The Case for Incumbency

2024 was the year of the election as countries across the world held elections. As the results of these elections came in a trend began to emerge. By the middle of the year it was evident that there was a broad backlash against incumbent governments. And it really was a global trend; ParlGov global research has tracked the elections of 10 major countries and for the first time since at least the 1950s every governing party lost vote share in 2024. 

For American political pundits the questions very quickly became, "would America follow this global trend?" But as quickly as the question was asked it became muddled. Yes, Democrats held the Presidency and the Senate, but Republicans held the House, and the Supreme Court was dominated by Republican appointed Justices. Control of the federal government was divided, and while Democratic President Joe Biden experienced low approval ratings with a 56% disapproval rating as of the midpoint of 2024, House Republican governance was conspicuously dysfunctional and the Supreme Court has experienced a net disapproval rate since the overturning of Roe v Wade.

To further complicate the question the Republican Presidential nominee was a former President who had lost his previous reelection campaign to the current President, and by July 21, 2024 Joe Biden had dropped out of the race to be replaced by his Vice President, Kamala Harris. This created a real question of whether Trump would be burdened with the incumbency label, having served as President less than four years prior, or would Harris as a member of the current administration.

We have since gotten our answers to "which candidate did voters identify as the incumbent" and "whether America would follow the anti-incumbent trend". Republicans were able to hold on to the House of Representatives while winning the Presidency and flipping the Senate, placing control of the government squarely with Republicans. This result was foreshadowed as far back as January 2024 when Gallup asked voters whether President Biden and most members of Congress deserved reelection.

This finding was reaffirmed in October when Gallup again asked whether members of Congress deserved to be reelected.

As evident by these results, the sentiment towards incumbents deteriorated over the nine months from January to October 2024 when only a plurality, not even a majority, of respondents felt their own Representatives in Congress deserved to be reelected. 

This begs the question, though, what exactly do voters want when expressing an anti-incumbency preference. To be anti-incumbent is an abstraction, after all. To move from an abstraction to a fleshed out position one must first identify who, or what the incumbent is; and as stated earlier, that was at the very least muddled in the 2024 election cycle. Even after the incumbent is identified a set of policies, or lack thereof, must be prescribed to the incumbent that voters do not support. 

In the case of the 2024 election the economy was the most important issue to voters. President Biden and Democrats were seen as both responsible for the inflationary environment of the past few years due to increased federal spending in response to COVID, as well as the Inflation Reduction Act and Infrastructure Investment and Jobs Act, and were viewed as not having taken action to meaningfully reduce prices. 

Given that voters ranked the economy as their top issue in the 2024 election, and thought candidate Trump would be better able to handle the economy, it would seem that the nature of voter's anti-incumbency preference was motivated by economic policies. 

Outside of issue preferences of the 2024 election, there is some additional polling that may help further shape the anti-incumbency concept. First, pollsters find that an increasing percentage of Americans identify as independents rather than Democrats or Republicans. Furthermore, a majority of Americans believe we need a third-party option in our elections.

Second, between 2000 and 2020 32% to 43% of the eligible voting population is not participating in elections. Of that cohort between 11% and 18% reported not voting because they were not interested or felt their vote would not make a difference and 8% to 25% reported not voting because they did not like the candidates or campaign issues. This is a significant portion of the population that is disengaged from our political system because they do not think that system is responsive their beliefs.

These findings would seem to suggest anti-incumbency is not necessarily partisan or even issues based, but rather a rejection of our political system as a whole. Viewed in this light, the anti-incumbency vote was perhaps more of an abstraction than a rejection of a specific set of policy prescriptions after all. 

If I was to synthesize each of these poll results into a single expression of voter sentiment it would be this; voters want a responsive government and do not feel like that is what they are getting. This is further supported by the disapproval ratings for all three branches of government. But what does it mean for a government to be responsive? 

There are two ways in which a government can be responsive; it can be responsive to voters or it can be responsive in its ability to perform its duties, what I will call actionability. While distinct, these are not necessarily mutually exclusive qualities of government.

I do want to be careful here not to diminish or discredit the dissatisfaction expressed by voters in the series of polls cited above, or in the 2024 election. After all, I too want a government that is responsive to my political persuasion! However, I think it would be a mistake to not address the contradictions between voters' expressed preferences and their revealed preferences. Namely, voters' expressed preference for a government with greater actionability and their revealed preference of, at best, vacillating between a normal governing party in the Democratic Party and a party that is seemingly uninterested in governing in the Republican Party. Now some may say this is an unfair characterization of the Republican Party, but this is a party that failed to adopt a party platform in 2020 and the 118th Congress, led by Republicans, was by far the least productive Congress in at least 40 years. Whether or not one agrees with Democratic policy positions and governance, if voters desire a government with actionability then they should be voting for politicians other than what Republicans have offered.

Voters are not fully at fault for this contradiction, however. Our system presents voters with a theoretical false dilemma, but practically only allows for two legitimate options in any election. Not only is there not only a Democratic and Republican side to any issue, but voters also do not consistently hold Democratic or Republican perspective on every issue. People are complicated and their political perspectives are equally heterodox. 

While there are two ways in which a government can be responsive, my read of the motivation behind the anti-incumbency vote is a frustration with the actionability of government more so than the responsiveness of government to voters. As we saw with voters' top issue being the economy, voters were responding to inaction from the Biden administration on prices. Further, Gallup has also found that a majority of Americans would like their elected leaders to compromise on issues. That's not to say that responsiveness to voters is not also important, so I do want to come back to it, but first I would like to examine the second form of responsiveness; government's actionability. 

