Looking for the latest stories? We're now at cironline.org

Confusion about Oakland’s voting system may have affected election

Alameda County
Registrar of Voters
The ranked-choice ballot has three columns to indicate a voter's priority.

One out of every 10 Oakland voters showed signs of confusion about how to vote for mayor using the city’s new ranked-choice voting procedure, according to a computer analysis of returns obtained by California Watch.

The confusion was so great that it may have flipped the final results of the extraordinarily tight mayor’s race between former state Senate leader Don Perata and city council member Jean Quan, the analysis shows.

The analysis of voter registrar data was performed by an elections expert who is not from Alameda County and not affiliated with any of the candidates or campaigns.

The analyst asked not to be identified by name lest he be caught up in controversy over the election’s outcome.

He said that more than 9,700 of the 97,940 Oaklanders who voted in last week’s election made mistakes that reflected fundamental misunderstanding about the new system.

“There were a lot more people confused and potentially messing up their ballots than there were votes in the spread” between Perata and Quan, the expert said.

As of Monday, Quan was leading Perata by 1,876 votes, records show. Alameda County Registrar of Voters Dave Macdonald said he's unsure when the vote count will be complete, the Associated Press reported. The election was a week ago today.

Advocates say ranked-choice voting saves taxpayers the substantial cost of holding runoff elections.

The system requires voters to list, in order of preference, three candidates for each office. And in Oakland’s election, some voters struggled to figure the system out, the study shows.

The expert’s workup shows that 90 percent of the 97,940 participating Oakland voters seemingly understood how ranked-choice voting works.

Of the total, 70 percent listed three candidates in order of preference, as instructed. Another 11 percent listed only their first and second choices, and about 9 percent only marked a first choice.

After that, confusion seemed to set in.

More than 5 percent of voters marked the same candidate for their first, second and third choices, the analysis shows. But a voter can only vote for a candidate once, so for these 4,900 voters those second and third choices went uncounted.

In another sign of confusion almost 1 percent of voters – 924 – cast their third-choice vote for their first-choice candidate, meaning that their third choice went uncounted.

Another 472 voters listed two or more candidates as either their first, second or third preference, canceling out those votes as well.

The voting for mayor was at the back end of a ballot that included elections for governor, U.S. senator and a long list of other state and local candidates and measures. As perhaps another sign of confusion about ranked-choice voting, 1,304 voters skipped the mayor’s vote entirely, the study shows.

In 2004, San Francisco* became the first California city to use ranked-choice voting. This year it was used for the first time in Oakland, Berkeley and San Leandro.

Here are results of the analysis:

 

# Ballots

% Ballots

 

97970

100.00%

 
     

