Best methods for electing the SF Board of Supervisors — summary of the Rose Institute report

This is a summary of the best methods for electing the SF Board of Supervisors from the report “Re-Assessing San Francisco’s Government Design” by the Rose Institute of State and Local Government and commissioned by TogetherSF

Over the past fifty years, the City of San Francisco has alternated back-and-forth between electing Supervisors “by district” or from the City “at-large,” that is city-wide. District-based systems prioritize representation of neighborhoods and communities, while at-large systems prioritize citywide needs. San Franciscans have long struggled to balance those competing values. The report by the Rose Institute of State and Local Government emphasizes that any system of representative elections involves trade-offs, and analyzes the nature of those trade-offs. The report recommends that San Francisco should consider a “mixed” system, a hybrid of districts and at-large, with the at-large supervisors elected using a specific method known as Proportional Ranked Choice Voting (PRCV).

Below are excerpts from the Rose Institute report which summarize the pros and cons of districts, at-large, mixed and Proportional Ranked Choice Voting methods of electing the Board of Supervisors.

At-large/citywide elections – used in San Francisco from 1980 thru 2000


  • “At-large members were much more likely to report focusing on the city as a whole than district members”
  • “there is some evidence that at-large elections are more competitive than district elections, in part because they attract far more candidates”
  • “would likely elect Supervisors to the Board who have a more citywide perspective”
  • “At-large election systems do not have districts, so are not susceptible to gerrymandering”
  • “at-large cities are more likely to permit the construction of multi-family housing”
  • “Asians are almost perfectly represented amongst San Francisco’s at-large elected officeholders”
  • “Women…were far more successful winning representation using at-large elections than they have been using district-based elections”


  • Unrepresentative election results: “plurality at-large election systems are at a greater risk of electing governing boards where racial minorities are under-represented”
  • “because a cohesively voting majority can elect every seat under this system, [plurality voting] carries a greater risk of under-representing political or racial minorities on the Board”…even “effectively preventing a minority faction from winning any representation”
  • “at-large elections generally produce far worse geographic or neighborhood representation than by district elections”
  • “campaign expenditures are frequently higher in at-large elections”
  • “experts have found that it is common for many or most of the city councilors elected at-large to live in one area of the city—typically a middle-class white area”
  • “at-large cities are less likely to equitably distribute the burden of [housing] construction… permitted 36% more multifamily housing in minority neighborhoods than white neighborhoods”
  • “an overemphasis on Downtown and business interest perspectives”
  • “local at-large election systems have a far greater risk of being struck down under federal or state voting rights laws”
  • “Few of the stakeholders interviewed for this report who were critical of district elections for the Board wanted a return to a purely at-large system.”

District elections – used in San Francisco from 2000 thru today


  • “Neighborhood representation…is clearly enhanced by district elections”…Prioritizes representation of local neighborhoods and communities.”
  • “Numerous studies have found that transitioning from at-large to by district elections increases the representation of racial minorities, provided that the minority population is large enough and meets a certain level of geographical compactness.”
  • “Reduce the costs of campaigning” because “in district based elections…the electorate may be small enough for a door-to-door canvassing to reach most voters”


  • Unrepresentative election results: “In San Francisco, it is not clear that districts would necessarily provide more diverse representation”…”Whites are significantly overrepresented on the current Board, whereas Asians, the next largest racial group…are the most underrepresented”
  • “district elections are uniquely susceptible to gerrymandering” and “San Francisco’s process, in particular, has been the repeated subject of political controversy and accusations of gerrymandering”
  • “creates weaker incentives for individual supervisors to consider the needs of the City as a whole” and “Incentivizes supervisors to focus on issues of parochial concern to their districts” and leads to “narrowly-focused ward politics”
  • “limits voters to electing one Supervisor instead of 11”
  • “city councils whose members are elected by district report somewhat more council conflict than in cities where all council members are elected at large.”
  • “districts are less competitive than at-large elections, in part because they attract far fewer candidates.
  • “a majority of the former Supervisors consulted for this report, expressed the view that a different electoral system might better serve the City’s needs.”

Mixed system (combination of districts and at-large)


  • “San Francisco needs a better blend of community and citywide oriented Supervisors to meet the City’s significant challenges. In theory, mixed election systems…should provide both perspectives”
  • “proponents of a mixed system argue that it combines the best attributes of both systems: district members ensure no part of the city is neglected… at-large members ensure that the most pressing needs of the city are also addressed”
  • “because mixed systems generally result in better descriptive representation than pure at-large systems, they are less vulnerable to a voting rights challenge under the CVRA;
  • “adding an at-large element to San Francisco’s election system would likely make Supervisors more resistant to NIMBYism and may better align Supervisors’ incentives with the needs of a City in transition”
  • “some studies suggest that mixed cities produce policy outcomes somewhere between purely at-large and purely by district cities, suggesting a balancing effect is taking place”
  • “another study, looking at housing production, found that mixed cities produced more housing than by district cities”
  • “one study found that cities were more likely to adopt smart growth policies as the percentage of at-large seats increased on a city council”


  • “this system is likely to produce less neighborhood diversity than a purely by district system”
  • “racial minorities in mixed systems are more likely to be elected in the district seats than the at-large seats”
  • “candidate campaign costs in mixed systems also fall in between the two pure systems: more expensive on average than purely by district elections, but less expensive than purely at-large systems”
  • “if San Francisco kept an 11-member Board but reduced the number of district seats to add at-large seats, the remaining districts would grow in size, which would likely also increase campaign costs for those seats”
  • “because mixed systems have districts, the potential for gerrymandering would remain; bloc voting in the at-large seats could also create the risk of over-representing the political majority as to those seats”
  • “mixed systems are not immune from suit under the CVRA, but are likely to be less vulnerable than purely at-large systems”
The report recommends that San Francisco consider using Proportional Ranked Choice Voting for the at-large part of the mixed system.