What I think is being overlooked by the anti-incumbent voter is the value of institutional knowledge. In the past several years we have seen examples of the benefit of legislators with deep institutional knowledge and those without. Democratic Representative Nancy Pelosi was able to effectively govern the House of Representatives as Speaker of the House with a very narrow, ten seat majority, because she understood the levers of power available to her. Republican Representative Kevin McCarthy, on the other hand, may be the prime example of a legislator who did not have deep institutional knowledge; his tenure as Speaker of the House also oversaw a narrow, ten seat majority, was extremely dysfunctional and punctuated by his own caucus removing him as speaker and McCarthy resigning soon after. While both wielded the Speaker's gavel with similar majorities the Senate and White House were aligned Pelosi and not McCarthy, creating different incentive structures for both; Pelosi to enact meaningful legislation to advance Democratic priorities and McCarthy to obstruct and frustrate Democratic policies. To that end, McCarthy could be considered a success. Although, as the earlier polling cited, American voters want their elected officials to compromise on at least some issues. Further, control of Congress has been divided before and not been as unproductive as the 118th Congress and no other Speaker has been removed by their own caucus.

Congress is a complicated institution with a set of arcane rules operating in a larger governmental structure that only adds additional layers of complexity. Institutional knowledge gained from having served in Congress allows for individual legislators to have actionability within this system and as a result create a government with actionability. This is not to suggest that institutional knowledge is the only knowledge valuable to a legislator. There is value in having a Congress made up of Senators and Representatives with a diverse set of backgrounds and expertise. Expertise that allows legislators to effectively regulate certain industries and backgrounds that makes legislators genuine representatives of their constituents. 

Beyond institutional knowledge, an incumbent will have existing relationships with other legislators who they know they can work with to enact, amend, or block legislation. More senior legislators are more likely to get high ranking positions within party leadership and on committees. This allows those legislators to set the agenda for the legislative session as a whole or for specific committees. Ultimately, seniority means influence and legislators with influence have greater actionability to be able to deliver results for their constituents.

This is to argue that if voters want a government with actionability, basing their vote on an anti-incumbency zeitgeist is not the means to achieve that particular end. What this is not an argument for is an inherent virtue of career politicians. Politicians that do not need to be responsive to voters often end up being lazy or corrupt, or both. In fact, to ensure we have a government with actionability we need a government that is responsive to its voters. 

To that end a majority of voters think that there should be term limits for members of Congress. While on its face support for Congressional term limits and anti-incumbency may seem to express the same sentiment I think there is a nuanced distinction between the primary expression between support for each. The anti-incumbency vote did actually result in a responsiveness to voters, at least in electoral results. As such, and for the reasons laid out earlier, this seems to be an expression of frustration with the governments actionability, or lack thereof. Imposing term limits, on the other hand, appears to be an attempt on the part of voters to reclaim some government responsiveness.

The issues laid out with a blanket anti-incumbency vote apply just the same to term limits. Further, to impose term limits on Congress is to treat a symptom of the problem—entrenched politicians—and not the underlying issue—an electoral system that makes it overly difficult to remove lazy, incompetent, or corrupt politicians. Congressional term limits would remove both those legislators that are effective along with those that are ineffective, without discretion and with no mechanism to ensure those ineffective legislators are replaced with effective ones. 

Gerrymandering, campaign finance issues, and our first-past-the-post electoral system all contribute to a system where elected officials become entrenched and become less accountable to their constituents. In a better system, not even an ideal system, elections would be the mechanism to remove unpopular elected officials, making term limits obsolete. Improving our elections systems would allow voters to retain the effective and popular representative while removing the lazy, incompetent, and corrupt. 

I would argue the reason we do not have a government with actionability is precisely because we do not have a government that is responsive to its voters. To fix the former the latter must be addressed.

The Secondary and Tertiary Problem: Caucuses and Closed Primaries

American elections are incredibly disaggregated; each county has their own board of elections that is responsible for executing an election while adhering to the election rules established by each state. General elections are, for the most part, uniform in how they are administered. From state to state, there will be different requirements to register to vote and different rules regulating how, where, and when an individual can vote; but, generally speaking, the process to vote in Boise, Idaho is functionally the same as the process to vote in Baltimore, Maryland on that first Tuesday following the first Monday in November. The same cannot be said of our primary and caucus elections.

States and political parties have more control over how primary and caucus elections are conducted. To the point where the election process during the Iowa Caucus will be unfamiliar and perhaps foreign to someone used to a normal primary election. The extent to which individuals are able to participate varies widely from state to state as well; from open primaries that allow any registered voter—regardless of party affiliation, to participate—to closed primaries that only allow voters registered with the party to participate.

This disparity in the administration of primary elections has drawn the attention of good-governance groups who make the argument that closed primaries are locking some people out of the primary process, namely independent voters, and have the effect of electing more partisan representatives. My home state of Ohio currently has partially open primaries, meaning that voters can request either party's ballot on the day of the primary election. However, Republicans in the state are looking to change how Ohio conducts its primaries, in September 2023 Senator Michele Reynolds (R – Canal Winchester) sponsored a bill which would close Ohio primaries. So lets examine whether the openness of a state's primary does have any noticeable effects on its elected representatives.

Given that the election system and the openness of that system vary from state to state we are able to isolate each and look at whether they have an effect on the outcome of elections. First, we will look at the caucus and primary system to see how the different systems affect turnout—are voters being shut out of primaries and caucuses. Second, we will look at primary access and its affect on the partisan makeup of Congress—does the primary type affect the partisanship of elected representatives.