41

0.04%

Two or more votes in 1st, 2nd, and 3rd

1

0.00%

Two or more votes in 1st, two or more votes in 2nd, skipped 3rd

18

0.02%

Two or more votes in 1st, 2nd, and vote in 3rd

2

0.00%

Two or more votes in 1st, skipped 2nd, two or more votes in 3rd

15

0.02%

Two or move votes in 1st, skipped 2nd, skipped 3rd

5

0.01%

Two or more votes in 1st, skipped 2nd, vote in 3rd

63

0.06%

Two or more votes in 1st, vote in 2nd, vote in 3rd

7

0.01%

Two or more votes in 1st, vote in 2nd, skipped 3rd

9

0.01%

Two or more votes in 1st, vote in 2nd, same vote as 2nd in 3rd

3

0.00%

Two or more votes in 1st, vote in 2nd, two or more votes in 3rd

1

0.00%

Skipped 1st, two or more votes in 2nd, skipped 3rd

1304

1.33%

Skipped 1st, 2nd, and 3rd

33

0.03%

Skipped 1st, skipped 2nd, vote in 3rd

45

0.05%

Skipped 1st, vote in 2nd, vote in 3rd

59

0.06%

Skipped 1st, vote in 2nd, skipped 3rd

16

0.02%

Skipped 1st, vote in 2nd, same vote in 3rd as 2nd

63

0.06%

Vote in 1st, two or more votes in 2nd, two or more votes in 3rd

99

0.10%

Vote in 1st, two or more votes in 2nd, vote in 3rd

8700

8.88%

Vote in 1st, skipped 2nd, skipped 3rd

145

0.15%

Vote in 1st, skipped 2nd, vote in 3rd

281

0.29%

Vote in 1st, same vote in 2nd as 1st, skipped 3rd

4953

5.06%

Vote in 1st, same vote in 2nd and 3rd as 1st

703

0.72%

Vote in 1st, same vote in 2nd as 1st, different vote in 3rd

69242

70.68%

Vote in 1st, vote in 2nd, vote in 3rd

10245

10.46%

Vote in 1st, vote in 2nd, skipped 3rd

921

0.94%

Vote in 1st, vote in second, same vote in 3rd as 1st

786

0.80%

Vote in 1st, vote in 2nd, same vote in 3rd as 2nd

103

0.11%

Vote in 1st, vote in 2nd, two or more votes in 3rd

14

0.01%

Vote in 1st, two or more votes in 2nd, skipped 3rd

13

0.01%

Vote in 1st, two or more votes in 2nd, same vote in 3rd as 1st

64

0.07%

Vote in 1st, skipped 2nd, same vote in 3rd as 1st

16

0.02%

Vote in 1st, same vote in 2nd as 1st, two or more votes in  third

* This corrects a typographical error.

Update: Rob Richie, executive director of FairVote, a Maryland nonprofit that promotes ranked choice voting, disputes many aspects of the preceding analysis: among other things, he contends that undervotes in the mayor’s race (i.e. skipping the race entirely) were actually higher in 2006, when Mayor Dellums was elected, than they were this year. That undercuts the notion that undervotes this year reflected confusion with ranked choice voting, he argues. More of his critique is posted in comments.

 

Comments

Comments are closed for this story.
judy1939's picture
Mr. Williams- It's interesting that you are choosing to highlight ranked choice voting on your blog. I am very curious, however, about the source of the "study" you base your article on: you do not identify the name of the study, the organization or university, etc., which conducted the study, the size of the sample, etc., etc. Surely, MR. Williams, if you are basing this article on a reliable study done by a reputable organization they would not be adverse to being identified with their findings and you would not be adverse to identifying your source. Since you have kept all this a mystery, I can only assume that your source is not reliable enough to see the light of day. I am sorry that your standards for your self as a journalist do not require you to go to more reliable studies which truly warrant the attention of the public. Please try again--and better--next time. Judy Cox
Lance Williams's picture

Judy -- The study is based on data from the registrar. There is no "sample," as you put it -- it's a study of all the ballots cast in the Oakland mayor's race.
I posted the study itself, so readers can see exactly where the conclusions come from and how they were derived.
L

mfraser's picture
Mr Williams - you post this as a joke, correct? I checked the calendar and since it's not April 1, I'm confused! The 150 or so people who voted twice for 1st, 2nd and or 3rd did show some confusion, but all your other points are completely contrived and you present no real evidence of ANY confusion by any of the other voters. If I'm right, how does an error rate of 150 out of 100,000 compare with other elections using traditional head to head competition with only one vote allowed? Similarly, 1.3% don't vote for Mayor; well, how does that compare with other prior elections? Less, equal, more? Please, don't present this data as evidence against ranked choice voting when in truth it is nothing of the sort, and perhaps is evidence strongly in SUPPORT of the system, which we don't know absent any comparison with other balloting systems. Really, do I have to point out the many problems in the Florida vote and throughout the country with the 2000 Presidential election? Surely this is a lower error rate than that traditional head to head, one vote only election. Obviously, marking the same person 1, 2 and 3 is functionally the same as marking just one person in first, and leaving all others blank. There is a certain rhetorical underscoring which people might derive some small satisfaction or amusement from in listing a person 1, 2 and 3, or other motivations as well I couldn't pretend to exhaustively be able to list. They might as well write an exclamation mark after their #1 choice, and I expect that is ALL they were trying to do, in effect. Don't you think most or all of these people would have just listed one candidate if in some way doing what they did invalidated their vote? Do you think they thought they HAD to vote the same name three times, but desperately wanted to actually list others? Have you ANY evidence for that strained assumption? You acknowledge almost 10% of voters declined to list a second and third choice, and about 1% declined to list a third choice. This could mean many things, but for you to conclude 'error based on misunderstanding' in these cases or in the case of listing the same name more than once demonstrates a substantial bias to try to find fault with they system, or perhaps just a deep misunderstanding of how the system works on your part, not the voters'. What about people who write in 'Mickey Mouse' on traditional ballots? Do they really think they have a chance of getting Mickey Mouse as their elected official? If you truly believe this sort of reasoning sans evidence, how about the idea that maybe voters on a traditional 2 person ballot might think they are to list the candidate they are voting AGAINST? After all, that could be easily misunderstood, couldn't it? There is a study done in a UC Berkeley Law Journal 20-something years ago which examined recidivism among those who had been executed. Guess what - that answer was 0! No crimes at all ever committed again by those who had been executed!! Perhaps your next post should be an analysis of that 'study,' as it would make a nice pairing with your 'work' on this one. To the contrary, everything you detail and present as data gives me reason to believe almost 100% fundamentally understand the system in large measure, and are simply using the ballot to express their preferences and to cause consternation with folks, like, err, you. p.s. would be nice if your comment box actually allowed us to hit return to break up the text! At least in my comment, it just comes through in one long block...
Good4Rcv's picture