Proportional ranked choice voting (at-large version of ranked choice voting)


  • “PRCV generally leads to improved minority representation over plurality at-large and single-member districts in racially integrated cities”
  • “in cities where the minority community is geographically dispersed, PRCV may be the strongest option for providing diverse representation, and has been used as a voting rights remedy in several cities”
  • “PRCV system is likely to elect Supervisors who are more ideologically and demographically representative of the electorate than a plurality at-large system”
  • “PRCV’s chief benefit is that it produces representation in fairly close proportion to how the electorate is voting. As a result, this system is more likely to result in a demographically diverse Board than plurality voting, which also lessens this system’s legal risk”
  • “since PRCV is an at-large election system, PRCV Supervisors may be likely to have a more citywide perspective than district Supervisors”
  • “PRCV is designed to maximize the effectiveness of a person’s vote, so that as many voters as possible contribute to electing a representative of their choice.,, typically closer to 80% to 90% of voters will have their vote count toward a candidate ranked on their ballot.”
  • “in PRCV at-large, there are no districts to gerrymander”
  • “competitiveness increases under PRCV”
  • “constituencies that could elect someone “include neighborhoods, ethnic groups, or communities that are not based on a place or demographic identity, like small business owners, renters, or bicyclists”
  • Neighborhood representation: “Candidate slates under PRCV may recruit candidates from different areas of a jurisdiction and encourage those candidates to focus their campaigning there”…“if candidates adopt a neighborhood campaigning strategy, elected officials may have a closer focus on the region where they draw most of their support”
  • “PRCV may improve the chances of LGBTQ+ candidates in San Francisco where, outside a few neighborhoods, the LGBTQ+ community is relatively dispersed. There is some evidence that women…who are also geographically dispersed, may benefit”
  • “competitiveness increases under PRCV [because] candidates seem to be more likely to campaign as part of slates, giving voters a clearer picture of the policy cleavages between candidates”
  • “One study of the municipal use of PRCV in five Ohio cities found improved consensus building was observed”…electoral pressure on PRCV candidates in Cincinnati “to appeal for second- and subsequent-choice votes from supporters of other candidates appeared to reduce tension on the council.”
  • “PRCV is sometimes described as a candidate-based proportional representation system, where a majority of the electorate will elect a majority of seats but a substantial minority…will also win representation, in proportion to their share of the vote”
  • “San Francisco is likely better positioned than most jurisdictions to adopt PRCV, since a PRCV ballot is identical to a single-seat RCV ballot” and “in a PRCV election, voters are able to rank the candidates on their ballot in order of preference, just like in a single-seat RCV election”
  • “because candidates are able to win election by targeting a much smaller electorate, this might reduce the cost of campaigning”
  • “existing research on single-seat RCV suggests that most voters are able to cast a ranked ballot without difficulty… voters are perfectly capable of ranking” and the “overwhelming majority of voters…say they understand how it works.”
  • “a later study… found that voters reported similar levels of understanding RCV (87%), plurality elections (86%), and state top-two elections (89%)” and “an academic study found that the overvote rate in RCV and non-RCV races were comparable.”


  • “[a] scholar… found that PRCV tended to undermine political party (or faction) discipline among elected council members and as a result made governing coalitions less stable” and “argues that these early PRCV systems were short-lived because they undermined party cohesion”
  • “PRCV may not be better for minority communities where there is a high enough degree of residential segregation that majority minority districts can be drawn”
  • “it is possible that the voter error rate in a PRCV election will be higher than in a single-seat RCV election, especially since there will be more candidates in an average PRCV election than the average single-seat RCV election”
  • “PRCV elections are still at-large elections; thus, …there would still be a risk of the system being challenged and overturned under… the state CVRA. However, because the election threshold is lowered in proportion to the number of seats to be elected, PRCV is far less likely to dilute the voting strength of communities of color in jurisdictions with racially-polarized voting”
  • “the campaign finance effects of PRCV have also not been well studied. In theory, PRCV campaigns should be less expensive than traditional at-large races because the vote threshold to get elected is far less”
  • “the policy incentives under this system, and whether or how these incentives differ from traditional at-large elections, has not been well studied”

Conclusion of the report:

  • The report recommends on page 65 that San Francisco should consider a “mixed” system of both districts and at-large, “harmonizing neighborhood representation and a greater focus on the needs of the City as a whole;” in addition, the report recommends that San Francisco should consider electing the at-large supervisors by using Proportional Ranked Choice Voting because “PRCV is likely to elect a more representative body” that balances the needs of neighborhoods with citywide interests.

“After careful review, we believe that a mixed system for electing the Board of Supervisors that harmonizes local representation and citywide perspectives… as a way to balance the benefits and costs of at-large and district election systems…could serve San Francisco well and deserves serious consideration…However, we do not recommend that the City revert to fully at-large elections….mixed systems provide governing boards with a more balanced policy perspective that accounts for both neighborhood and citywide needs…The report recommends San Francisco consider an alternative approach to electing at-large seats called Proportional Ranked Choice Voting (PRCV). San Francisco was the first major city to use RCV in modern times and could be among the first to use PRCV.”