The Independent Party of Oregon’s Secretary of State race was a serious stress test for STAR Voting. With three Democrats, an Independent, a non-affiliated candidate, and only one Republican, this is the type of scenario where most voting methods don’t cut it. Not only that, but it was anybody’s race. All 6 candidates were viable, and for voters, there was no easy way to predict in advance who the front-runners were.
With standard Choose-One voting, elections like this generally end predictably; Most voters realize that voting third party could waste their vote, so even in elections in which they don’t prefer the major parties many people end up strategically supporting the front-runner on their side.
This election was different. The Independent Party primary used STAR Voting, a newer voting method which most voters were trying for the first time and it included candidates from across the political spectrum. The electorate was also unique, and only registered Independents and non-affiliated Oregonians were eligible to participate. If any election was going to take the pulse of Oregon’s swing voters it would be this one.
After an in depth analysis and crunching the numbers from every angle possible, we are pleased to announce that STAR Voting successfully elected the candidate who best represented the electorate, but how did we get there?
A New Voting Method
Voters scored candidates from 0 up to 5 stars, the two highest scoring candidates advanced to an automatic runoff, and those finalists were Kim Thatcher and Ken Smith.
As we can see in the chart above, Kim Thatcher was preferred by a majority of those who had a preference between her and Ken Smith, but in this election a true majority didn’t exist. The electorate was made up of multiple distinct factions.
STAR Voting gives voters the opportunity to score as many candidates they wish and then it uses that high-quality data to find the winner who best represents the voters. Of course, people can just give a top score to their favorite if they choose, but STAR Voting gives voters a strong incentive to show their preference order, even among the candidates they don’t love.
While the idea that voting could be this expressive may be novel to some, it ensures that whether or not your favorites can win, your vote can still make a difference and help you get better representation.
Let’s take a closer look at the voters who didn’t express a preference between the two finalists:
The two finalists in the Independent Party of Oregon Secretary of State primary were Republican Kim Thatcher and Independent Ken Smith. 29.8% of voters expressed no preference between the two finalists.
Here is a breakdown of those “no-preference” voters:
- 21% of voters would have preferred one or more of the Democratic candidates to either of the finalists.
- 6% of voters would have preferred non-affiliated candidate Armand “Rich” Vial to either of the finalists.
- 1% of voters preferred both finalists equally over all others.
The bulk of the no-preference block was made up of supporters of the various candidates who did not make it into the runoff. Most supported the 3 Democrats or the non-affiliated candidate, Rich Vial.
These voters had the opportunity to express secondary support for other factions and effectively coalition for better representation- but they didn’t. For some this may represent lazy voting. Some voters may have had a preference between the other candidates, but didn’t take the time to do that research or mark each bubble. Other voters may not have had a preference for either Kim Thatcher or Ken Smith.
In practice, when voting blocks choose to withhold support from other factions entirely they are choosing polarization, but if a voter gives even a single star to candidates they prefer to their worst case scenario it’s enough to break through and find a better consensus winner if one exists. STAR Voting incentivizes coalition, but that can only work if the voters choose to provide that data. Since none of these factions chose to show secondary support across party lines in sufficient numbers, a consensus candidate wasn’t able to be identified.
Did voters in this election lean right or left?
In order to break down the runoff and its implications further we have to start making assumptions: Voters who preferred Democrats are likely left leaning and those who preferred Republicans are likely right leaning. We could guess that the fact that Rich Vial supporters picked the non-partisan candidate with Republican ties over the other Independent implies that they may be right leaning. Those who preferred Thatcher over the Democrats are likely right leaning. Ken Smith’s website clearly positions him as the “true Independent” in the race so we’ve put his voters in the middle of the spectrum. Lastly, one could assume that most voters who preferred a Democrat would likely prefer an Independent over a Republican.
Of course, it’s worth noting that people do not fit neatly into boxes, and that when we looked at the data there were actually 5 voters who gave a tied top score to a Democrat and to the Republican.
The next graphic shows how many voters preferred each candidate as their first choice. Note that 11% of voters gave a top score to multiple candidates, so the percentages add up to 111%.
As you can see, the conservative block has slightly more voters, but this electorate is pretty evenly balanced. So we took a look at the data from one more angle as well.
- 51.4% gave Kim Thatcher a non 0 score.
- 57% either preferred Rick Vial or supported Kim Thatcher with a non-0 score.
Did the voters who showed no preference between the finalists actually have no preference?