The trick to authoring comments in multiple paragraphs is to click on the "Enable rich-text" link below the comment box, use basic html to do formatting, and click on the "Preview" button to verify the formatting before clicking on the "Save" button to publish your comment.

For example, the first paragraph and this paragraph look like this in the comment box:

<p>The trick to authoring comments in multiple paragraphs is to click on the &quot;Enable rich-text&quot; link below the comment box, use basic html tags to do formatting, and click on the &quot;Preview&quot; button to verify the formatting before clicking on the &quot;Save&quot; button to publish your comment.</p>

<p>For example, the first paragraph and this paragraph look like this in the comment box:</p>

paulmitchell's picture
Despite the inability to use the return key, the above commenter is dead right. Not voting down ballot is not an indictment of RCV, nor is bullet voting (only voting for your absolute choice). These are valid choices for informed voters. In fact, voting for a second or third candidate could be construed as a "confused" vote because you're potentially giving the race away to a candidate that isn't your first choice. I almost always bullet vote for school board or "vote for three" races because it is a stronger vote. Additionally, in down ticket races like State Assembly you often have 10-20% non voting. The numbers you cite are actually low.
ekbrooks's picture
The author is making incorrect assumptions about the voters. I talked to people who only voted for one person for mayor because they didn't want to give any 2nd or 3rd votes to anybody else. Some people voted for the same candidate for all three places because they did not want anybody else. Those were not errors or misunderstandings. The ones that are errors are where they voted for more than one person at rank #1, thus invalidating their vote altogether. Some people don't put anybody because they don't know anything about any of the candidates. If every one of those errors would have been votes for Perata, that still is only the current number of votes that Quan is ahead of Perata. There is no reason whatsoever to believe that all of those "lost" "error" votes were intended for Perata.
Good4Rcv's picture

Apparently Oakland voters understood RCV elections much better than some journalists and purported election experts. Even if the statistics are accurate, the problem, as Mark Twain would remind us, is in their interpretation. It is understandable why this so-called expert does not want to be named or to publish this "study". It is also unfortunate that California Watch is participating in spreading this misinformation.

Here are a few examples:

First, in any election, it is typical that some voters will mark their choices for state and federal offices but skip some or all local offices and measures. To suggest that 1,304 (1.33%) voters skipping the mayor's race is a sign of voter confusion about RCV is a gross misinterpretation of the numbers and historical patterns.

Second, it is entirely appropriate for a well-informed voter to vote for only one candidate. There are several ways that a voter can mark a ballot for only one candidate, all of them counted equivalently. There is no particular reason, in terms of impact on the election, to prefer, for example, marking just a first-choice vote for Perata and leaving the 2nd and 3rd choice blank, or marking all three choices for Perata.

An example of a serious flaw in the study, or in Mr. Williams interpretation of it, is drawing the conclusion that the 2587 voters who voted for Perata for all three choices might have changed the outcome of the election if they had either left the second and third choices blank or voted for other candidates after Perata. That conclusion is just wrong.

Third, the 473 voters who overvoted are the only ones who clearly showed signs of confusion, but not necessarily about RCV. Only 229 of those even had the potential, had their overvotes been corrected, to impact the published vote totals and are not enough at this point to change the winner. The percentage, less than half a percent, is at or near the overvote rate for old-fashioned non-RCV elections.

The results published so far are only partial; there are still a lot of ballots being counted. We don't even know who will win the mayor's race or by what margins. So it is still too early to be speculating about what might have changed the outcome.

Regardless of who wins the mayor's race, RCV has already made Oakland voters winners:

  • RCV helped offer voters a larger choice of strong, competitive candidates, avoiding back-room deals while nominations were being decided.
  • RCV encouraged greater, more inclusive participation by Oakland voters by having a single election in November.
  • RCV enabled voters to express their preferences more fully, voting three preferences rather than just one or two.
  • RCV is counting the votes more fairly and more intelligently to decide the winner with multi-round runoffs.

