In November’s midterm elections, control of the U.S. House of Representatives will be determined by the thinnest of margins.
One of the most closely contested races is in Alaska, an election that is unusual for two reasons. First, the election is essentially a rematch of a special election held in August. Second, the election will be conducted using ranked-choice voting, a system for conducting elections with more than two candidates whose popularity in the United States is quickly spreading.
In August’s special election, Democratic Party candidate Mary Peltola defeated Republicans Nick Begich III and Sarah Palin and, as a result, she is favored to win again. However, Peltola’s victory was the result of an extremist bias that is inherent to the method used by ranked-choice voting to select winners. It is thus entirely possible that the fate of the House will turn on a flaw in the ranked-choice voting system rather than the preferences of the Alaskan voters.
The popularity of ranked-choice voting stems from the fact that it permits candidates with extreme political views and narrow bases of support to run in elections without acting as a spoiler for one of the major party candidates. For example, in the 2000 presidential election, Democrat Al Gore lost to Republican George W. Bush in Florida by a smaller margin than the total number of voters who voted for Green Party candidate Ralph Nader.
In ranked-choice voting, if no candidate gets a majority of the vote, candidates are iteratively dropped from the election and their support is reallocated among the remaining candidates until one candidate has a majority. In Florida in 2000, then, Nader would have been eliminated in the first round and his support reallocated between Gore and Bush. If most Nader supporters preferred Gore to Bush as their second-place choice, these votes would have swung Florida, and the presidential election, to Gore.
However, ranked-choice voting makes it more difficult to elect moderate candidates when the electorate is polarized. For example, in a three-person race, the moderate candidate may be preferred to each of the more extreme candidates by a majority of voters. However, voters with far-left and far-right views will rank the candidate in second place rather than in first place. Since ranked-choice voting counts only the number of first-choice votes (among the remaining candidates), the moderate candidate would be eliminated in the first round, leaving one of the extreme candidates to be declared the winner.
This is exactly what happened in the special election in Alaska. Peltola was declared the election winner under a ranked-choice voting system, but the more moderate Begich was preferred by more of the electorate. In the official election results, Begich received just 52,536 first-place votes, compared with 74,817 for Peltola and 58,339 for Palin, and was therefore eliminated after the first round.
However, we analyzed the anonymized voter data made public by the Alaska Division of Elections, a dataset that includes the complete ranking of candidates for every scanned ballot in the election and comprises over 99 percent of the total votes counted. We found that Begich won head-to-head contests against Peltola by over 8,000 votes (86,385 to 78,274) and against Palin by over 38,000 votes (99,892 to 61,606).
What should be done? Locales interested in holding elections with more than two candidates should use a different vote counting system. There are many voting methods that ensure moderate candidates will tend to beat extreme candidates in the presence of a polarized electorate.
Three well-known and well-studied methods are Borda’s rule (where voters allocate points to candidates based on their ranking of candidates), Copeland’s rule (which gives candidates points for the number of times they beat another candidate in a one-on-one competition) and Kemeny’s rule (which reverses as few votes as possible to produce unanimity among the voters). There are many other vote-counting systems that share a similar tendency toward moderation when the electorate has many voters with extreme, countervailing views.
Consider Borda’s rule, which in a three-candidate election would assign 2 points to a voter’s top-ranked candidate, 1 point to candidates ranked second, and 0 points to candidates ranked third or not ranked at all. Adding the points from all of the voters, the candidate with the most points wins. After removing write-in candidates, Begich would receive 186,277 points under Borda’s rule, Peltola would receive 168,238 points, and Palin would receive 145,435 points, meaning that Begich would have won the special election.
In fact, Copeland’s rule, Kemeny’s rule, and a wide variety of other well-studied rules would all have selected Begich as the winner.
The ranked-choice system that is being used around the country to conduct elections with more than two candidates is biased towards extreme candidates and away from moderate ones. This is just as important of an issue for the future of American democracy as the other important policy concerns surrounding voting in our country.
When there are more than two candidates, it is not just about counting votes accurately. How you determine a winner from the tallied votes matters too. Given our current polarized political environment, Alaska and the other states that have adopted ranked-choice voting are doing it wrong.
Nathan Atkinson is an assistant professor at University of Wisconsin Law School. Scott C. Ganz is an associate teaching professor at Georgetown University’s McDonough School of Business and a research fellow in economic policy studies at the American Enterprise Institute.