Before looking at the turnout results of primaries compared to caucuses, we should outline the differences between the two. Primary elections look more like general elections, with voters reporting to polling stations and casting a ballot in a voting booth. Vote-by-mail and early voting is also often available during primary elections, giving voters more opportunities to cast their ballots. Caucuses can be thought of as party meetings. These are held on a single evening at a single location in each county or voting precinct. Candidate surrogates will make the case for their candidate and party members will debate before the vote is held.

As may be clear, it is easier to vote in a primary election than it is in a caucus. Barriers to voting are inherently problematic in a representative democracy—after all, if not all voters are represented its not a democracy that is particularly representative. If we focus on this for a moment longer, the consequences of the barriers to participation in caucuses become more problematic. Being held on a single night means that voters that are not able to take time off work, find child care, or able to be relieved of any other obligations they may have are unable to participate. Furthermore, being held at only one location in each county or precinct means that those with limited mobility will have difficulty getting to the caucus location. The population of people that we are describing here is going to tend towards poor or working-class adults, people with disabilities, and the elderly that are being systematically excluded from caucus elections.

Looking at Presidential primaries and caucuses from 2008 to 2020, there is a large disparity between turnout of registered voters in states with primaries compared to states with caucuses. This includes election years when both parties have an open field without an incumbent candidate.

In some cases, like Iowa, the results of these elections hold an outsized influence as they are the first election to be held. For further discussion on this point see The Primary Problem.

Given that turnout in caucuses is extraordinarily low, averaging 5.6% in the last four primary and caucus cycles, it would seem that we should move on from this system entirely. It is not even a difficult case of arguing what new, alternative system we should use. We have another system in place in a majority of states that sees almost 7x the turnout in our primary election system.

That said, I do not think we should be satisfied with an average turnout of 35% in primary elections, or even the 42% average of 2008 and 2016 when both parties had an open field during the primary. The candidates in the general election should not be determined by less than half of registered voters in any given year.

Looking more closely at primary turnout specifically we are able to observe another distinct cleavage between the two primary types; open and closed primaries. As previously mentioned, there is a spectrum of primary access from open to all registered voter to closed to all but registered party members, depending on the state. One might intuit that there is a direct relationship between the openness of a primary and participation by voters. 

In Presidential primaries from 2008 to 2020 there is a consistent 4% greater turnout in those primaries that are open compared to those that are closed. Even so, turnout for open primaries remains too low at an average of 36% overall, or 44% for the 2008 and 2016 primary years.

To add another layer to why low voter turnout during primary and caucus elections is problematic; not only do obstacles to voting make it more difficult for marginalized communities to participate in the election, but more partisan voters are more likely to participate in primary and caucus elections. So not only are specific groups' views and perspectives not being considered during primary and caucus elections, but the more extreme views of the parties are weighted more heavily.

This leads nicely into our second topic; the partisan effect of open versus closed primaries. Similar to the effect on voter turnout, one may also intuit that the openness of a primary would correlate to how partisan elected representatives are. 

DW Nominate tracks how partisan Senators and Representatives are based on their voting record—with 0 being non partisan; negative values being left leaning and positive values right leaning. Using this metric we are able to group the Senators and Representatives from each party based on whether they are from a state with an open primary system or a closed primary system. Refer to the table below for the results for elected officials for the congressional sessions from 2011 through 2021.

 These results are a bit more muddled than the voter participation results in the Presidential primary and caucus elections, above. In the Senate it is fairly clear that open primaries result in less partisan politicians, particularly for Republican Senators. And for Congress as a whole, the results are as expected, with open primaries having a moderating effect on the elected representatives. However, in the House the results do not follow the expected trend, or the results of the Senate and Congress. This is likely because there are additional factors affecting House elections, namely partisan gerrymandering. Through gerrymandering, elected officials functionally negate any potential moderating effect of an open primary by drawing single party majority districts. 

In defending her bill to close Ohio primaries, State Senator Reynolds argues that a closed primary is needed to prevent voters from switching their party affiliation the day of the primary to manipulate the results of the election—voting against a candidate instead of for one. Proponents of the bill have not been able to provide evidence that this kind of election manipulation has occurred at a scale which affected the outcome of an election. In fact, the proponents of this bill have not been able to provide any evidence that this has occurred at all; proponents including Secretary of State, and head elections official, Frank LaRose.

Those opposing closing Ohio primaries, and opposing closed primaries more generally, argue that closed primaries suppress voter turnout by excluding independent voters and by excluding independent voters limit the moderating effect on the election resulting in more partisan representatives. I have attempted to validate the conclusion of those opponents to closed primaries herein, however, I would like to add a bit of nuance to the attributed cause of excluding independent voters.

While it is true that voters identifying as "Independent" has steadily increased throughout the new millennia, these independents are more partisan that independent voters of the past. That is according to Michigan State University political scientist Corwin Smidt who found in his paper, Polarization and the Decline of Americas Floating Voter, that between 2000 and 2004 voters who identified as independents voted more consistently for a single party than those who identified as strong partisans from 1972 to 1976. This suggests that if a state was to close its primary these voters could register with one of the major parties without concern that they may want to vote in another party's primary in the next cycle. It is also the case that the increase in voters identifying as Independent has been driven primarily by Gen X and Millennial voters. Younger voters tend to be less politically engaged and those who are less politically engaged are less likely to participate in primary elections. So low participation of independent voters may be more of a function of their cohort's political engagement.