Certainly there are some opportunities to improve on Oakland's use of RCV for future elections, but Oakland voters and election staff have done well with RCV's introduction.

Charles's picture
"More than 5 percent of voters marked the same candidate for their first, second and third choices, the analysis shows. But a voter can only vote for a candidate once, so for these 4,900 voters those second and third choices went uncounted." “In another sign of confusion almost 1 percent of voters – 924 – cast their third-choice vote for their first-choice candidate, meaning that their third choice went uncounted.” Lance, I have to fundamentally disagree with your hypothesis that the 6% of voters in the quotes above were “confused”. In the first case, this is a viable strategy in campaigns surrounding ranked choice voting. If a candidate can keep voters from voting for his/her opposition in any way (including voting for your first choice more than once or leaving the 2nd/3rd choices blank altogether), then that candidate stands a better chance of winning or not losing through the ranked choice process. Although new, ranked choice campaign messaging has often become all about ‘vote for me and me alone’. In the second case, isn’t it possible that 1% of voters simply didn’t approve of or have a third choice? I think the opinion of a political strategist (maybe one who worked on RCV/IRV campaigns in Oakland or San Francisco) and not just the analysis of some annonymous ‘elections expert’ would have benefitted this blog.
Rob Richie's picture
This is from an email I just sent to Lance and several colleagues at California Watch: I'm afraid Lance Williams has swallowed hook, line and sinker the spin of at least one ranked choice voting opponent with his latest blogpost. I'm sorry that Lance didn't reach out to RCV defenders for an alternate perspective. But perhaps Lance already has reached his conclusions about ranked choice voting -- he did a highly negative piece before the election and did not respond to my note to him about it.... L's go some facts that should have gone with his post. 1. Lance has this seemingly damning conclusion: "More than 5 percent of voters marked the same candidate for their first, second and third choices, the analysis shows. But a voter can only vote for a candidate once, so for these 4,900 voters those second and third choices went uncounted." What Lance neglects to says that of these voters, 3,627(73.2%) voted for Perata or Quan as a first choice. That means their second and third choices would not have been counted anyway. It also means their votes counted in the final round. Indeed, of all the ballots that counted in the initial round (as 99.8% of mayoral ballots did) and yet had a potential error (overvote in 2nd and 3rd rankings, undervote, skipped rankings, duplicated rankings), three in four counted in the final round. See the attached spreadsheet (one from a longer analysis that has a lot more numbers). But Lance never reports this fact. 2. Lance suggests that all of these ballots showed confusion -- indeed, that's his lead sentence. But that's simply not true. For some voters, for example, voting for one candidate first, second and third is a way to emphasize their "bullet vote" for one person.. They won't help their first choice by doing so, but may well be saying,"This is my candidate. Period." This action can be particularly true if you believe your first choice is going to be in the final two, as Perata backers could logically predict. 3. He suggests undervotes (skipping the race entirely) are a sign of potential voter confusion. And yet he knows that there were elections for governor, U.S. Senate and statewide ballot measures that drew many voters. The great majority in Oakland voters also voted for mayor, but that doesn't mean everyone felt the knew enough about the race to do do. Even so, 27% more voters participated this year than when Ron Dellums was elected in the 2006 open seat race for mayor. But Lance says nothing about that turnout number. He says nothing about the intriguing campaign finance angles to this story. He just works with an anonymous tipster and reports that spin without talking to advocates, even though he knows where to find them. I'm not saying that Oakland RCV was an absolutely perfect election. But I will say this. This is hack journalism, seemingly unworthy of a site with your aspirations.
joel on best's picture
I think you can safely conclude that those who put more than one vote in either the 1st, 2nd, &/or 3rd choice boxes were confused. That adds up to under 500 votes. As pointed out above, the rest that the reporter says MAY have been confused votes may very well NOT have been confused - there's at least as much reason to conclude they were not confused as that they were. But that doesn't make a stirring lede, does it? And we must keep things stirred up - first rule of media in the Fox News era - mustn't we? So in comes the word "may". It's discouraging to see this coming from the center of investigative reporting. Nice photo, though.
girleypearl's picture
I think the reason this information is being blogged about is because the democratic power structure in California doesn't like RCV. I am a very liberal democrat but I detest the political power machines that use their power and money to influence local races. I do not like it that Diane Fienstein lent her name to Perata's campaign. I do not like it that Perata fought to keep RCV off the ballot and then used "confusion about RCV" as his campaign strategy. We Oaklanders voted yes for RCV in 2006. I don't care who in Sacramento or Washington doesn't like it. Please leave us alone.
jfern's picture
OK, I calculated some similar statistics here.

