News / Articles

By Celeste Landry in consultation with Pat & Frank Venturo and Gaythia Weis 

This past June 22, New York City residents used IRV to rank their party’s primary candidates for mayor. IRV works this way: rank your favorite #1, your second-favorite #2 and so on.  If no candidate gets a majority of #1 rankings (aka votes) in the first round, then the candidates with the fewest votes are eliminated sequentially until only two candidates remain; the higher vote-getter of the two finalists is elected. 

The NYC Republican primary had only 2 declared candidates, so the election result was identical to a plurality result, needing only 1 round to elect a winner.

The NYC Democratic primary had 13 declared candidates. It took 8 rounds to elect a winner. The NYC ballot only allowed a voter five rankings so no voter could rank all thirteen candidates. This limited a voter’s expressiveness (though nowhere near as much as “choose one” plurality).  

The following table shows the tabulation for each round. Below the table you’ll find an explanation of what happened. If you don’t want to get into these weeds, you can skip to the end of the article for a comparison of the NYC election with the upcoming Boulder Mayor’s election and lessons learned from New York. Click here for a full sized PDF of the table.

Explanation of the Democratic Primary Results 

At the end of Round 1, the undeclared write-in candidates were eliminated due to extremely low vote totals, entered as 0% in the table.  At the end of Round 2, the lowest vote-getter among the declared candidates was eliminated.  At the end of Round 3, four more candidates were eliminated simultaneously because their combined vote total was less than the number of votes for Shaun Donovan, the lowest vote-getter of the candidates named in our list.  Unfortunately, for Donovan, he did not get enough transfer votes from supporters of the eliminated candidates to survive the end of Round 4.  The end of Round 5 saw three more candidates eliminated simultaneously because together they had only 12% of the vote while Andrew Yang, the lowest vote-getter of the leading-pack candidates, had 13% of the vote.  When Yang was eliminated at the end of Round 6, the transfer of his supporters’ votes to Kathryn Garcia allowed her to leapfrog ahead of Maya Wiley.  Given that Yang and Garcia campaigned together toward the end of the campaign, the large transfer from Yang to Garcia is not surprising.  

The last three candidates in the race, Eric Adams, Kathryn Garcia and Maya Wiley, were viewed respectively as conservative, centrist and progressive.  When Wiley was eliminated at the end of Round 7, many of her supporters “exhausted” their ballots by not ranking either Adams or Garcia on their ballot.  (Ballot exhaustion occurs when a ballot is no longer countable in a tally as all of the candidates marked on the ballot are no longer in the contest.) Wiley’s supporters’ exhausted ballots accounted for almost 8% of all the ballots cast in the election.  In all, about 15% of NYC voters exhausted their ballots and didn’t have a say in the final round of tabulation.

We can probably assume that most of the Wiley “exhausted-ballot” voters would have preferred Garcia over Adams and would have tipped the scale to Garcia if only they had ranked Garcia on their ballot.  [It’s also possible other considerations besides political leanings, such as race or gender, would have been factors.  For instance, Adams and Wiley are both Black.  Garcia and Wiley are female, and NYC has never had a female mayor.] 

Some of the media attention prior to the election included a fair amount of discussion on how voters could strategically rank their ballot. An opinion piece by Rob Richie of FairVote, a well-known national organization which advocates for all forms of RCV, gave this advice: “First, because New York limits rankings to five, the best way to guarantee your ballot counts in every round of the tally is to include at least four of the five most viable candidates.” If Wiley’s supporters had followed Richie’s advice, then Garcia would probably be the Democratic mayoral nominee and presumptive winner of the November election."

Comparisons to Boulder’s Mayoral race in 2023

According to reports, NYC Democratic voters seem pleased with IRV, particularly being able to provide more information on the ballot.  In 2023 the city of Boulder will start using IRV to elect its mayor whenever there are 3 or more candidates.  However, some significant differences exist between the NYC and Boulder mayors: 

• NYC’s mayor is a “strong” mayor with substantial power.  The Boulder mayor is a “weak” mayor with a vote equal to the other 8 council members.
• NYC’s mayor serves a 4-year term.  Boulder mayors, now and under IRV, serve a 2-year term.
• NYC is a large city and had 13 candidates with at least 4 viable candidates.  Boulder is a much smaller city and is not likely to have nearly as competitive an election.  The more competitive the election, the more likely the “instant runoff” will be triggered. 
• NYC has been electing its mayor for a long time.  Boulder voters have never directly elected their mayor before, instead electing council members who choose one of their own to serve as mayor. 

   

  Lessons Learned from the NYC primary election

   

• Instant-runoff voting means that the voting process is complete once the ballot is cast – instead of having a later runoff.  Note that IRV is not called “instant-runoff tabulation.”  Because the elimination-and-transfer tabulation is done round by round, changes in one round can have a big impact on future rounds.  In plurality voting, even if not all the ballots are counted, a winner is usually declared once a candidate has an insurmountable lead, i.e., the outstanding ballots won’t change the winner.  An insurmountable lead is harder to gauge in IRV if the tabulation progresses to second, third or later runoff rounds.  
• The NYC mistake of including test ballots in the first report of IRV results did not help instill confidence in the city’s Board of Elections but should otherwise have no bearing on IRV.
• IRV does not eliminate negative campaigning, although it may reduce negative campaigning.  Adams, Yang and the Yang/Garcia joint campaign were all criticized by opponents for political, rather than policy or administrative, reasons.
• A lot of people still don’t understand IRV’s properties and that RCV is an umbrella term for several forms of ranked voting.  For example, a Reuters article   about the NYC primary stated, “Most U.S. elections are ‘winner take all’ but some major cities have gone to ranked-choice voting, which supporters argue is more democratic.” The sentence should have stated, “Most U.S. elections are plurality (or first-past-the-post) but some major cities …”
• People confuse ranking and rating.  A NY Times op-ed  reported that “A mix of Times writers and outside political experts assessed the [mayoral] contenders’ performances and ranked them on a scale of one to 10” until they were alerted that rating, not ranking, is done on a numeric scale.  This type of rating is called score voting; voters may score every candidate.  Score voting is the most expressive of the main voting methods under consideration in the US. 

We hope that you are as excited as the Voting Methods Team about all the attention on better voting methods generated by the NYC mayoral primary.  Please share this article with anyone who might like our analysis of the election.