help_outline Skip to main content
Shopping Cart
cancel

News / Articles

The “Correct” Winner is Squeezed Out in the Alaska Special Election

Marcus Ogren | Published on 10/3/2022

    - Why Boulder and Broomfield Should Care

  By Marcus Ogren with assistance from other members of the LWVBC Voting Methods Team 


Boulder and Broomfield are going to be using Instant-Runoff Voting (IRV) in 2023 so let’s look at the recent Alaska special IRV election to better understand how the instant runoff works. The big headline in that election was that Democrat Mary Peltola defeated Republican Sarah Palin to win a US House seat in a deep red state.


 
A blank IRV (single-winner form of RCV or Ranked Choice Voting) ballot




               How ¼ of Alaska IRV voters filled in their ballots

In Alaska voters weren’t limited to just saying who their favorite was - they also got to specify a second choice. What most interests our Voting Methods Team is that the other Republican, Nick Begich, would have defeated either Peltola or Palin in the final round, but he didn’t win the seat! Why???

To understand why, we have to take a closer look at Instant-Runoff Voting.

Here’s how the ranked ballots were tabulated in the first round, looking only at 1st-choice rankings. (We’re ignoring the write-in votes.)

Peltola: 40.2%
Palin: 31.3%

Begich: 28.5%

 

In IRV, when no candidate gets a majority in the first round, the candidate with the fewest votes is eliminated, and their votes are transferred to the next choice marked on their supporters’ ballots. In Alaska the eliminated candidate was Begich. If a Begich supporter marked “Peltola” as their second choice, that ballot’s vote went to Peltola; likewise with Palin. If the Begich supporter did not mark a second choice, then that ballot was “exhausted.” The final-round results were:

Peltola: 48.4%

Palin: 45.6%

Exhausted: 6.0%

Peltola had more votes in the final round, so she was the winner.

 

The IRV tabulation only allowed election observers to see the Peltola vs. Palin head-to-head matchup. But what if we look at the Begich vs. Peltola and Begich vs. Palin matchups? We can do this because the State of Alaska provides the ballots’ full rankings. We find that Begich would have won against either Peltola or Palin because of the number of second-choice votes he would have received from the eliminated candidates’ supporters.  
               

Begich vs. Peltola:


Begich: 46.6%

Peltola: 42.1%

Exhausted: 11.3%

Begich vs. Palin:


Begich: 53.7%

● Palin: 33.7%

Exhausted: 12.6%


Begich would have defeated either opponent by a larger margin than the margin by which Peltola beat Palin. He was the overwhelming second choice of both Palin and Peltola voters; both preferred him over the other candidate by roughly a 10-to-1 margin. Begich was the clear compromise candidate, but he did not have enough 1st-choice votes to advance to the second round.

 

This phenomenon is called a “center squeeze”: A centrist (Begich) is defeated before the final round even though he’d have beaten either opponent in a one-on-one competition.








center squeeze IRV

A Center Squeeze
(reprinted with permission from The Center for Election Science)


This isn’t the only center squeeze election that’s happened under IRV. The famous 2009 Burlington, VT mayoral election saw a center squeeze, and there were three probable center squeezes in the 2022 Australian federal House elections. Unfortunately, we don’t always have access to the lower rankings to know whether or not a center squeeze has occurred.
 
Since the most popular candidate was not elected in Alaska, does this mean that Alaska was wrong to adopt IRV? Probably not. Before we rush to judgment, it’s worth noting that the same thing might have happened under the old Plurality system if we assume Palin would have beaten Begich in a Republican-only primary.
 
Consider the 2017 Alabama US Senate election, a Plurality election with party primaries and a general election: The highly controversial Roy Moore won the Republican primary, beating Luther Strange. Democrat Doug Jones then defeated Moore in the general election, despite Alabama being even redder than Alaska. Many analysts believe that Strange would have beaten either Moore or Jones if the full electorate had weighed in – making Strange a victim of the center squeeze.
 
Insofar as IRV deserves blame for the election of the “wrong” winner in Alaska, it should be blamed only for replicating the old system of party primaries, not for making things worse.
 
Are there voting methods that would have outperformed IRV in Alaska and elected Begich? Yes. Condorcet (“con-door-say”) methods guarantee that, if there is a candidate who wins head-to-head against all the other candidates, that candidate wins. Other voting methods that avoid the center squeeze include STAR Voting and Approval Voting. (See this article on the 2009 Burlington election for how STAR and Approval Voting might have fared in a center squeeze scenario.)
 
Because no voting method is perfect, our Voting Methods Team wants cities and states to evaluate multiple voting methods before deciding which will work best for them. This fall, Seattle voters will get to decide whether to keep the top-two Plurality for their primary elections or replace it with top-two Approval Voting or Bottoms-Up RCV.
 
And stay tuned: Alaska is going to have another IRV election in November! Peltola was only elected to complete the term of the deceased Don Young, and she, Palin, and Begich will all be competing again in the regular November election to serve in Congress in 2023-24. Will the coming IRV election be a repeat? Or will some Republicans who prefer Palin now (strategically) rank Begich first so their party will win the seat? Only about 4.4% of Palin voters would have to do this if nothing else changes. Even Palin may prevail if more Begich voters list a second choice this time. In any case, the November election will be the perfect opportunity to see what lessons Alaska voters and candidates (and we in Boulder and Broomfield) have learned from other Instant-Runoff Voting elections.


Colorado Flag
Alaska Flag
Vermont Flag
Alabama Flag

    Colorado Flag
    Alaska Flag

    Vermont Flag
    Alabama Flag