In 2008, Oregon voters rejected a ballot initiative for a Top Two system by a nearly two-to-one vote. In 2014, Top Two may be back on the ballot, but this time tied to a famous Oregonian name and with a twist: the ability to vote for multiple candidates in the preliminary election, increasing the chance that many general elections will shut out candidates from all but one party.
Under Top Two, as used in Washington and California, partisan primaries are replaced with a preliminary election to determine who can appear on the ballot in November. All candidates run in the preliminary election irrespective of political party preference. Every voter has one vote, and the two candidates with the most votes advance to the general election, even if both have the same political party preference.
We have written on Top Two extensively:
- Our 2013 policy perspective on fixing Top Two in California
- Our report on election results in Washington
- Our analysis of California’s Prop 14 prior to its implementation
- Our reporting on Arizona rejecting Top Two in 2012
Most of our writing about Top Two has been critical of the system as Washington and California have chosen to implement it. Political science scholars have presented evidence that current forms of Top Two have not accomplished much. The Washington/California model for Top Two limits voters’ general election choices and occasionally renders absurd results, such as the 2012 race in California’s congressional district 31, where it advanced only two white conservative Republicans to the general election in a majority-minority and majority Democratic district.
The backers of the new Oregon proposal seem to acknowledge that the Washington/California Top Two model has been less than satisfactory, and so they have added a twist likely to generate controversy of its own. Under the Oregon proposal, voters would not be limited to voting for only one candidate in the preliminary election; they could cast equally weighted votes for as many candidates as they want to. This system, called approval voting, is intended to help resolve the sort of vote-splitting that allowed two candidates of the minority party to advance in California district 31 (although FairVote has expressed doubts about its ability to do so as well as ranked choice voting).
The new Oregon proposal is backed by Mark Frohnmayer. Although Mark Frohnmayer is relatively new to politics, his family is not. His father is Dave Frohnmayer, former Oregon attorney general and gubernatorial candidate. His uncle is John Frohnmayer, former chairman of the National Endowment for the Arts. The Frohnmayers were known as moderate Republicans. That this proposal is attached to the Frohnmayer name gives it a boost of momentum.
Frohnmayer apparently believes that the issues with Top Two should be largely addressed by the use of approval voting in the preliminary election. If voters vote honestly and enough candidates run, this proposal should result in two candidates from the same party appearing on the general election ballot nearly every time.
If a district is majority Republican, for example, then the majority will likely approve of two or more Republicans, such that both the first and second-place finishers in the primary will be Republicans. The same would be true for Democrats in Democratic-majority districts. Even statewide races could often end up with two people of the same party, as even a slim partisan advantage in the preliminary could easily result in two Democrats or two Republicans being the only candidates on the ballot for governor or U.S. Senator.
In theory, such outcomes in highly partisan districts would give voters a contested general election choice between the only candidates who had a real shot to win – that is, such districts are likely to elect a member of the majority party anyway. The proposed Top Two system would allow all voters in the general election to choose which member of the majority party will ultimately win.
There are a couple problems with this result, though. First, as the infamous Sherman-Berman race in California’s congressional district 30 in 2012 demonstrates, when candidates are too similar in terms of policy positions, they may just resort to mean-spirited personal attacks to win, hardly a result promoting moderation. Second, even if the system did result in more “median” candidates being elected, it would come at the cost of a lack of diverse viewpoints in the debates and a lack of choices outside the majority party preference on the ballot. It would be as if the 2008 presidential race in November had presented voters with a choice of only Barack Obama or Hillary Clinton.
Further, there are reasons to think that candidates, parties, and voters will engage in strategic behaviors that undermine the goals of the system, just as they do under Top Two in Washington and California.
Consider the 2010 race for Washington’s state senate seat in legislative district 38. Union-backed interest groups wanted to unseat moderate incumbent Jean Berkey and replace her with the more progressive Nick Harper, but they knew that in a one-on-one race, Berkey would likely beat Harper. Fortunately for them, former Republican Rod Rieger, a long-shot candidate running under the “Conservative” party label, was also running in the preliminary election. The groups managed to propel Rieger to the general election by campaigning for his election alongside their ads for Harper. They thus squeezed Berkey out of the general election entirely, and Harper went on to easily win.
Strategies like these would still be possible with approval voting in the preliminary round. In a Democratic district, backers of a particular Democrat would prefer not to face another Democrat in the general election. With approval voting, they could simultaneously vote for their favorite Democrat and for a weaker non-Democrat in the field. It would be as if Harper supporters could not only send out mailers for Rieger, but actually vote for him as well. As long as that candidate also gets support from others, such a strategy could be viable for strategically manipulating who makes the general election.
