Fix the Top Two Primary: Admirable Goals Don’t Justify Indefensible Outcomes
On June 3, California held primaries for nearly all major state and federal offices, including governor and 53 U.S. House seats. Since the approval of a 2010 ballot measure, the state has used a “top two primary,” a version of which is also used in Washington. In the top two primary, all candidates for an office run against one another, and voters select a single candidate. The two candidates with the most votes advance to the November election, regardless of party affiliation.
Sounds simple, right? What could go wrong?
Plenty, as it turns out. Top two backers claim the system can work and just needs more time, but the verdict is in: it should be reformed and other states should improve the model when considering it. What it does right – shaking up otherwise noncompetitive contests, expanding voter choice in primaries, and creating incentives for more representatives to care more about the general election electorate than they do in traditional primary systems – does not justify a process that can turn general elections into a mockery.
We urge reformers to move forward, not back – that is, we should build upon what top two gets right rather than regress to the traditional primary system. But consider this series of problems, all of which are exacerbated by the system’s central flaw: having only two candidates advance to the general election even when they both clash with the preferences of most general election voters.
- Indefensible outcomes: Distortions in turnout and spilt votes are delivering indefensible outcomes. Hundreds of thousands of votes are still to be counted, but we already know that in congressional district 25, only two Republicans will advance in what should be a swing district. Even worse, the key statewide contest for controller nearly advanced two Republicans despite Republicans collectively only capturing 45% of the vote. Congressional district 31 is within 200 votes of having two Republicans advance to the November ballot for the second consecutive election – even though 53% of 2014 primary voters backed Democratic candidates and Barack Obama carried the district by a 16% margin in 2012. Democrats likely will end up advancing in both the controller race and district 31, but just barely. And in both cases that Democrat will be heavily favored to win in November despite having won only a plurality of the vote for Democratic candidates.
- Abysmally low and unrepresentative turnout: Voter turnout hit an all-time low of less than a quarter of California’s registered voters, with the electorate significantly less representative of the population than the November electorate. The final turnout likely will be four percentage points lower than the previous low in California, and a recent Los Angeles Times poll provides concrete evidence that primary voters are disproportionately old, white, wealthy, Republican and partisan as compared to general election voters. Traditional primaries are also problematic for turnout, but Top Two is so far not reversing California's downward trajectory.
- Lopsided races: Despite top two advocates focusing on the handful of contested races, we can expect mostly boring, no-choice elections in November after mostly boring, lopsided primaries. Of 53 House races, for example, 41 were won by at least 20 percentage points, with only 12 races won with less than an absolute majority in the primary. More than half of all House races will feature a general election between a Democrat and a Republican after each handily defeated other candidates of their party in the primary.
- Shutting out third parties: Third parties and independents were eliminated in every single federal and state contest where at least one Republican and one Democrat ran. They have advanced in at most five congressional races total, all of which were lopsided contests where the majority party candidate earned more than 73% of the vote. Every statewide race will feature a Republican versus a Democrat, meaning the overwhelming majority of the state will not have a chance to hear from or vote for a single non-major party candidate in the general election. This barrier to general election ballot access for challengers of the status quo is historically high and contradicts the fact that a representative democracy should provide representative ballot choices.
- Split votes: In numerous federal and state elections, there was a vigorous contest for second where a candidate may have advanced only due to “split votes” among like-minded candidates. These outcomes are analogous to winning a key election with 30% of the vote, when as many as 70% of voters may have opposed the winner.
- Democrat-Democrat and Republican-Republican contests: There were two much-publicized congressional races where a liberal incumbent (Democrat Mike Honda) and conservative incumbent (Republican Tom McClintock) are facing candidates of their party in November, and allegedly may need to more toward the political middle. But both incumbents won by huge margins – McClintock by 56% to 23% and Honda by 49% to 28% -- and will be heavily favored to increase their percentages in November. Turnout will drop sharply from what it would be if both major parties were on the ballot, given that many backers of other parties will just skip the contest. (They can't even do a write-in candidate, as write-ins are banned.) Even those these contests are more interesting than what they would be without intra-party contests, that comes at a cost: general elections without representative choices.
Reformers Should Go Forward, Not Back
Many editorial writers seem lost in a fog of uncritical support for the “open primary” because they like the fact that every voter can vote in every primary contest and are oddly enchanted with the idea of a general election that only has candidates of one party – apparently with a hope that it will help more centrist candidate. But California’s results show the problematic nature of the system, and how it even falls woefully short on its central premise of boosting moderate candidates.
Given such outcomes, some of the Democrats in power in California are talking about repealing Top Two. But reformers should not go back to California’s old, discredited system that led to mostly lopsided contests in both primaries and general elections. Far better would be to take what’s good about the current system and make it better. We see fixing top two as easy as 1-2-3.
