San Francisco and Ranked Choice Voting: New Data is Asking, Not Answering Questions


[Image courtesy of KQED]

UPDATE: A commenter below passed along the link to the USF study mentioned in the article … it can be found here.

Ranked choice voting (RCV) continues to be a hot topic in San Francisco. A new report from researchers at the University of San Francisco suggests that only one-third of the city’s voters completely filled out their ballots in the November 2011 election for mayor, sheriff and district attorney.

Opponents of RCV have seized upon the data to suggest that the city should go back to more traditional methods of voting, and charge that the low rate of completion suggests the $300,000 spent on an RCV public education campaign was insufficient and/or wasted.

I haven’t seen the full report yet, but I’m pretty confident that the data contained therein shouldn’t be seen as a definitive indictment of RCV; rather, the data should tee up questions that election officials, policymakers and researchers alike should seek to answer before reaching any conclusions about RCV – either in San Francisco or in general:

  • + To what extent does San Francisco’s RCV data correspond to successful adoption of RCV in other cities?
  • + Change is always hard; is the San Francisco data analogous to similar data from jurisdictions adopting new technologies or rules for voting?
  • + Does the low rate of ballot completion in certain minority neighborhoods – as low as 9% in Chinatown and Bayview – reflect confusion about the process or a strategic choice to reject all but voters’ first choice?
  • + Does San Francisco’s policy of limiting voters to three choices – and thus discarding ballots of voters if all three choices have been eliminated – disproportionately affect certain voters?
  • + And finally – did the public education campaign help at all, and was its effect enhanced or muted in different areas of the city?

San Francisco’s debate about RCV is being repeated elsewhere in California and nationwide, and it’s important to use research to answer questions about its performance for voters. Data like the USF study is tremendously helpful (and long overdue for such a hot topic) but those of us who care about elections should be careful not to allow a small handful of data to determine policy – or worse, to use a small sample to reinforce policy preferences or prejudices.

In other words, let’s not jump to conclusions on RCV – or anything else in the election sphere – until we’ve had an opportunity to collect a little more data.

5 Comments on "San Francisco and Ranked Choice Voting: New Data is Asking, Not Answering Questions"

  1. Thanks for your thoughtful comments here, Doug.

    There indeed is much to review here. Proponents of ranked choice voting in San Francisco have assembled a variety of information that provides a fuller picture of what’s going on there than some of what’s been publicized that includes:

    * a large increase in elected people of color, seemingly tied to less polarizing campaign tactics and more guaranteed turnout levels in decisive elections

    * much higher turnout in final rounds of elections in most RCV races than most runoff races (indeed, a majority of runoffs in San Francisco in 2000-2002 were won by a candidate who had fewer votes than the first round-leader in November, which is a common pattern in federal primary runoffs as well)

    * summaries of the only exit poll surveys done in San Francisco (in 2004-2005) that show overwhelming support for RCV and capacity to use the system.

    * evidence that undervotes (people skipping the race) have declined by significant numbers since adoption of RCV

    Relating to the data you cite, one key aspect of it is that more than 90% of voters in San Francisco ranked more than one candidate in at least one race. In the mayoral race, about 99.6% of voters cast valid ballots. 19 in 20 didn’t do anything “quirky” with their ballot, and of that remaining small group, half had their ballot count in each round of counting (for instance, someone who ranked mayor Ed Lee first, second and third).

    More broadly, we had six cities using RCV elections these years. In all of them, election officials stepped up and did a great job. In Portland (ME), say, voters handled their new system exceptionally well — more than 99.8% valid ballots in a 15-candidate race for mayor — without the city spending anything on voter education. It shows that getting a good ballot design with clear instructions is the single most important thing to do.

    Finally, we need to have a national conversation about how to handle elections with more than two choices. Having a reactive one every time a third party candidate runs for presidentt or some other high office is not productive — but get ready for it again next year, if Americans Elect indeed ends up on the ballot in all 50 states and a strong candidate earns its nomination. Already it’s looking like former New Mexico governor Gary Johnson may well be on all or nearly all state ballots as a Libertarian.

  2. The negative characterization of a third of voters indicated three preferences for all three offices is surprising. The Sheriff’s race had three dominant candidates (Mirkarimi, Miyamoto, Cunnie) and one minor candidate (Wong). In a three-way race, voters have little reason to rank all three of the candidates. The third rank is irrelevant as the race will be decided before the vote would transfer to that rank. It makes more sense to rank two candidates (unless one supports the minor candidate, which few do by definition).

    Similarly, the District Attorney’s race had three or four dominant candidates and one or two minor candidates depending upon how one looks at it.

    In the Mayor’s race, the almost third of voters who ranked Ed Lee first had little incentive to rank other candidates because of the widely reported pre-election polling results showing Lee far in the lead.

    In light of the actual dynamics of the contests, the finding that a third of voters used all three ranks in all three races is a testament to RCV and RCV education rather than an indictment.

  3. There are indeed many questions to ask and it will take more than one election and more than one study to ask them, let alone answer them.

    By the way, the actual preliminary report from the USF team (from usfca dot edu slash centers slash mccarthy slash research ) is much more accurate than the uncharacteristic, tabloid-style journalism of the referenced New York Times article. It is not incorrect to mark only one choice, the target of the education campaign was not voters who marked less than three choices, and ballots are exhausted, not discarded, during an RCV tabulation. Exhausted ballots are similar in nature to the typically more prevalent wasted votes in a plurality election. In a city that has dealt with stolen cast ballots and allegations of ballot boxes found floating in the bay, the misapplied terminology of “discarded ballots” should be acknowledged for what it is: carefully chosen, politically motivated, inflammatory language.

    Certainly there are opportunities to improve voter understanding and use of ranked choice voting. Good research can help identify those opportunities as well as what are effective remedial measures.

  4. Thanks for the link to the USF study … will add it to the post!

  5. Steve Rosenfeld | December 27, 2011 at 7:09 pm | Reply

    As a 20-year San Francisco resident who always votes, I have a very different view that election scholars (and certainly Rob Richie) may frown upon. The reason so many ranked choices ballot slots are not filled in is because nobody knows who these other choices are. We barely know who we are voting for — apart from the top of the tickets. We get city ballot mailings with candidate statements. Sometimes they are not very elucidating. We have lots of races and ballot initiatives, and I strongly suspect people are too busy to sit down and figure out there top three picks. The campaign consultants send slate mailers, which people take to the polls. A few say put me in the top-two slot. RCV has made a complex ballot more complex not simpler; hence the blanks… Maybe over time people will get used to thinking in terms of their top three choices… But I have never heard a conversation in a coffee shop or anywhere like that just yet… My unscholarly two cents…

Leave a comment

Your email address will not be published.