To shine some light on this question we took a closer look at how expressive voters were with their scores.
367 voters, or 49.4% of the electorate, expressed a nuanced preference order and gave a score to at least two candidates.
Of those who voted no-preference in the runoff, 59 voters in total, or 38%, expressed support for multiple candidates. For these expressive voters we can determine that the choice to not show a preference between Thatcher and Smith was intentional.
The other 62% of no preference voters bullet voted for their preferred Democrat only. A “bullet vote” is essentially a vote just like we do in traditional ‘Choose-One’ Voting.
For some, bullet voting may be an honest but polarized vote. For others it may be a lazy vote or a vote by someone who didn’t feel they had an informed opinion on all the candidates. What this election clearly demonstrates is that bullet voting in STAR is not a good strategic vote.
In total, 138 voters expressed a preference for Democrats in a race in which the finalists were a Republican and an Independent. Those voters had an opportunity to help decide which finalist won, but chose not to.
What would have happened if voters overall were more expressive?
STAR Voting strongly incentivizes voters to show their full preference order and voters are instructed to only give the candidates the same score if they “don’t have a preference.”
Kim Thatcher, the winning Republican candidate only won the race by 13 votes over the other finalist, Independent Ken Smith. If just 14 of the no-preference voters had expressed just a slight preference for the Independent candidate over the Republican, it could have turned the election. For partisan voters in STAR Voting the lesson is to not write off the Independent or 3rd party candidates. Democratic leaning voters in this election could have effectively coalitioned with the Independents to get better representation and the same goes for the Independents.
To the best of our knowledge, Kim Thatcher, the Republican, was the only candidate who campaigned and worked to get out the vote for this election. If the Independent had better campaigned and outreached to democratic-leaning swing voters, and if he was able to convince them that he was a serious candidate, he could have won.
Would that have been a more representative outcome? If voters who didn’t show a preference between the finalists had been more expressive would it have changed the results? What if those who had preferred a Democrat had given all other Democrats 4 stars or had given Ken Smith and Rich Vial non-zero scores? What if voters who preferred Vial, the right-leaning non-affiliated candidate, had also given 4 stars to the Independent and/or to the Republican? What if more voters who preferred the Republican had expressed their preference order?
To answer these questions we created a hypothetical simulated election, assigning each voter a hypothetical set of expressive scores and preferences based on their preferred candidate or candidates as well as their position on the political spectrum relative to the candidates.
For voters who didn’t win, it would be logical for them to look at the results, think back to their ballot, and resolve to vote more expressively next time. Let’s take a look at what might happen if voters had the chance for a do-over.
A hypothetical do-over with all voters fully expressing their preferences and degree of support:
To create this scenario, we broke down voters by their preferred candidates and then assigned each block a set of scores and preferences based on the assumptions explained above.
As predicted, in our hypothetical scenario Ken Smith -the Independent- would win. In this election Smith was not the most popular candidate by a long-shot, but he was a compromise candidate who would have probably left the majority feeling like they didn’t win, but like it could have been worse.
Ultimately it’s up to the voters to decide if they are willing to accept that kind of compromise? In this real election they didn’t, and so Ken lost.
The learning curve or should we say ‘un-learning curve’:
It’s a logical assumption that Democrats in this race would have preferred the Independent to the Republican, but it’s important to remember that that isn’t information we can glean from the ballot data. The best we can expect from STAR Voting, is to crunch the voter preferences that voters decided to share, and then determine the most representative winner based on that data.
In order for a voting method to produce the best results possible the voters will need to learn that it’s in their best interest to provide all the data needed to find not only majority preferred candidates, but also consensus candidates when a simple majority doesn’t exist. Voters can still vote with a polarized or partisan mentality, but with STAR Voting it isn’t going to give them an advantage anymore.
Some opponents of STAR Voting have argued that voters could game the system by simply giving their favorites a top score, despite the fact that simulated data has consistently shown that the STAR runoff strongly incentivizes honest voting. This election clearly refutes that argument again, this time with empirical data.
We believe that over time voters using STAR Voting would become more expressive, not less. With STAR Voting the runoff round encourages voters to show their full preference order to ensure they always have a say between the finalists.
Did the candidate who best represented the electorate win?
Yes. Kim Thatcher-R was the candidate who was the highest scoring overall, she also was the candidate with the biggest block of supporters. In the runoff, Thatcher was preferred by more voters than Smith, the other finalist, and she was also the Condorcet winner, meaning that in head to head match-ups she was preferred over any of her opponents.