I do not have an alternative theory for why closed primaries reduce voter turnout and produce more partisan representatives. It may be the case that states with closed primaries correlates closely to states with stricter voting laws more generally, having the knock-on effect of more partisan elections with lower turnout. The Election Law Journal's Cost of Voting Index throws that theory into doubt, however, with the "cost of voting" appearing to be fairly evenly distributed among open and closed primaries. It may be the case that by design closed primaries present barriers to entry, even for registered, partisan voters and it is this structure that is the problem. Whether this is the case or not it would seem there is little risk in abandoning closed primaries in favor of open primaries.

Data: 

Presidential Primary Turnout 

House Partisanship 

Senate Partisanship 

Forward to Reform: The Case for Overhauling the Mechanisms of Elections

In the 2020 Democratic Presidential primary Andrew Yang had a moment. He was a fresh face to the Democratic party with a background in business instead of politics, and he differentiated himself by strongly advocating for universal basic income. He eventually dropped out of the presidential primary and a year later he tried to capitalize on his political momentum by running for Mayor of New York. Again, he did not make it out of the primary. Fast forward another year and Yang is back in the headlines, this time announcing that he is launching a new third party, the Forward Party.

When explaining his reasoning for starting a third-party, Yang loves to throw around data; the percentage of Americans that identify as independents, and the number of Americans that want a third-party, and the number of Americans that want their elected leaders to compromise. The problem is, Yang’s understanding and analysis seems to begin and end here. He is right, to some extent, about the polling; and I will dig into that more a bit later. But a major third-party does not exist in America for of a lack of third-parties. There is no reason to believe that Yang’s Forward Party will fare any differently than any of the other 80+ third-parties in the country. The problem is structural and without addressing the structural issues a legitimate third-party will be dead on arrival. However, it is much easier to create a party to make it appear as though you are solving a problem than it is to actually reform a system. This is particularly true when to reform the system you need those with power in the system to change the system that they derive their power from.

The thing is, I do not disagree with Yang’s assessment, though I think either ignores or omits key information to fit his narrative, and where we diverge is in his proposed solution. So, lets start to break his argument down.

To start us off, lets look at some polling from Gallup on party identification. It is true that most Americans identify as independents, and that has been the case since 2009 according to Gallup’s polling. However, as I will examine more later, most Americans are casting their ballots for either of the two major parties.
 

While Americans that identify as independents is a plurality, this may be a bit misleading. First, I guess it needs to be stated, while there is an Independent Party that is not what this poll is identifying. This poll is identifying “independents”, as in those Americans that do not identify as either a Democrat or a Repbulican. In the 2020 election there were more than 80 named third-parties included on ballots across the country. Some, like the Libertarian and Green parties, will have more support than others, but spread that 41% across those 80+ parties and Democrats or Republicans are going to have the plurality.

Second, Michigan State University political scientist Corwin Smidt found in his paper, Polarization and the Decline of the American Floating Voter, that between 2000 and 2004 voters who identified as independents voted more consistently for a single party than those who identified as strong partisans from 1972 to 1976. This, Smidt argues, is due to parties that are more ideologically sorted now than they were fifty years ago, which means that voters could not use party identification as a heuristic when casting their ballots in the same way they can today. In turn, this has resulted in increased partisanship and voter calcification. That explains the voting behavior, but not the self-identification. That likely has to do with the general dissatisfaction with the two major parties and Congress, as we will examine later. 

So, despite the lack of a clear conclusion that can be drawn from Americans’ self-identification as independents, we did have two other data points to look at that might offer some clarity. If we again turn to Gallup polling, we will see that a clear majority of Americans favor the introduction of a major third-party.

Despite this expressed preference, Americans continue to elect Democrats and Republicans. As discussed above, this is due to increasing partisanship—which extends to independent voters as well—and voter calcification as voters become less likely to switch their votes between the parties, and negative partisanship as each side increasingly sees the other as an existential threat. This is not to say that the respondents to the poll lied, or the poll was fundamentally flawed. The dissatisfaction with the two parties is real and is reinforced by congressional approval ratings that remain in the 20% - 30% range, but something is preventing Americans from acting on their preference for a third-party alternative. Before we get into those factors lets look at the last data point as I think it is essential to understand the forces at work in our current system.

Lastly, turning again to Gallup, a majority of Americans believe that it is more important for leaders in Washington to compromise than it is to stick to their beliefs.
 

However, these results are immediately recontextualized if we take this analysis just one step further. Turning to Pew, they too find that a majority of Americans want elected officials who will compromise with the other party. The important difference in the the Pew survey is they followed the initial question by asking partisans if elected officials of their party should compromise with the opposition party. Support for compromise in this context fell precipitously.

Partisanship, particularly negative partisanship, is the likely culprit for this seeming hypocrisy. But I would argue that this desire for the opposition to compromise is the manifestation of a symptom—i.e. negative partisanship—not the underlying ailment. Negative partisanship is the result of a zero-sum system we have created where you either win an election or you lose an election. Those in the minority have found ways to frustrate the majority to prevent them from fully enacting their policy agenda; but for the most part if you win then you get to govern and if you lose, well, win the next one. And this is where the problem lies. Our election system incentivizes two parties because (1) you can win with only a plurality of the vote, (2) if you do not win an election, you do not have a part in governing, and (3) structures within our electoral and legislative systems incentivize the party in the minority to be nothing more than an opposition party.  

