What happened in CD31 in 2012?
This is the “Miller-Dutton” congressional election. It is also a commonly referenced example of something going wrong with the nonpartisan top-two election procedure.
The story takes place in California’s 31st congressional district, a fairly safe Democratic seat east of Los Angeles in San Bernardino County. The Republican (two-party) presidential vote share in 2012 was 42%. Yet, two Republicans, Bob Dutton and Gary Miller, advanced to the general election; Miller won the seat. What happened?
Redistricting set two unusually well-qualified Republican candidates on a path to conflict with each other that would also prove disastrous for the Democratic Party. Gary Miller was an incumbent U.S. House member with many years of experience in public life. As an incumbent member of congress, Miller possessed significant financial advantages over his opponents in both parties. Bob Dutton, though, was also a well-qualified candidate: had been the Republican’s leader in the State Senate. He also had long experience with public office and campaigns.
Four Democrats—Pete Aguilar and three others—also ran. Aguilar, the strongest Democrat, had been active in local politics in the city of Redlands, but both Republicans were very experienced campaigners, armed with meaningful amounts of campaign cash. The Democratic Party failed to mobilize enough resources and endorsements behind any candidate. The result was a divided Democratic vote amid a low-turnout election; paired with an even split of the Republican primary vote, only the two Republicans advanced.
Aguilar, the leading Democrat, fell short of second place by 1,376 votes. Miller finished in first with 26.7%; Dutton came in second with 24.8%; and Aguilar came in third with 22.6%. In the general election between Miller and Dutton, Miller won 55.2% of the vote and the seat.
This case of the district’s majority party being shut out of the general election may be a troubling outcome, but it is the exception that proves the rule: these are rare events. Out of the several hundred top-two elections, this is the only seemingly safe Democratic seat to be lost in this fashion. Furthermore, to pull this off in 2012, it took such low turnout (with no serious contest in the Democratic presidential primary to generate excitement, and an uncompetitive Senate race as well) that the sum of the vote of the two Republican candidates also exceeded the sum of the vote earned by the four Democrats in the primary – so, in the actual primary electorate at least, it was not a safe Democratic seat. Aguilar’s campaign also successfully reached relatively few of the available voters: he obtained only 14,181 primary votes in a district with 106,820 registered Democrats. While these conditions could recur, it is the sort of thing unlikely to happen very often. Miller would not even contest the seat in 2014; Aguilar would go on to win it that year and hold it every year since.
How should we count turnout?
This question returns to the topic of meaningful participation, highlighting:
- Same-party general elections shift the meaningful choice from a lower-turnout primary to a higher-turnout general election.
- Even with the roll-off of “orphaned” voters who have no candidate of their own party on the ballot, participation in the same-party general election is arguably more meaningful for those who choose to vote because they are doing more than selecting a near-certain general election loser.
Consider Representative Nanette Barragán of California’s 44th Congressional District in 2016, a heavily Democratic district in Los Angeles. This district’s Republican two-party presidential vote share in 2012 was 14 percent, declining to 13 percent in 2016 before increasing to 20 percent in 2020. To say that it would be hard for a Republican to win the seat would be an understatement. It is precisely the sort of district that would have been won, in practice, in the primary under the old system. How did Barragán get elected?
The 2016 primary contest featured eleven candidates, including eight Democrats. Barragán’s most serious rival was Isadore Hall III, an experienced Democratic politician, who declined to run again for State Senate and entered the contest supported by the endorsements of many elected officials in the state party. Hall placed first in the primary with 40% of the vote; Barragán came in second with 22%. If this had been the result of a partisan primary, Hall would have won the primary and, almost by default, the seat. With the nonpartisan top-two system, though, the contest continued on to the general election. Barragán defeated Hall 52.2% to 47.8% in the general election.
What is the right way to think about turnout, then? Should we focus on the primary, or on the general election? In the primary, 100,276 votes were cast, including 90,141 split among the eight candidates from the Democratic Party. In the general election between Hall and Barragán, 178,413 voters participated. More broadly, comparing the primary participation in all of California’s congressional districts in 2002-2010 to the general participation in CD44 in 2016—the election that put Barragán in first instead of second—no other congressional partisan primary election in 2002-2010 had more voters participate. In fact, the 2016 Barragán-Hall general election had more than double the average voter participation in the state’s 2002-2010 congressional primaries.
