What Fargo Reveals about Approval Voting

While contextually an improvement over plurality elections, approval voting has downsides compared to ranked choice voting

Fargo’s recent elections show how approval voting offers more options for voters than plurality voting. But it also provides the latest evidence that far fewer voters with approval voting support second and third choice candidates than in ranked choice elections and exposes how voters can manipulate outcomes.

“I would probably bet that every candidate says just vote once because that has more power as a vote.”

• Fargo mayor Tim Mahoney in announcing his successful re-election campaign in February

“The problem is the more candidates you vote for, the more it dilutes the impact of your vote."

• Rep. Shannon Roer Jones in January as she launched her campaign that resulted in a third place finish

For three decades, FairVote has studied our elections and advocated for reforms to improve them. We're particularly well known for our expertise on ranked choice voting (RCV) - also known as instant runoff voting and preferential voting. As our reform impact has grown, we’ve heard a lot from backers of other election methods to pick a single winner - methods like approval voting, score voting, Condorcet voting, and “score-then-automatic-runoff” (STAR) voting.

Every election method has its pro's and con's, making any quest for perfection the wrong goal. But at FairVote we believe that ranked choice voting is distinctly better in practice than alternative winner-take-all methods, which explains our ongoing focus on winning RCV. RCV also has proven viability, including widespread international use and, in the United States, two states using RCV for all their federal elections, seven more states using RCV for at least some voters in some federal elections, more than 50 cities using RCV in their most recent elections, more than 100 colleges and major organizations using RCV, and pro-RCV bills becoming law this year in four states. RCV has won the last 13 times on city ballots, by an average of nearly 30 percentage points, with several more ballot measures coming in November 2022.

The FairVote website compares RCV to several alternatives. We’ve felt a need to make such comparisons due to encountering opposition to RCV from backers of other methods seemingly upset that we don't devote energy to advocating for their favorite method.

Fargo uses approval voting to elect its mayor for the 1st time

In that spirit, let’s turn to the June 14th city elections in Fargo, North Dakota, where the city’s two-term mayoral incumbent Tim Mahoney handily defeated six challengers. This was the first election in the United States where approval voting was used to directly elect one candidate. (Fargo first used approval voting in 2020 to elect two city commission seats and this year in St. Louis, approval voting was used to narrow the primary field to two runoff candidates). There were aspects of the election that advocates of approval voting will like. But the downsides confirm for us the value of why FairVote will stay focused on making ranked choice voting the norm across the United States.

Let’s start by examining election results. The North Dakota Secretary of State presents results for this year’s elections for mayor and for two seats on Fargo’s city commission. In the mayoral election, it shows the incumbent winning easily, with 41% of the vote and more than twice as many votes as the second-place finisher. In the city commission contest with 15 candidates and two winners, the results page shows the top two candidates with less than 14 percent of the vote, including Denise Kolpack with 13.9 percent and incumbent Dave Piepkorn with 12.6 percent.

The percentages of the vote are misleading. As often seen in election results for multi-winner elections - as in the two-seat city commission races in Fargo before 2020 — the results are tied to showing the number of votes in an approval voting election for two seats rather than the number of voters. In this year’s commission race, voters cast an average of three votes, while in 2018 they had cast an average of two votes. In 2018, Piepkorn also seemed to win with a very low share of the vote of 16.6%. The media failed to grasp the right math after elections in 2016 and 2018, with editorials like this one in 2017 using the misleading “votes” denominator in calling for approval voting and this 2018 news story that again used the “votes” denominator rather than voters.

In reality, Piepkorn won with the support of about 30% of voters in 2018 and won with a somewhat higher share of 38% in 2022 – a result far short of a majority of the vote, but not less than 20%. Approval voting advocates have conducted a spirited media campaign to draw attention to this discrepancy. As a result, the media is improving its coverage of the issue. For now, though, anyone scanning the results on the state elections results website will see very low shares for winners. This is a perception problem for approval voting advocates to address wherever it might be tried.

Substantive issues tied to tactically withholding approval votes

Those problems with reporting approval voting election results are fixable. But there are more substantive concerns that are more troubling. City commissioner Dave Piepkorn clearly has his supporters, but his win as a political conservative in 2018 helped spur on more progressive approval voting advocates, as featured in this news story and in this letter to the editor arguing Piepkron’s re-election “highlighted the need for approval voting” (ironically making use of the secretary of state’s misleading voting percentages).

With well under 40 percent of the vote, Piepkorn may have been re-elected only due to a fracturing of the majority vote – although, just like in 2018, the approval voting system doesn’t provide enough evidence of voter intent to know. In a field of 15 commission candidates for two seats, voters on average cast their approval votes for only three candidates, just one vote more than the two votes cast by voters in 2018 before Fargo had approval voting.

The fact that so many voters may not have elected anyone highlights a big contrast between approval voting elections and ranked choice voting elections: that is, voters in ranked choice voting elections are far more likely to show support for multiple candidates and elect more definitively representative winners.