I think it is worth lingering on this last point for a while longer. We should assume that parties are rational actors within a system and the parties have two primary objectives within that system; to win elections and to wield power. When making their electoral case, parties can either make an affirmative case for themselves or a negative case for their opponents. Assuming majority control is realistically realizable for the minority party, they are going to behave in a way to maximize their electoral performance. The minority party understands that the majority party will get credit for legislative accomplishments, whether those accomplishments are bipartisan or not. However, the minority party understands the inverse is true too; the majority party will also get the blame for partisan gridlock. While a minority party may not be able to easily make an affirmative case for themselves, they are able to take away the majority party's ability to make their affirmative case. In the process, the minority party is making a negative case against the majority party—"what have they done to fix the problems of the nation when they held the majority?" 

This is the argument Frances Lee makes in her book, Insecure Majorities, though she takes it a step further. Lee compiled the share of the national two-party presidential votes, share of House seats, and share of Senate seats of each party in the chart below. It shows that the relatively small and short-lived majorities in the period since 1980 is relatively unique in the post-Civil War era. As stated above, the parties have two primary objectives. But as Lee illustrates in the chart below, there were extended periods where achieving the first objective of winning a governing majority was out of reach for the minority. That left wielding power as the only achievable objective, and the only way to exercise what little power the minority party did have was to compromise and work with the majority party. 

We are able to see the effect Lee is describing if we look at the number of Senate cloture votes from 1917, when the cloture rule was introduced, to 2020. A cloture vote is the three-fifths majority needed to end a filibuster, and so is used as a proxy for a filibuster since there is not data collected on when a filibuster is initiated. If we overlay the chart below, from Tim Ryan Williams at Vox, we can see that the number of cloture votes begins to rise roughly around the period of insecure majorities that Lee identified begins. 

There is a thread through these three data sets that Smidt and Lee can help us follow. As Smidt argues, in the second half of the twentieth century parties realigned themselves, sorting along ideological lines. This resulted in more polarized parties, and allowed voters that were likely always ideological partisans to more easily and consistently identify the candidate that they align with. Shortly after, the Democratic majority that had held throughout most of the century broke down, but instead of being replaced by a large and enduring Republican majority, the majorities were small and shifting. Lee posits that as long as majority control was attainable within an election cycle or two the party was incentivized not to compromised in order to win the next election. This created a feedback loop where increasing partisanship lead to decreasing compromise which fed frustration with the opposing party among voters, which led to increasing partisanship. While this loop continued to spiral, voters became increasingly disillusioned with Congress and the two major parties. So, voters increasingly identified as independents, and say they want compromise and a third-party option. 

To this point I have just stated that our system does not benefit third-parties, despite the polling that would suggest it otherwise should. We can look at the current composition of the Congress to see the practical, real-world results of our current political system. Currently there are three independent members in the US Senate; Angus King from Maine, Bernie Sanders from Vermont, and Kirsten Sinema from Arizona—only Senators King and Sanders were elected as independents, Sinema left the Democratic Party at the end of 2022. On the other side of the Capitol there are no independent members of the House of Representatives. Most recently, Justin Amash, Representative from Michigan, was elected as a Republican but left the party in 2020 to join the Libertarian party. He did not seek reelection. We should be seeing a trend here—though a very small sample size, which is indicative of the systemic advantage to the two parties in itself, half of the independent representatives to serve in Congress since 2016 were elected as members of one of the two major parties and not as independents.

We can look at the highest office in the nation as well. Since the Civil War the Presidency has been held by either a Democrat or a Republican. Even in 1912, when former President Theodor Roosevelt ran as a Progressive after not receiving the Republican nomination, the Democrat, Woodrow Wilson, won the Presidency. In that election, Roosevelt was only able to garner 28% of the vote. This is impressive considering that he received more votes than the incumbent Presidential candidate William Howard Taft. 

It is the case that Taft and Roosevelt split the Republican base, and had one not been in the race the other would have likely defeated Wilson. This highlights one of the arguments against third-party candidates, they are viewed as spoilers to one of the two major candidates—because it’s a zero-sum contest. Ross Perot in 1992 spoiled George H. W. Bush’s reelection; Ralph Nader spoiled Al Gore’s election in 2000; Jill Stein spoiled Hillary Clinton’s election in 2016; and Gary Johnson spoiled Donald Trump’s popular vote victory in 2016.

There is another way to look at Teddy’s Progressive presidential campaign though; whether it had an effect beyond his own election. The 62nd Congress (1911-1913), had one Progressive member of Congress. After the 1912 election the number of Progressives in Congress waxed to 10 members. This was the high-water mark for Progressives however, within three election cycles there would be none remaining. This is, again, an impressive legacy of a third-party candidacy for President, though it was short lived. But that is kind of the point here. Either the Progressive party was going to flair up and perform well in an election cycle or two before fading away again, or it would have grown and eventually supplanted one of the two major parties, likely the Republican party.

With Teddy’s third-party candidacy as a standard lets look again at those modern examples of spoilers. Ross Perot, running as an Independent in 1992, won 19% of the popular vote. Not quite Teddy’s 28%, but still a significant portion of the electorate, and well above any other third-party candidate since. Perot ran again in 1996, this time running as a candidate of the Reform Party. He was unable to build on his momentum from the 1992 election and only received 8% of the popular vote. In 2000, Ralph Nader, running as the Green Party candidate, received 3% of the popular vote. And in 2016 Gary Johnson, running as the Libertarian Party candidate, and Jill Stein, running as the Green Party candidate, received 3% and 1% of the popular vote, respectively. As one might be able to infer from the discussion of independents currently in Congress, none of these third-party candidacies were able to produce broader electoral gains for their parties.