Top Two definitely can be altered in a way that would make it a powerful reform in Oregon and elsewhere, but this proposal misses the mark. The issue is less about how people can vote in the preliminary election, and more about what choices are available in the general election.
For example, advocates of this proposal could simply change their calendar. They could move the first round to November and elect the top candidate provided that candidate receives at least 60% approval. If no candidate achieves that threshold, there could be a December runoff between the top two. This approach would preserve voter choice in the critical autumn election and minimize some of the most common concerns about approval voting.
Alternatively, as we demonstrated using California and Washingto
With four candidates and a ranked choice ballot in the general election, voters could indicate their first, second, and third choices to ensure a final winner with majority support compared to their top rival after an instant runoff. It’s a proposal we call Top Four. Taking ranked choice voting into multi-seat elections would then allow for real representation of the diversity of political viewpoints people actually hold.
Mark Frohnmayer and other Top Two advocates are absolutely right about one thing: the polarization and dysfunction we see in legislative bodies today trace back to how we choose to conduct elections. Abolishing traditional primaries and experimenting with approval voting represent the kinds of creative thinking that can help us to make that connection and seek a real solution. Our analysis suggests that Top Four is a better approach, however: one that would generate real competition and real choices while paving the way for real representation.
This article was originally published by FairVote on October 18, 2013. FairVote is a nonprofit, nonpartisan organization dedicated to educating voters and enlivening the public discourse on how best to remove the structural barriers to a democracy. It was updated on Oct. 22nd.
Join the discussion Please be relevant and respectful.
My most recent attempt to give numbers to the strategy of trying to promote to the runoff a candidate you don't want to win in order to shut someone out that you want even less, is at https://docs.google.com/spreadsheet/ccc?key=0Am4BvpGhMUxLdFBzTmFOU3JqOVdmS2RWd3YyU2FqaWc&usp=sharing . Anyone trying to work such an example could start by copying mine and then change the numbers, perhaps add factions and/or candidates.
In regard to these passages: "The groups managed to propel Rieger to the general election by campaigning for his election alongside their ads for Harper. They thus squeezed Berkey out of the general election entirely, and Harper went on to easily win. [paragraph break] Strategies like these would still be possible with approval voting in the preliminary round. In a Democratic district, backers of a particular Democrat would prefer not to face another Democrat in the general election. With approval voting, they could simultaneously vote for their favorite Democrat and for a weaker non-Democrat in the field. It would be as if Harper supporters could not only send out mailers for Rieger, but actually vote for him as well."
How far could such a strategy go toward reducing overall voter satisfaction? The picture you are painting here requires us to think about at least two factions among the voters: the faction doing the strategy you describe, (call them "b") and the faction consisting of those they would like to make their victims (the "a" faction). Please describe a scenario where you list candidates and factions, and you list how much each faction values the event of each candidate's winning (use a scale of zero to one, where one means they will be fully satisfied with the election of that candidate and zero means maximally unsatisfied). I have attempted such specificity because I was convinced of your argument, but I have not found an assignment of numbers that results in the strategy working beyond reducing the satisfaction of the "a" faction to maybe 1/2 what it should be; they never get shut out completely.
Glad to see the fine folks at FairVote take an interest in the effort we're working on in Oregon, and sorry I missed the post here on the first time around. I'd like to address some shortcomings in the analysis that I think bear more thoughtful consideration.
- Labeling the effort under the bundle of "Top 2" is misleading. Of fundamental importance in _any_ runoff system is what method qualifies the candidates to advance to the runoff. California and Washington use "First Past the Post" voting, which has amongst its flaws minority rule and the mathematical inevitability of a two-party system - see: www.youtube.com/watch?v=s7tWHJfhiyo . The use of Approval Voting in the primary is not some incidental tweak, rather it completely shifts the outcome for both voters and candidates. Given that "Top 2" has become a pejorative term as it relates to 3rd-party and independent voices in the elective process, this association is unfair.
- The conclusion that "if voters vote honestly and enough candidates run, this proposal should result in two candidates from the same party appearing on the general election ballot nearly every time" based the party of majority control is also inaccurate. In Oregon, only 9 of 60 legislative districts have a major party majority representation, thus based on the logic of the article, it should be revised to say, "nearly 15% of the time." In every other district, enfranchised 3rd-party and non-affiliated voters will actually have a say in candidate selection. They may tip it to one of the major parties, they may have viable 3rd-party selections to choose from, but fundamentally and finally every voter will have an equal voice on each candidate's ability to advance.