The 2016 Barragán-Hall general election was also very competitive. Given prior research on the roll-off of orphaned voters, undoubtedly some of the district’s Republicans, if they voted in the election for other offices, skipped the general election contest between Barragán and Hall. None of them were required to do so by law, though, and those that decided to participate had an opportunity to meaningfully influence who would represent them in Congress. Although the district was heavily Democratic, it had 35,682 registered Republican voters and 79,512 No Party Preference voters shortly before the election. Under the old system, Republicans would not have been able to express a preference between Barragán and Hall; although the unaffiliated voters could have chosen to participate, they faced an additional barrier to doing so, and many would have likely not taken the Democratic ballot in the primary.
Something is lost, of course. The 35,682 Republicans in CD44 did not have an opportunity to express their preference for a Republican—however hopeless that candidate’s cause—in the general election. That is a different kind of participation: a group of voters preference-signaling rather than selecting. There are (small-d) democratic reasons to care about both kinds of participation. Arguments about the top-two system, though, should take into account the nature of this trade-off. In partisan primary systems, in the lopsided districts that make up most of the legislative districts of all kinds in the country, “the main purpose of the minority party primary is to choose the loser at the general election.” The top-two system offers potentially meaningful participation to all voters at both stages of the election, particularly in districts that ultimately field competitive same-party races.
What does moderation mean?
One source of confusion for journalists and practitioners trying to understand the top-two system: when political scientists say someone is more or less moderate, what does it mean? Consider moderation in these three different ways:
- How candidates present themselves in the election.
- How officeholders vote once in office.
- What officeholders would vote for, if they could just enact their own preferences.
Assessing candidate ideology in all three ways can tell us something about the top-two system. It is possible to be moderate in one way and not in another: someone can sound moderate in a campaign and not vote that way; conversely, someone can sound extreme as a candidate and then turn out to be a centrist in office. And a politician who neither sounded moderate in the election nor voted moderate as a member of the legislature could turn out to be a moderate if circumstances provided an opportunity to express his or her real preferences.
Not all studies use the same definition, and not all studies use the same methods to try to measure these considerations, either. For example, some ways political scientists use to estimate ideologies of legislators treat them as fixed over time, meaning that any detected moderation could only come from officeholder replacement. Other measurement strategies allow for legislator changes over time but are not comparable across states. (For an excellent review of these issues in the top-two context, see the paper by Eric McGhee and Boris Shor in Perspectives on Politics: they use multiple measures in their paper, and explain them clearly.)
It should not then be surprising that one paper on moderation—focused on how candidates present themselves—finds different results from another paper on how legislators behave. Focusing on different time periods or different levels of office, as well as other methodological differences easily explains how high-quality scholarly papers can seem to point in opposite directions. More research will be required to further unpack this question.
It is also important to point out that the discussion about moderation often assumes a single-dimensional ideological space—a liberal to conservative spectrum. That may not well describe American politics; debates about the nature of political ideologies and the ways we ought to think about them are also ongoing in political science.
While the usual interpretation of the Proposition 14 language is that “less partisan” and “more practical” means “moderate,” there is another possible interpretation based on the context of the proposition’s passage: it means doing whatever is necessary for passing a budget on time and getting the mess in Sacramento off the front page of the newspaper. This gets to the third point about moderation: what officeholders might do in other circumstances. Due to the simple-majority budget reform, and the large Democratic majorities in the legislature along with Democratic governors, those same circumstances just have not come up again—legislators needing to take a tough vote to get the budget done. That kind of moderation, voting for a solution in a time of crisis, is very hard to measure without having a crisis requiring those tough votes.
How does top-two compare to final-four or final-five?
As a practical matter: we don’t know.
The Final-Four/Final-Five rules would allow voters to choose any candidate in the primary. The four (or five) candidates with the most votes would advance to a general election conducted with ranked-choice voting.
No state has yet used a final-four procedure; Alaska will for the first time this year. No state has adopted a final-five procedure. So, as of yet, there is no evidence to compare.