What, then, is the solution here if not to try and create a major third-party? As I have argued throughout, because our system is zero-sum there is not room for a major third-party. That is not to say that the Democratic and Republican parties are interminable or immutable; throughout our history we have had parties rise and fall—the Federalists, the Democratic-Republicans, the Whigs—but one party was always supplanted by another. Furthermore, there has never truly been a period of multi-polar governance in this country; the Presidency has always been held by one of two major parties and though there have been more third parties in Congress at various points, they never wielded significant power and eventually faded away. See Figure 1 below for additional detail.

The solution is to remove the mechanisms that are currently in place in our electoral system that creates a zero-sum contest. Some argue that the Electoral College should be removed in favor the more direct popular vote; but this only addresses Presidential elections and still allows candidates to win with only a plurality, not a majority of the electorate. A more comprehensive set of solutions is needed to address all elections—whether that is the adoption of ranked choice voting or a proportional system. These alternate mechanisms for electing governing officials allows voters to cast ballots for third-party candidates with less worry that they will be “spoiling” the election for their preferred candidate of the major parties or throwing their vote away on a candidate that does not have a chance of winning. This would allow Americans to act on their expressed preference for a major third-party and for them to be able to vote in a way that aligns with their ideological identification—i.e. independent.

Further, if truly multi-polar governance was achieved elected officials would be able to compromise with members from parties outside of their own party without having to negotiate with someone that holds diametrically different views, as is the case in the current system. This would functionally make Pew’s survey cited above moot as elected officials could compromise to achieve policy goals without undermining their beliefs. However, if multi-polar governance was achieved further reforms would be needed to facilitate effective governance—thinking specifically about the filibuster in the Senate and the 60-vote threshold to functionally pass any legislation.

I recognize that this sort of reform is not likely to be enacted since reform would have to come from the very people benefiting from the current systems. This is the conversation we should be having though, not whether a new third-party with substantial resources is going to upset the status quo. It very well may be a disrupter in the political system, but if history is any indicator this new third party will either fade away in a few election cycles or take the place of one of the two major parties. Either way, in the long-term, equilibrium returns to a two-party system. 

The Primary Problem

The Presidential Primaries as we understand them are a relatively recent development in America’s electoral politics. Not that America's electoral system has been immutable; from changes to electing the President and Vice President on a single ticket to the direct election of Senators, America has been willing to change its processes for electing its governing officials. The modern Presidential Primary process we are familiar with today is a response to the riots at the 1968 Democratic National Convention in Chicago.

To understand the adoption of the modern system, we should set the stage leading up to the 1968 DNC. In 1968 Lyndon Johnson was President and had only four years early won reelection in a landslide victory. However, despite passing some legislation popular with the Democratic base, like the 1964 Civil Rights Act and the 1965 Voting Rights Act, civil strife at home continued to grow. The Civil Rights movement continued its fight for equality and the Vietnam War continued to drag on, distracting from Johnson's domestic agenda—'The Great Society'. The absence of a clear motivation or objective for the Vietnam War—other than the increasingly broad policy of 'Containment'—along with the draft and apparent lack of any progress towards victory galvanized the public against the war,
particularly among young Americans. As a result, the anti-war movement became a focal point of the 1968 Democratic Presidential Primary.

Tapping into this zeitgeist, Eugene McCarthy, Democratic Senator from Minnesota—not to be confused with Joseph McCarthy, Republican Senator from Wisconsin, of 'Red Scare' and 'McCarthyism' fame two decades prior—ran a primary campaign against President Johnson on an anti-war platform. Meanwhile, President Johnson, an adept politician, read the tea leaves and announced that he would not seek reelection on March 31, 1968. Four weeks later, Johnson’s Vice President, Hubert Humphrey, declared his candidacy for the Democratic nomination for President.

Leading up to the Convention, held during the final week of August 1968, only 13 states and Washington DC held primary elections. Humphrey was not on the ballot in any of these primary elections, yet he won the Democratic nomination from delegates from the remaining 37 states controlled by party leaders. Outraged by the party leadership's flagrant power grab, anti-war protesters confronted police outside the convention hall, which devolved into a riot.

In response to the chaos of the 1968 DNC, the Democratic Party decided to hold primaries or caucuses in every state to decide their nominee for President. The Republican Party followed suit shortly after.

Though not nearly as dramatic as the events of the 1968 DNC, tumult during the 2020 primary has again pushed Democrats to rethink how they conduct their primaries. During the 2020 Iowa Caucus, the first contest between Presidential candidates, a failure of the electronic voting software delayed results, caused confusion, and ultimately led to conspiracy theories over stolen votes. The discussion whether to replace Iowa—and New Hampshire—as the first primary states predates the 2020 hiccup, however. Some have pointed out that both Iowa and New Hampshire are whiter and more rural than the country as a whole, and so, are unrepresentative of the country. This matters because the early states in the primary process winnow the field of candidates, effectively selecting the candidates later states will vote for. Voters in two states that are predominately white and rural are going to have different considerations than a more diverse and urban electorate.

President Joe Biden has weighed into the discussion, proposing South Carolina be the first primary state. President Biden owes his presidency to his primary victory in South Carolina. Heading into another primary in a year, the President likely wanted to discourage any potential challengers, and if one should arise he would want an early, decisive victory to quash any concerns over his electoral chances before they arose. If there was any doubt that President Biden’s suggestion of South Carolina was about politics and not representation, FiveThirtyEight looked at which states were most representative of the Democratic Party’s base. They found South Carolina ranked 46, behind Iowa (42) and New Hampshire (34).

