by Aaron Klemz
Feb 5, 2013, 7:00 AM

Second choice: RCV lessons from 2009

Note: Gary Schiff stopped by Drinking Liberally last week and visited with the people in attendance, including Jon Oulman, the owner of the 331 Club, where DL meets. It is unknown if Gary Schiff is Jon’s second choice, or first choice, or whatever. He just happened to be standing next to Gary when the camera went off. [Steve]

The race for Minneapolis mayor is already crowded, and getting bigger every day. Schiff, Winton, Hodges, Samuels, Cherryhomes, the list goes on and just keeps growing. One question that we don’t have much experience with is what impact ranked choice voting will have on a competitive mayoral race.

2009 was the first year that Minneapolis used ranked choice voting (RCV). RCV lets voters choose a first, second and third choice for each office. In single-seat races, RCV tabulation techniques set an election threshold of 50% + 1 vote and first looks at first choice votes to see if any candidate exceeded the threshold on the basis of those votes alone. If no candidate meets the threshold, the candidate who received the lowest number of first choice votes is eliminated and their second choice vote is then transferred to that candidate. This process continues until a candidate reaches the threshold. If a voter ranks only one candidate and that candidate is eliminated, then the ballot is considered “exhausted.” If only two candidates remain and neither meets the threshold, the candidate with the higher vote total wins.

There were several races in 2009, but few close single seat races. The closest Minneapolis election in 2009 was the Minneapolis Park Board District 5 race between incumbent Carol Kummer and challenger Jason Stone. Stone had challenged Kummer in 2005, and the 2009 race was a rematch. Despite the change in voting procedure, Kummer won by a narrow margin similar to her victory margin in 2005. In the 2009 race, a third competitive candidate, Steve Barland, entered the race. In addition there were two other candidates who did little campaigning. This race represents the best example we have so far of the unique dynamics of an RCV election in Minneapolis.

9,382 voters cast ballots in the precincts of District 5 out of 42,413 registered voters, for a turnout percentage of 21.6%. This is slightly higher than the citywide turnout of 20%. However, we cannot credit interest in this race for bringing more people to the polls, since there was a substantial undervote, where voters did not cast a vote in the MPRB District 5 race. Of 9,382 voters, 1,534 cast no vote in the District 5 race, an undervote percentage of 16.4%. This seemed quite high, so I wanted to see if RCV was partially responsible for this by comparing the undervote to the 2005 race. In fact, the undervote in 2005 was slightly higher (16.8%), so we can probably attribute the undervote to a lack of voter interest and knowledge about this race.

Candidate 1st Choice Votes Reallocated 2nd/3rd Choice Votes Final Total
Kummer 2964 (37.8%) +656 3620 (46.13%)
Stone 2788 (35.5%) +483 3271 (41.68%)
Barland 1541 (19.6%) -1541
Peterson 369 (4.7%) -369
Looney 151 (1.9%) -151
Exhausted Ballots 957 957 (12.19%)

In a race that was decided by a very thin margin in 2005, the presence of at least one more credible candidate and the possibility that second choices might influence the final decision leads to fascinating and unexplored areas of campaign strategy. Stone professed to not worrying too much about people’s second choices, since he believed that he and Barland appeal to an anti-incumbent feeling, and they would likely receive second choice votes from each other’s supporters.

In some ways, this race represented a paradigm case for RCV, with three viable candidates. However, the results in this race demonstrate that a few “truths” about RCV are questionable, and highlights a couple of opportunities that are unique to candidates in RCV elections.

Analysis of Results

In my opinion, Kummer adapted better to the change in the voting system and was able to prevent second choice votes from tipping the election to Stone. Given the close result, this race could have been decided by second (and third) choice votes, but it seems that many people did not choose to rank beyond the first choice. There were 2,096 votes cast for candidates who were eliminated in the second round, and of those 957 were exhausted after the reallocation (45.6%). This also debunks the notion that RCV results in a majority result for the winner. In this case, the high number of exhausted ballots meant that the winner won with a plurality of the vote. While it is possible that some of these ballots ranked two or three candidates who all were eliminated, it is likely that most of these exhausted ballots ranked only their first choice. We have no way of knowing whether this was true of ballots counted for Stone and Kummer, but it seems likely that a similar trend held true for their supporters as well.

Some of Kummer’s late literature used comparisons that attempted to demonstrate the similarity between Kummer and Barland. For example, a direct mail piece used a series of columns that looked like this:

Barland Kummer Stone
Supports an Independent Park Board X X ?
Supports full funding for sports fields X X ?

This is a clear attempt on the part of Kummer to sway Barland supporters to rank Kummer as a second choice. For much of the race, all candidates seemed to operate as though second choice votes didn’t exist. All of the literature and websites of the candidates exhorted voters to “rank me #1” or “make me your first choice.” It’s true that it is awkward to ask voters to make you their second choice, but in this case second choice votes really determined the result of the election. In fact when asked by a reporter who their second choice was, only Stone would offer a second choice, Barland.

One of the core arguments of RCV supporters is that RCV discourages negative campaigning, as negative campaigning risks alienating voters who will not rank you a second or third choice. However, this race demonstrates that aggressive, even negative pieces can be used effectively. In this case, Kummer correctly identified Barland as the third place candidate who was going to attract a significant number of votes. Therefore, identifying the concordance between her positions and Barland’s presented no risk to her first choice vote total. It was unlikely that she would cause voters to rank Barland first and Kummer second. And if she was correct that Barland would finish third, she’d get those votes back when the second choices were reallocated.

Stone believed that Barland’s supporters would be anti-incumbent votes, and that he would naturally get a majority of those votes. As it turns out, Stone finished third in second choice votes, behind “nobody”/exhausted and Kummer. He could have explicitly linked his campaign to Barland’s the way that Kummer did, particularly attending to whatever anti-incumbent / anti-status quo positions that he took. Stone also might have geographically targeted his second choice messages in the areas where Barland was best known and likely to draw the most votes. Given that Barland was widely considered to be running in third, there’s little downside to this strategy. It wouldn’t make much sense to attempt to get Kummer voters to rank Stone second, since there was a lot of enmity between them in 2005.

My argument is that all campaigns in RCV races should create an explicit strategy to maximize their second choice votes among trailing candidates. Kummer’s choice to frame Stone as “questionable” on issues that appealed to Barland’s voters may well have made the difference in this race. At the least, it prevented Stone from overtaking Kummer after the reallocation of second and third choices.

In the Minneapolis mayor race, it will be the second and third choice votes that will decide the election. Given the breadth of the field and their bases of support (ideological, geographic and otherwise) each campaign needs an effective strategy to maximize these votes from the supporters of other candidates. In future installments, we’ll look at individual candidates and what those strategies might look like.

Thanks for your feedback. If we like what you have to say, it may appear in a future post of reader reactions.