13% of Oakland voters ranked two different candidates and 15% of Oakland voters ranked one candidate By comparison, in San Francisco’s first RCV election, only 59% of voters ranked three different candidates. Even among voters who did not vote for three different candidates, four in five had their ballot count in every round of counting.
Furthermore, in 2010, only 1.5% of Oakland voters either didn’t vote or cast an invalid ballot in the mayor’s race. But in the 2006 Oakland mayoral election (which used the previous non-RCV method), fully 2.9% of voters either didn’t vote or cast an invalid ballot. That means that the the “dropoff” in the 2006 mayoral election was almost twice as high as in the 2010 RCV mayoral contest.
These results show that a significant majority of Oakland voters successfully used the RCV system to express their preferences among the mayoral candidates.
How did ranked-choice voting impact participation in the election?
Far more Oakland voters participated in electing the mayor in 2010 than in 2006. By the time all of the ballots have been counted, approximately 106,000 Oakland residents will have voted in the mayoral election. This compares to only 83,891 voters in the June 2006 election won by outgoing mayor Ron Dellums. Thus, 27% more Oakland voters have participated in this year's mayoral election, the first in which ranked-choice voting has been used, compared to the previous mayoral election.
Why is it taking so long to determine the winner of the Oakland mayoral election?
In Oakland, a large number of voters chose to vote by absentee ballot. Because each absentee ballot must be opened and verified by hand, it takes a long time to process them. Provisional ballots also take a long time to process because each one must be researched to ensure that the voter is eligible to vote.
This slow process is unrelated to RCV. After all, the statewide contest for Attorney General also is taking a long time to determine the winner due to the closeness of the contest and the need to process so many absentee and provisional ballots. For the same reason, in 2006 it took two weeks before Oakland knew if Ron Dellums had won the non-RCV mayoral election. Whether for RCV or non-RCV contests, if an election is close, all the absentee and provisional ballots must be tallied before you can be sure of who the winner is.
If the election is not close, then preliminary results provide enough of a snapshot to know with a high degree of certainty which candidate is going to win. For example, RCV also was used to decide three Oakland city council seats, three board of education seats and the city auditor, and because those races were not close we have known since election night who the winners will be. Only the mayor's race is still unknown due to the closeness of the contest and the slowness of processing absentee and provisional ballots -- not due to RCV.
How long does it take to run the RCV tally?
Running the RCV tally is easy and takes only about five minutes per race. It's as simple as walking to your laptop computer, hitting a few keystrokes on your keyboard giving the command to run the tally, and two minutes later you have the results for 100,000 ballots. It is really fast. Running the tally automatically generates the round by round vote total reports like the type you can see at http://www.acgov.org/rov/rcv/results/rcvresults_2984.htm. In addition, it takes no additional time to scan 2nd and 3rd choice rankings on voters’ ballots because those rankings are automatically scanned into the system at the same time as voters’ 1st choices.
If the tabulation is so fast, then why the delay in reporting results?
Besides the slowness of processing absentee and provisional ballots, the registrar of voters has made a policy decision to not run the RCV tally every day. Because the ranked-choice tabulation process is nearly instantaneous, it can be run and re-run as often as officials choose. The registrar of voters could have released preliminary results on election night and every day since, showing updated preliminary results as votes continued to be counted. All of the "waiting" is the result of the registrar’s policy, not due to ranked-choice voting.
But don’t you need to have all ballots processed before you can run the tally?
No, not at all. At any point, you can run the tally on the ballots that have been processed up to that moment, and any results are by definition "preliminary." But that is no different than non-RCV races, such as in the Attorney General race where Harris and Cooley are going back and forth as they process the absentee and provisional ballots. Yet that's not seen as a reason not to release preliminary results in the attorney general race.
RCV is no different, except that you must run the tally in order to give the true preliminary results. Just giving more first rankings gives a misleading perception; in the mayor's race, it caused the public and the media to think that candidate Don Perata had a 10% lead for three days when in fact his lead opponent Jean Quan was picking up far more runoff rankings from the supporters of other candidates. If the registrar had run the tabulation on election night, the public would have known immediately that this was a close race. In order to truly give "preliminary results" it is necessary to run the tally, which is easy and quick to do.
How could a candidate with the most 1st choice rankings lose to another candidate?
To win a ranked-choice voting contest, the winning candidate must have both a strong core of support as well as a broad base of support. Polarizing candidates who have a strong core of support but lack a broad base typically start out as front runners but end up losing to candidates who have a broader base. For example, Don Perata had 35% of first rankings but that was not close to a majority and meant that 65% of voters preferred another candidate. Jean Quan picked up a significant more number of runoff rankings from the supporters of other candidates and was able to take the lead. With RCV, the winner is the candidate who has a broad base and is preferred by the most voters in that election.