Still, let's take the assertion at face value. The reality today is that in a heavily gerrymandered district, the dominant party always wins the general election. So the election is OVER in the primary. So only ~1/3 of the voters even had a say at all on who gets the office, and maybe only 1/6 of them actually supported the winner. I'd take two candidates with the same party label, voted on by ALL voters in the primary any day.
- The contention that getting to an actual moderate selection well-liked by a solid majority of the populace must come at the cost of diversity of viewpoints is flatly false. The election process, taken as a whole, would provide independent viewpoints a much greater voice in the process as they would be on equal footing throughout the full cycle. Currently, the publicly-funded primary process advertises only to voters and only for candidates of the two major parties. By the time independent voices are even allowed into the ring, they've already lost. The mechanics of First Past the Post voting make them automatic spoilers and make it impossible for them to build legitimate movements. Under an approval primary, a Ralph Nader gets 15% of the vote and can build a platform over time, instead of tossing the election to his ideological opposite and pissing off everyone even moderately inclined to support him. If Barack Obama and Hillary Clinton were actually the two top approved candidates across the country, then why would it not have been far better for ALL voters to have the choice between the two than just the third that checked the box "Democrat??" How is that democracy?
- The gaming scenario outlined also misses the mark. The author from FairVote conflates Harper backers - i.e. those who are willing to nefariously send out mailers to encourage voters for Harper's ideological opposite in order to game a broken First Past the Post voting system, with Harper _voters_ - the voting electorate who check his name on the ballot. Those voters are much more likely to approve of the two candidates they like than they are to try to squeeze the one candidate they're OK with (an incumbent no less) and take the chance that they'll actually knock out the one they actually like. This is a legitimate problem with Washington's system and one that the Oregon proposal fixes completely.
- Top 4 IRV. Ok, I get it. This is actually an ad for FairVote's idea of good election reform. I was once a big fan of IRV myself. Until I learned that it still leads to a two-party dominant system: http://rangevoting.org/TarrIrvSumm.html and that it has some weird math problems that may make it better for some voters who are lazy and stay home on election day: http://rangevoting.org/CompleteIdioticIRV.html . Or until I really thought about how likely it would be in any case to have 4 viable candidates willing to spend the better part of a year fighting over a single job.
But at the base level, ALL 5 of the major problems FairVote identifies in http://www.fairvote.org/assets/Uploads/Fixing-Top-Two-in-California5.pdf are ALSO fixed by the Oregon Fair and Unified Elections Initiative:
* Incumbents and Major Parties: Approval voting in the primary guarantees that an incumbent's challenger in the general election has the highest approval of all potential challengers. This significantly increases the likelihood of a competitive general election race.
* Vote-splitting Unfairness: This is fixed by approval voting in the primary
* Voter turnout distortion: According to the Open Primary folks at the national level, 75% of Californians not affiliated with the major parties don't know they can now vote in the primary election. Further, Oregon uses a vote-by-mail system and sees consistently above-average voter participation nationally. This will also allow a much clearer communication with voters about changes to the primary process.
* Lack of Competition Still the Norm - see above - Approval voting in the primary guarantees that an incumbent's challenger in the general election has the highest approval of all potential challengers. This significantly increases the likelihood of a competitive general election race.
* Association Rights: The Approval Voting Primary measure show's both a candidate's party choice (as signified by his or her voter registration) and any official party endorsements. This preserves the free speech and association rights of both the candidate and the party.
And all this without changing the design of the ballot.
Mark Frohnmayer ought to amend his initiative so that it abolishes the primary, and then all candidates would run in November, using approval voting. The typical legislative race in Oregon only has 3 or 4 candidates every year anyway, counting all candidates from all parties. Abolishing the primary and using approval voting would shorten the campaign season, allow candidates to enter the race in August, and not exclude anyone from the general election campaign season. One disadvantage of the current Frohnmayer proposal is that after February of an election year, no candidate could enter the race. There might be unexpected events later in the year that would create a demand for a new candidate who hadn't entered as early as February.
@RichardWinger Mathematically, Approval Voting with a top 2 runoff beats Approval Voting alone on measures of voting system fairness (and also beats IRV by a wide margin). While I think the reform you suggest would almost certainly be an improvement over what we have today, I don't think it's possible to get the voters to get it there in a single step. Even something as seemingly simple as changing the primary date has significant consequences at the local and national level, and would likely sink such a reform out the gate.