In Fargo, we estimate about 60% of voters cast votes for only one mayoral candidate despite having a choice of seven candidates. In contrast, let’s turn to the 20 Utah cities using RCV last year, including in a number of mayoral races and a number of two-seat city council races very much like Fargo’s elections. Here’s what we can learn from the full ranking data provided by Utah County for seven Utah cities. Even though not every mayoral race was competitive, 88% of voters indicated at least a second choice, and 92% of voters ranked a winner among their top three choices. Nationally, among the 50+ cities using RCV, it’s typical for that same share of voters — about nine out of ten voters — to rank two or more candidates.

The reason is straightforward. Voters with ranked choice voting can indicate a second or lower choice without fear that ranking multiple candidates will have any adverse electoral impact on their first choice. With approval voting, giving a vote to a compromise candidate can cause your favorite to lose. Indeed, in Fargo, two of the leading mayoral candidates made that point in this year’s campaign. Incumbent Tim Mahoney said in February that “I would probably bet that every candidate says just vote once because that has more power as a vote.” State representative Shannon Roer Jones, who finished in third place, said on a radio show in January that “The problem is the more candidates you vote for, the more it dilutes the impact of your vote."

Approval voting presents a conflict: either voters will withhold honest preferences due to engaging in tactical voting or it will violate the principle of majority rule. That is, either voters will withhold votes from compromise candidates or they might instead cast those votes and enable a candidate with little first choice support to defeat a candidate who might well be the first choice of 55 percent of voters.

The problem of potential voter manipulation

A challenge for advocates of any new election method is to make sure voters are aware of it. Inevitably, many voters won’t have heard much about the new rules and certainly not enough to know how to try to game it or to dismiss errant suggestions for gaming it.

For ranked choice voting, backers need to assure voters that ranking a backup choice won’t hurt your first choice. Some voters mistakenly might think their rankings are converted into differential values of points or that all second choices might be added to all first choices (rather than the reality where only votes from losing candidates count for the next ranked choice). But one thing about which we can be reassured: a voter who is highly informed about RCV will use their rankings exactly the same as a voter who knows nothing about RCV, but has the same preferences and follows the instructions on the ballot. While election method theorists love to play with blackboard analysis to suggest RCV can be gamed, it just isn’t done in the real world.

It’s different with approval voting. Once you understand the math of how voting for a second candidate can cause your first choice to lose, you must weigh whether it’s most important to help your first choice win by withholding votes for compromise candidates, even if that risks electing your least favorite. While crystal clear for someone who understands approval voting, this feature of approval voting could lead to other voters inadvertently undermining their favorite candidate by simply following the instructions on the ballot that invite them to vote for multiple candidates.

Consider this year’s commission elections in Fargo. It's possible that different ways of engaging with the system by backers of the third-place finisher Al Carlson (a former senate majority leader) and second-place winner Dave Piepkorn changed their order of finish. Both are Republicans in a city that leans Democratic, so they likely shared some voter support.

Piepkorn finished second in a winning position with 5,834 approval votes, while Carlson had 4,839 votes in a losing position. Suppose 500 strong Piepkorn voters liked Carlson, but, due to being schooled in the math of approval voting, decided not to vote for him so as not to harm Piepkorn. And suppose 500 of Carlson's strong supporters also this year cast votes for Piepkorn because they didn’t realize that might cause Carlson’s defeat. If they had swapped their tactics, Piepkorn would have earned 500 fewer votes and Carlson 500 votes more - and changed the outcome. Carlson voters alone might have handed the race to Piepkorn if 1,000 of them also cast a vote for Piepkorn as a fellow Republican who was a compromise choice. 

Just the potential for more informed voters having a tactical advantage over other voters shows how important voter education is so that as many voters as possible fully understand the pro's and con's of using their approval voting options. Of course, the more voters understand how approval voting works, the more likely they are to withhold approval votes of any candidates other than their favorites - leaving elections open again to “spoilers” and unrepresentative outcomes.

Pointing out flaws with any election method is always possible because there is no perfect one. Kenneth Arrow proved how every ranked voting system will have some flaw that makes it seem illogical. Every “cardinal” system like approval voting that is based on ratings will invite manipulation due to how support for a second candidate can harm the chances of one’s favorite. Plurality voting is rife with problems like unrepresentative winners and “spoilers,” while traditional runoffs can winnow the field to two unrepresentative choices and often result in far lower turnout than in the first round.

That said, approval voting’s issues are real and may get worse over time, as suggested by useful analyses of its flawed history in campus elections that show the trend toward more voters moving to “bullet voting” for only one candidate. That’s why FairVote and FairVote Action are going to focus on teaming with a growing chorus of allies on how to make voting come to mean ranking. Let Americans rank the vote!

Rob Richie @Rob_Richie.