I didn't look at ballot order, but did look at whether there was something strange about a ballot with overvotes or multiple votes for the same candidate

0 valid votes, everything blank: 1304

0 valid votes, some overvotes: 60

1 valid vote, others blank: 8792

1 valid vote, but overvotes as well: 110

2 votes, plus an overvote: 303

2 valid votes for the same candidate, no overvotes: 361

2 different votes, no overvotes: 10435

3 votes, all the same: 4953

3 votes, 2 of them the same: 2410

3 different votes: 69242

Anyways, voting multiple times for the same candidate is equivalent to leaving your ballot blank. Leaving your 1st preference blank is equivalent to leaving your last preference blank. Now, the 473 people who had some sort of overvote may have been pretty confused.

lovelivelaugh's picture

There is no argument or evidence here that supports your claim that voter confusion on these ballots somehow "flipped the final results" of the race, as you state in the beginning but don't follow-up on.

You seem to suggest that somehow this confusion benefited Jean Quan in particular, but that makes no sense. Quan is winning because the majority of Oakland simply doesn't like Don Perata and thinks she's the better candidate.

Thus, this just sounds like part of Perata's post-election backroom campaign to de-legitimize RCV because voters used it to defeat him, not responsible journalism.

There was at least one arrogant candidate in this election who told his supporters to put him 1, 2 and 3. And there is always some small margin of ballots that have mistakes in every election.

Rob Richie's picture

Further evidence of this article's bias. In 2006, there was also an open seat race for mayor of Oakland, won with Ron Dellums getting a bare majority. It was a "vote for one" ballot in June, with a statewide primary for governor and some other statewide races --- but those statewide races were not as contentious as this year's statewide votes. And yet the dropoff in 2006 was nearly double this year.

* In 2010, 97,970 Oakland voters cast a ballot as of Friday afternoon, and 96,501 had a valid vote in the mayor's race. That means 1.5% of voters (1,469) didn't vote in the mayors race.

* in 2006, 86,379 Oakland voters cast a ballot, and 83,891 had a valid vote in the mayor's race. That means 2.9% of voters (2,488) didn't vote in the mayor's race.

But how does Lance report this fact? He writes: "The voting for mayor was at the back end of a ballot that included elections for governor, U.S. senator and a long list of other state and local candidates and measures. As perhaps another sign of confusion about ranked-choice voting, 1,304 voters skipped the mayor’s vote entirely, the study shows."

Simply appalling news judgment.

Manny's picture
Though the attention the Oakland Mayor's race in garnering is interesting, I think it is to deflect spotlight from the SF District 10 race.

21 candidate ran, because in SF, RCV has been around for a few elections and candidates have discovered it can be a crapshoot, and many people toss their hat in.

There were 15,520 valid ballots cast by voters.

In the end, Malia Cohen was declared the winner with 3,693 votes. Elections software say that's 51.51%.

In reality, that's 23.8% support of the voters. Hardly a mandate or a majority that IRV promises.

361 voters, or 2.3% made mistakes on this complex ballot, overvotes, which means they voted for two or more in a ranking. That invalidates their vote.

The difference in votes between 1st and 2nd is 216

In SF, a 2.3% overvote, iver very significant. Couple that with the winner getting 23.8% support, they've got some 'splaining to do.

Mary Mazzocco's picture
I am one of those 10,000 voters who marked a first choice and a second choice, but no third choice. I was not confused. The instructions on the sample ballot and the ballot itself were very clear. I simply didn't have a third choice. Honestly, I do not understand the media frenzy over these election results. Perata brought strong negatives into the race; you either like him or you really, really don't. He was unlikely to be the second choice of many voters. Kaplan and Quan are both popular in the city, hold similar political views, and urged their voters to choose the other as their second choice. Even had there been a traditional runoff, I would have predicted that unless Perata was unable to get a majority in the first round, either Kaplan or Quan would be the ultimate winner. IRV simply sped up the process.

via Twitter

© 2013 California Watch   /  development:  Happy Snowman Tech   /  design: