Voices & Choices

University of Colorado Student Government election highlights challenges for approval voting

University of Colorado Student Government election highlights challenges for approval voting

Editor’s note: FairVote is publishing this guest blog to share useful analysis that helps explain our support for ranked choice voting as preferred means to reform American elections. We do not endorse plurality, single-choice elections over approval voting.

On April 13, 2018, the University of Colorado student government (CUSG) held its annual Spring Election. While more than 50 colleges and universities in the U.S. use ranked choice voting (RCV) for student elections, CUSG uses a less common method known as “approval voting.” Under approval, voters don’t rank their candidates, but they may vote for as many candidates they want, with each vote carrying equal weight. The results of this recent CUSG election confirm the challenges that FairVote has previously documented with approval voting, and they help explain why electoral reformers like me continue to focus their energy on RCV.

The table below shows the results for the April 2018 CUSG Executive race. Students normally run for the executive position in slates or “tickets” of three candidates. The name of the ticket appears as a single option on the ballot, and the ticket with the most votes wins the three seats. There were five tickets running for executive this year: Ignite CU, EMPOWER, BOLD, PRIDE, and Prism. (One independent candidate was also on the ballot, which is allowed, but the CUSG election code doesn’t say how the two remaining seats should be filled if the independent wins.) Note that because voters cast different numbers of votes under approval voting, the number of votes and percent of votes is not as useful a measure as the percent of voters - which reflects the proportion of voters who "approve" of the slate.

Candidate

Votes

% Votes

% Voters

Ignite CU:

2487

29.0%

38.9%

EMPOWER:

2279

26.6%

35.7%

BOLD

2146

25.0%

33.6%

PRIDE:

707

8.2%

11.1%

Independent, Charly Mendoza

501

5.8%

7.8%

Prism

459

5.4%

7.2%

Total Votes

8579

   

Total Voters

6390

   

As detailed below, this election highlights three weaknesses of approval voting relative to RCV. And it isn’t the first such approval election to do so. In fact, every past University of Colorado election with more than two candidates demonstrated them as well. The student government at Dartmouth College experienced similar issues with approval and ended its use of it in 2016. Prior to that, the IEEE association and Dartmouth Alumni Association also abandoned approval voting on similar grounds. Let’s give these problems another look in the context of this recent CUSG election, comparing them to how ranked choice voting works in practice.

1. Approval voting does not ensure a winner with broad-based support.
Ignite CU won votes from only 39 percent of all voters, and because voters could back more than one candidate, that amounted to only 29 percent of all votes cast. By neither metric did Ignite win a majority (more than 50 percent) of the vote, and the results don’t provide any other evidence of broad-base support for the winner. The results don’t tell us whether Ignite could have beaten any of the other tickets in a head-to-head race, especially against EMPOWER and BOLD, who also earned support from more than a third of voters. In fact, Ignite could have theoretically been the last choice of 61 percent of voters. Some proponents of approval claim that it will elect “consensus” candidates, but the data from this and other real elections fail to support this assertion.

In every ranked choice voting election this year, by contrast, the winner was ranked in the top three by at least 60 percent of the voters. These races include the San Francisco mayoral election and the Maine Democratic primary for governor, both crowded fields of seven or more candidates where no candidate received more than 40 percent in the first round. In other words, RCV winners are not only earning a majority of the vote against their strongest opponent in the final round — the familiar guarantee of ranked choice voting — but also earning significant 2nd and 3rd choice support from backers of their top opponents.

2. Approval voting encourages high rates of bullet-voting.
Approval voting suffers from what we believe to be a significant defect: a vote for one’s second choice hurts the chances of one’s first choice being elected. As a result, many voters naturally feel pressured to vote for only their first choice, a practice known as “single-shot” or “bullet” voting. Given that only 8,579 votes were cast from 6,390 voters in the CUSG election, a large majority of voters — mathematically somewhere between 65 percent and 93 percent — cast a bullet vote in the executive race. This is similar to our findings from Dartmouth student government elections under approval voting, where between 78 percent and 98 percent of ballots were bullet votes.

The higher the rate of bullet voting, the closer the outcomes of approval voting will resemble the outcomes of regular plurality voting. Indeed, one key resemblance between approval and plurality was the subject of problem #1 above: candidates being elected without demonstrating broad-based support. If approval were used in modern political elections, we expect campaigns would encourage their supporters to bullet vote, broadening the practice further and making the outcomes indistinguishable from those of plurality voting.

Under RCV, a vote for your second choice never hurts your first choice, so voters are much more likely to vote for more than one candidate. Less than 15 percent of voters ranked only a single candidate in this year’s Maine Democratic gubernatorial primary and the mayoral elections in San Francisco and Santa Fe, for example. Of course, some amount of bullet voting is inevitable regardless of the voting method, as there will always be voters with no preference beyond their first choice. But if we enact a new voting system under which bullet-style voting is the norm, we will have changed the system in name only and still be stuck with plurality outcomes.

3. Approval voting takes guesswork to choose your favorite front-runner.
Imagine if voters had known with high probability, before the election, that Ignite and EMPOWER would be the top-two vote getters. Then many would have likely voted differently. In particular, more voters would have approved of either Ignite or EMPOWER (at least one but not both) to ensure they would have a say over which of the two ultimately prevails. With only 4,766 votes for Ignite and EMPOWER combined, somewhere between 25 percent and 60 percent of  voters had no such say in this CUSG election. Had these voters guessed the top two in advance, they could have had more impact on the result. That’s the reality of approval voting: a voter’s success relies heavily on their skills as a political prognosticator.

With ranked choice voting, the voters doesn’t need a crystal ball to have a say between the top-two finishers. They just need to include one of those front-runners in their ranking, which can always be guaranteed by ranking all the candidates. Granted, not every voter will do that, and some jurisdictions currently limit the number of rankings, but in practice the percentage of voters that do have such a say is impressive. This year, more than 90 percent of valid ballots continued to the final round in Maine’s Democratic primary for governor and in San Francisco’s and Santa Fe’s mayoral elections.

Days after voting had concluded, the CUSG election took a surprising twist. The EMPOWER ticket was disqualified by a CUSG election tribunal and were prevented from winning any office. While EMPOWER didn’t win the executive race, all those who had bullet-voted for them wound up with no say at all in the race — probably somewhere between 1100 and 1500 voters. Had this disqualification happened in a ranked choice voting election, that candidate could have simply been crossed off each voter’s ballot, and since bullet voting is uncommon with RCV, a higher fraction of voters would have continued to have a say in the result.

Approval voting does offer key advantages over single-choice plurality voting. While plurality pressures many voters to abandon their true first choice, voters should never fear voting for their favorite under approval. If your first choice can't win, approving of your backup choices helps fend off the "spoiler" effect. On the other hand, if your favorite is viable, additional approvals might defeat your first choice, and some voters may be unwilling to express equal "approval" for their later preferences on the ballot anyway. Navigating the strategic dilemmas of approval voting can be tricky and error-prone, but it offers a superior voting experience to single-choice plurality nonetheless.

While there has been ample study and hand-wringing over the theoretical properties of voting methods, what ultimately matters to a method’s success is its consistency with how real voters think and how the method works in practice.  These real, practical challenges with approval voting — the lack of majority winners, the high rate of bullet voting, and the reliance on political prognostication — illustrate why we continue to prefer ranked choice voting for competitive elections.

Greg Dennis, Ph.D. is policy director of Voter Choice Massachusetts (VoterChoiceMA.org)

Join Us Today to Help Create a More Perfect Union