It is worth observing, though, that these systems are very closely related to the top-two. The top-two differs only in the sense that “ranking” in the second round is unnecessary; the chosen candidate is “ranked first” and the alternative is ranked second by default. They really differ in a single parameter: the number of candidates who advance.
This means the main difference is likely to be found in the meaningfulness of the primary. There is good reason for campaigns to care a great deal about primary campaigns in California: any time three or more candidates are in the race, each faces the risk that he or she could come in third and miss the general election. Candidates in California need to balance turning out an excited group of their core supporters in the primary with being able to win either a cross-party or same-party general election. Those incentives largely disappear in a top-four or top-five contest.
The potential benefit from a top-four (or top-five) election would be avoiding outcomes like CA’s CD31 (discussed above). If, though, fewer than five (or six) candidates file for the office, there is effectively no primary at all.
Nevertheless, these systems are more alike than they are different. They are nonpartisan in the structure of their candidate competition. They erect no formal barriers to the participation of unaffiliated voters. They do all this and still conform to the single-member district system used in American elections, and can be adopted by states without too much change from their other current institutions. We will have to wait to see how the new Alaska system operates, and also how the “Final Five” works if any other state chooses to adopt it.
What should we watch for in 2022 and 2024?
The coming elections under the top-two system will offer much to watch. Will we see more uncontested elections? Will same-party general elections occur about as often as they did before? Will either party be “shut out” of a seat it should win? Are the same-party elections competitive or uncompetitive?
The 2022 elections provide an opportunity to see effects only revealed by the confluence of the nonpartisan election system and the work of the independent redistricting commission. Unlike in 2012, though, potential candidates and election consultants know more about how the top-two system operates. Generally speaking, incumbents have found separate districts in which to compete this year, avoiding a high-profile conflict such as the “Berman-Sherman” race in 2012. A notable exception is AD34, which features two incumbent Republican members of the Assembly.
Nevertheless, the large margins Gavin Newsom ran up in the 2021 California recall election, and the state’s overall political climate, have created a statewide context somewhat similar to 2014’s (reelecting Jerry Brown), but a national battle for control of Congress may generate more participation and interest in the general election.
Beyond redistricting, the 2022-2024 election cycle will also see the impact of California’s changes to redistricting. The first round of officials elected to the state legislature, under the new term limits, are approaching the end of their term-limited period. Enough legislators are running for different offices or retiring that CalMatters termed it “the Great Resignation.” We will start to see what a public service path looks like with the combination of the redistricting commission, the term limits, and the top-two system.
 For primary vote totals, see California Secretary of State, “Statement of Vote: June 5, 2012 Presidential Primary Election,” https://elections.cdn.sos.ca.gov/sov/2012-primary/pdf/13-sov-summary.pdf
 California Secretary of State, “Report of Registration as of October 24, 2016: Registration by U.S. Congressional
 Note that even though two Republican candidates in the contest evenly split the Republican vote, this did not result in a Republican-on-Republican general election; together, the two Republicans mustered only about 9%, with the leading Republican winning 5.5% of the vote.
 California Secretary of State, “Report of Registration as of October 24, 2016: Registration by U.S. Congressional District.”
 If, of course, one ran. This district did not attract a Republican candidate in 2012 or 2014, and districts are often unopposed in the general election with partisan primaries.
 The pithy quotation is from Benjamin Highton, Robert Huckfeldt, and Isaac Hale, “Some General Consequences of California’s Top-Two Primary System,” California Journal of Politics and Policy 8, no. 2 (April 7, 2016), https://doi.org/10.5070/P2CJPP8230564.
 McGhee and Shor, “Has the Top Two Primary Elected More Moderates?”
 Kousser, Phillips, and Shor, “Reform and Representation.”
 Grose, “Reducing Legislative Polarization.”
 For example, see Hans Noel, Political Ideologies and Political Parties in America, Cambridge Studies in Public Opinion and Political Psychology (New York: Cambridge University Press, 2013), https://doi.org/10.1017/CBO9781139814775.
 Ben Christopher, “What’s behind the ‘Great Resignation’ of California Lawmakers?,” CalMatters, January 10, 2022, sec. Politics, http://calmatters.org/politics/2022/01/california-legislature-great-resignation/.