While there are some objective measures over which state is most representative of the country to go first in the primaries—Illinois according to FiveThirtyEight—I think this begs the question, should any one state be first in the primaries? After all, voters in a Midwest state, even a representative one, will have considerations other than voters in the plains dealing with drought, or boarder states dealing with immigration, or coastal states dealing with rising sea levels.

Iowa and New Hampshire will make the case for themselves to continue to hold their positions as first in the nation. Some of these I find convincing; others, not so much. These states will argue that because they have held this position in the primaries for so long the residents of the state understand the importance of their selections and take this responsibility seriously. This, I’m not so convinced by, but we’ll look into voter motivation a bit later. The argument continues; because both Iowa and New Hampshire are relatively small, candidates are able to easily visit every major town and every county; meeting with voters across the state. This 'retail politics' is important for candidates to hone their message and actually "touch grass". I am a bit more persuaded by this argument; politicians should understand the interests and concerns of the people they are seeking to represent, however, the early campaign message is being tailored to an electorate that does not look like the nation. And lastly, the media markets in these states are relatively cheap. This gives newcomers who do not have a campaign war-chest or a robust fundraising network in place the ability to compete. I think this too, is a valid argument. We should not want a system that is biased towards a ruling class and only allows the wealthy to break onto the political scene. 

So, the question then becomes, can we achieve these benefits while also improving representation? To this point we’ve done a bit of stage setting, but let’s complete the picture with a quick look at what primaries look like today. I opted to look at the 2016 Presidential Primary calendar because (1) it was an open primary for both Democrats and Republicans, and (2) the 2020 Primary calendar was interrupted and jumbled by the COVID-19 pandemic. In 2016 the primaries began on Monday, February 1, with the Iowa Caucus and ended Tuesday, June 7. It spanned five months and there were 20 different election days. The primaries also began seven months before the party national conventions that are typically held in August—though candidates are declaring their candidacies earlier and earlier in the election cycle.

Of the 50 state elections held by both parties to determine their nominee, Democrats held 14 caucuses and Republicans held 13. If you live in a state that holds the more common primary and are unfamiliar with caucuses; these are party meetings typically held during the evening, in a public place where candidate surrogates make speeches before voters cast their ballot. This creates barriers of participation for people who cannot take time off work, find transportation to the caucus location, find someone to care for their children, or cannot attend an hours long meeting during the week for a myriad of other reasons. Many others have made the case against caucuses so I will not go into detail here. As the main thrust of my argument is for greater representation, caucuses that systemically preclude full representation—specifically participants of specific groups like parents, the sick or disabled, or poor—should be retired in favor of primaries.

Now to get back to our question; should any one state be the first primary state? Inspired by FiveThirtyEight, I did my own analysis of the demographics of states based on race, educational attainment, urban and rural divide, occupation, and employment status. Unlike FiveThirtyEight, I looked at the country as a whole and not the demographic breakdown for one or both of the parties. I also included a few more categories than the team a FiveThirtyEight. Based on the metrics I used, Oregon should be the first state to hold a primary—South Carolina is ranked 26, Iowa 38, and New Hampshire 43.

To those who do not live in Oregon, it might seem like an odd choice (to those in Oregon it might seem odd!); Oregon is whiter and slightly more educated than the nation as a whole. However, its rural-urban divide, occupation, and employment rate trend very closely to the national average. 

If ranking by race alone Illinois would be first; educational attainment – Georgia; rural-urban divide and occupation – Oregon; employment – New York. These other states may seem like more natural choices, they are states that are often in the political discussion in a way the Oregon is not.  

If we were to select one of these states to be first though, what other arguments would we have in defense of any individual state over the others? Cards on the table, I think an average of each of these metrics should be used, but I have no good argument for Oregon, alone, to be first. There may be reasons why a state should not be first; the state already holds an outsized influence over politics and media—think New York or California—or, as has been argued, the state is unrepresentative of the country. However, I am not able to defend weighing one state's political quirks more heavily by placing it first in the primaries, even if it is the most representative state in the country. Particularly because those quirky political preferences are weighed so heavily by our primary process. 

I said I would return to voter motivation, so lets take a look at that now. Every four years as part of the horse race coverage there will be stories about 'electability'. No one really knows what electability means. It is a lot of vibes, but one could assume it means something like broadly popular and able to win elections. A candidate is able to make a strong case for this by, well, winning primary elections. But the strange thing about the discussion around electability is that it is a bit circular in its reasoning. Discussions over candidate electability often lack the context of 'favorability' or 'strength of support' polling, which might actually do a better job of telling us who voters like and support—i.e. who is electable. Instead, electability seems to be asking respondents who they think has the most support and is the most well liked. In the early primary states there are few metrics for voters to inform this decision; there are fewer poll results and no primary results. So, the electability metrics that are released are take into consideration by voters, and because partisan primary voters want their party to win in the upcoming general election, they vote accordingly. And here is the circular reasoning; voters think a candidate is electable and so will vote for that candidate, who wins the election because they were electable (or more accurately, perceived to be electable). After the first few primary results come in, some candidates will perform well and appear electable while others will perform poorly and drop out. This is the winnowing effect of the early states that we previously discussed; but what's more, these early states have defined a narrative for the winning candidates going forward—namely, that they are electable.  