- We must banish the idea that general elections provide sufficient choice if they limit voters to two candidates. We are used to two parties dominating contests, but it should be illegal to ban other options from the November ballot, which the Top Two system effectively does in almost every contest. Furthermore, while it can be interesting to have two candidates of one party make the general election, that does not make it right to eliminate every other perspective. Evidence from the 31st congressional district in 2012 shows that about half of the Democratic voters who make up a majority of that district’s electorate just skipped a contest between two Republicans. California could advance four candidates to the November ballot, rather than two. Doing so would dramatically increase the number of elections with more than one candidate from the majority party without shutting out the second largest party as long as it ran a candidate. A “top four” rule also would open up the ballot to far more third parties and independents, including in three statewide contests this year.
- With four candidates on the November ballot, ranked choice voting should be used to ensure the election of the candidate with the strongest overall support. Doing so would uphold majority rule and reward candidates who reach out to more voters. Using ranked choice voting to reduce the field to four in the primary would eliminate “spoilers” in that election as well.
- Just as candidates are able to indicate their preferred party, parties should be able to indicate their preferred candidates. More information is good for voters and better upholds first amendment rights of political association. The proposed ballot measures in Oregon and Colorado provide a model.
Let’s go deeper into some of the primary’s results to see what is working and what isn’t.
On the positive side, the top two primary has increased the number of competitive primaries while making them more inclusive. All voters, regardless of their party registration, can participate in the primary and back candidates they like. This inclusive alternative stands in contrast to primaries where each party holds a set of separate contests and particularly ones where unaffiliated voters are barred from participation. If we are going to continue to be the only nation in the world in which taxpayers fund party nomination contests, we shouldn’t limit voter choice.
It is also likely that the top two system leads to more legislators being responsive to the full spectrum of their constituents than under the old primary system. When another member of a legislator’s own party can make it to the general election, elected officials in districts that are ‘safe’ for one party must be more attentive to a wider swath of constituents, knowing they could face a real challenge in the general election. Under a traditional primary system, most incumbents know they are never going to lose the general election, and thus are only incentivized to be responsive to the primary electorate.
However, these benefits are overshadowed by deeply troubling problems that have arisen under top two primaries. Accounting for the number of remaining provisional ballots according to California’s Secretary of State, turnout in this year’s primary will end up as less than 25% of registered voters, which will be by far the lowest turnout in the history of California primaries. This low turnout follows on the heels of the first use of Top Two in 2012, which had the state’s lowest-ever turnout in a presidential primary. The top two system may not necessarily be depressing turnout, but it may be that it’s just too “wonky” a contest for most eligible voters when only winnowing the field down to two – as opposed to a traditional primary that, if competitive, at least had a real winner.
Furthermore, the primary electorate was once again distinctly older, whiter, wealthier and more partisan than the general election electorate – and more Republican, with the statewide Democratic vote about 5% less than it likely will be in November. A Los Angeles Times poll before the primary was revealing, as screening questions asked whether respondents had voted in one or both of the primaries in 2010 and 2012 and thought they would vote this year. Consider:
- Age: Just 28% of respondents who were 18 to 49 had voted in one primary, and only 12% had voted in both. Among respondents 50 and up, 63% had voted once and 43% had voted both times.
- Race: Those voting in both primaries in 2010 and 2012 by race included 17% of African Americans, 12% of Latinos, 19% of Asians and 36% of whites – 70% of racial minorities had not voted in either primary, compared to only 43% of whites.
- Income: Only 15% of those making less than $20,000 a year had voted in both the 2010 and 2012 primaries, while 35% of those making $100,000 or more had voted in both primaries
- Party registration: 42% of Republicans, 26% of Democrats and an eye-popping 11% of unaffiliated voters had voted in both primaries, with 72% of unaffiliated voters not voting in either 2010 or 2012.
Regardless of whether the top two primary contributes to such disparities and low turnout overall, it makes primaries even more important because it eliminates all but two candidates before general elections that don't even allow voters to write-in an alternative. When primaries matter more, low turnout in primaries is even more problematic. Such distortions in the electorate almost certainly changed outcomes in key races along what issues candidates highlighted on the campaign trail.
Furthermore, a lot of voters also seem more confused by “top two” ballots. In the U.S. Senate primary in 2012, more than ten times as many voters in Oakland invalidated their ballot as in their 2010 mayoral race with ranked choice voting, and this year 0.5% of gubernatorial ballots were invalidated in San Francisco, which is more than the 16-candidate ranked choice voting election in San Francisco that draw far more voters – undercutting claims that top two has special value for being “simple.”