In 2016 Iowa voted first on February 1 and a week later New Hampshire voted on February 9, then about a week and a half later the next election was held on February 20. In the Republican field, Ted Cruz won the Iowa Caucus; Rand Paul, Mike Huckabee, and Rick Santorum dropped out. Donald Trump won the next primary in New Hampshire; Jim Gilmore, Chris Christie, and Carly Fiorina dropped out. Two elections and the field of Republican candidates was halved. On the Democratic side Hillary Clinton won the Iowa Caucus; Martin O'Malley withdrew after, leaving the race between Clinton and Bernie Sanders up to the Convention. 

Following New Hampshire there is a week and a half before the next set of primaries for the media to start to create narratives for election based on the results 1.38% of the country. Even if California were to be first in the nation, at 12% of the nation's population, would we think this fair? Perhaps, but I think there is a better way. It seems to me, the primary calendar should be, to some extent, condensed and regional primaries should be held.

 Above is my suggested regional groupings for primaries. Refer to Figure 1 in the linked support below for further detail of this breakdown. When trying to determine the regional groupings I had a couple objectives in mind; (1) I tried to keep the regions compact and (2) I tried to ensure the population across regions was relatively consistent. The latter is why some groupings might look a little odd, like Pennsylvanian's inclusion in the Appalachia grouping instead of New England or the Midwest. However, if I were to group PA with New York in New England, or with Ohio and Illinois in the Midwest the results of these regions would potentially weigh more heavily. It is the same reason Texas and Florida are separated. 

Maintaining compact regions is important for one of the reasons that Iowa and New Hampshire argue they should be first; it should be as easy as possible for candidates to traverse the regions to see as many voters as possible. That will be easier in some regions than others, the Central region is very large, but that is in the interest of this region having a comparable population to the others. By having regions, for the most part, continuous it allows candidates to set routes and continually stop and visit with voters in different regions in a way that would not make sense if Florida was grouped with Washington. Sure, a candidate could drive, stopping a long the way, but that would be wasting valuable time on states that have either already voted or will be voting in a later primary, at the expense of some states that will be voting in the next primary. 

One might argue that this is exacerbating the problem; we are swapping out the interests of the first few states for the interests of the first region. This is a fair concern, but as we did with individual states, we can identify which region is most representative of the nation to prioritize that group first. Furthermore, while there may be some regionalism at play, the individual politics of each state are unique leading to a greater diversity of representation. 

Pressing on this further, one might point out that this system would continue to favor larger states. Candidates would spend their time campaigning in the largest states in the region. To this I would point out that in several regions there is more than one highly populated state and this would not obviously be a sound strategy as it would effectively be forfeiting in the other states, which leads to my next point.

With each region having a similar population there is no obvious leader, meaning that winning the first region may not be as significant as winning the first state. This is because the results of the first primary become a bit more complicated. A candidate might come out of the first set of primaries with the most delegates, but they  may not win a majority of the states or votes. Or maybe a different candidate wins in each state. Or perhaps the candidate that wins a plurality of the states or votes does not win the most delegates. Any of these results preclude a clean narrative of a front runner. Sure, there will likely be candidates that drop out, but the mixed results allow more to stay in. It may not remove the 'electability' question, but it at least would allow more candidates to credibly make an argument for it, allowing for later primary regions to vote for their preferred candidates.

It is also worth considering how this will affect the campaign strategies of candidates. Not that there is no strategy currently involved in primary campaigns, but this would allow candidates deploy their resources differently. Under the current system candidates take a risk not campaigning heavily in Iowa and New Hampshire; the strategy paid off for President Biden, but not everyone is a former Vice President with deep connections in Congress (specifically to Representative Jim Clyburn from South Carolina). 

Consider the New England region; less well-funded candidates could choose to forego media buys in the expensive New York media market in favor of flooding New Hampshire, Vermont, and Maine with ad buys. Sure, well funded campaigns with wealthy donors are going to be able to play in every market, but how is that different than how the campaigns are currently conducted? And if well funded campaigns do decide to go the route of expensive ad buys in the New York market hoping to win a knockout win with a big delegate haul, they will have fewer dollars for ad buys in other markets. This allows new comers to the national political stage to target their campaigns in hopes of breaking through other, smaller media markets and notching a couple wins to carry momentum into the next regional primary. 

To state the implications of this plainly; the adoption of regional primaries would include expensive media markets in nearly every region. However, because there are so many markets in play at the same time, it has the potential to dilute well-funded campaign's ability to flood all media markets making less well-funded campaign's media efforts in cheaper markets more potent. 

So, now that we are all convinced, we should have regional primaries, which region should go first? Based on my analysis, the order should be South Central first, followed by the Midwest, South East, New England, Central, Appalachia, and West Coast last. 

I think order matters a bit less if a regional primary calendar is adopted, for the same reasons given for adopting regional primaries in the first place. It has also been suggested that a rotating primary schedule should be adopted so that one region (or state) is not always first, which seems reasonable to me. 

Parties are largely in control of how primaries are conducted. Its true that Iowa and New Hampshire have state laws requiring that they are the first states to hold caucuses or primaries, but there are ways around this. I mention this because it means it is far easier to reform this specific part of our electoral system than it would be to reform general election processes; through Constitutional amendment or reform efforts in each of the 50 states. I would like to say that it is good that the Democratic party is again wrestling with the problems in their primary system, however, it seems clear that they have placed Democrat priorities over democratic principles. Primaries are where our electoral process begins; if that first step is unrepresentative then how can the elections that follow be representative?