As discussed, Top Two is creating far too many perverse outcomes tied to elimination of all but two candidates in fractured candidate fields. It would have been absurd to deny any candidates except Republicans access to the November ballot in a contested statewide race for controller simply due to split votes among Democrats and a progressive Green candidate, especially when those candidates together won 55% of the vote. Swing district congressional races should not be decided for one party due to arbitrary outcomes in June, as happened in District 25 and nearly happened for a second consecutive election in District 31.
It also makes no sense to have full choice in June when most voters aren’t paying attention and deny choice in November when candidates are actually elected. Take the 33rd congressional district in Los Angeles County, where Henry Waxman decided to retire, and 18 candidates ran in a wide-open primary that included prominent author and political independent Marianne Williamson, strong Democratic woman candidate Wendy Greuel, and well-financed pundit Matt Miller. Only Republican Elan Carr and Democrat Ted Lieu will advance with a combined vote total of 40%, and now Lieu will almost certainly win easily, turning an exciting chance for real debate into just another lopsided November election.
Other races where the outcome might have been as fairly decided by a coin flip or where general election voters are missing out on important debate include:
- In the statewide attorney general race, Republican Ronald Gold earned second place behind Democrat Kamala Harris with just 12.8% of the vote, barely ahead of three fellow Republicans who all earned at least 8%.
- In the statewide secretary of state race, Democrat Alex Padilla and a Republican Peter Peterson far outpaced a fascinating field that included reform-minded independent Dan Schnur, Green David Curtis and Democrat Derek Cressman, all of whom were raising important concerns about the future of California democracy that won’t be part of the general election.
- In Congressional District 24, Chris Mitchum likely will advance by a slim margin with less than 16% of the vote in a race where 55% of voters did not vote for the incumbent.
- In Congressional District 15, we still do not know whether Republican Hugh Bussell or Democrat Ellen Corbett will come in second. If Bussell keeps his current lead, the nearly 70% Democratic district will face a boring, lopsided partisan election. Whether California voters will have a competitive race or not will be decided by a just a handful of voters in a low-turnout June primary.
In races like these, narrowing the field all the way down to two creates distorted outcomes. By limiting their options to voting for one candidate, California does not allow voters to provide enough information to create any kind of meaningful result. In fractured fields it is important to know more than a voter’s first choice.
A Solution: Top Four with Ranked Choice Voting
There is a solution to Top Two’s problems – and in fact it’s rather simple. If the general election were expanded from the top two candidates to the top four candidates, there would be almost no chance of the most popular candidate being eliminated in the primaries, and the impact of low turnout electorates and split votes would be largely minimized. By expanding the field to four and adopting ranked choice voting in November to avoid split votes, we can ensure that candidates that best represent their constituents have a shot in the general election.
Consider the example of Waxman’s open congressional seat. With top four, some 70% of voters would have voted for an advancing candidate, as opposed to only 40%, and the general elections voters would have included two strong women candidates, independent Marianne Williamson and a second Democrat Wendy Greuel. Similarly, losing voters in the State Controller race would be reduced from 54% to a mere 11%, and California would never risk having a key statewide contest with candidates of only one party.
With ranked choice voting in November, voters would be able to rank three candidates in order of choice. Those rankings would enable the state to simulate runoff elections, with the last-place finisher eliminated before each round of counting, and everyone’s ballot counted in each round for the advancing candidate they ranked highest. Using ranked choice voting in both the primary and the general would be ideal, as it would reduce "wasted votes” all the more, but even using it just in the general election would greatly add to the value of general elections with four choices. Candidates would need to earn their vote to win and would do best by campaigning to voters rather than going negative on their opponents. They would need to be more inclusive in their campaigns and more representative by reaching out to be the second choice of backers of other candidates.
Longer-term, the solution for congressional and legislative elections would be to use ranked choice voting in a fair representation, multi-seat district plan. Under this alternative to winner-take-all elections, California would have fewer legislative districts and each district would typically elect either three or five representatives. Elections would be with ranked choice voting in a form that would nearly always result in more than 75% of voters helping to elect a preferred candidate – and often closer to 100%. Representatives would better represent their constituents in terms of ideology, gender, and race, and there would always be substantive debate about the future of California both in campaigns and in the legislature. We’ve posted example of fair voting proposals for California’s representatives in Congress and the state legislature.
It is clear that something must be done. California should reform Top Two, and those interested in imposing the California model of Top Two in other states like Oregon should consider improving their proposals. Top Two distorts outcomes, too easily resulting in elected officials who do not represent their constituents and too often resulting in general elections that shut out a majority of the electorate from being able to vote for a preferred candidate. By limiting voter choice and placing the most power in the hands of a small, unrepresentative group of primary voters, the top two system falls short of its promises. California policymakers should act to reform Top Two by advancing more candidates and using ranked choice voting, while backers of open primaries in other states should expand their vision of what we deserve in a fully representative democracy.