A New Presidential Candidate Typology

How are the many, many candidates in the Democratic presidential primaries different from each other? How are they different from Trump? These are literally billion-dollar questions: they will anchor the strategies behind the candidates’ campaigns and, once a nominee is chosen, decide how the undoubtedly nasty 2020 general election will unfurl. But despite these questions’ importance, they are also hopelessly vague.

Take, for example this attempt by Nate Silver to rank and compare different Democratic hopefuls with his “five-corners graphs.” In his world, the differences that matter between the candidates are their relative appeal across different social communities in the party, and the candidates with the most potential are those who have high appeal across five “key groups.” With their recent analysis of new candidate Amy Klobuchar’s chances, noticeable head-scratching was found online:

and an early reaction:

To put it another way, pundits like Matt Yglesias and Matt Karp think Silver’s typology is nonsense because

A. The “appeal score” assigned to each of the categories is not really based on current polling data, but on some polls and woozy expectations about the future. Hence these scores lack scientific credibility.

B. The categories themselves are suspect. The categories are intentionally overlapping, but they are a pastiche of demographic and ideological loyalties.

There is solid evidence that political attitudes now differ greatly between the college educated and those who are not, or between urban, suburban and rural households. Why aren’t those divisions easily seen among the five groups?

To the above I would also add

C. These categories present an overly cynical view of presidential politics, as if a successful nominee simply agrees to fight for “millennial interests” or “Black interests.” In reality Obama ran on “Hope and Change;” Bernie Sanders ran on a “political revolution;” Donald Trump ran on “Build the Wall” and “Drain the Swamp.”

How do these slogans, policies and viewpoints map neatly into some pre-election position shared by millennials or Black Americans? To use Occam’s razor, it’s likelier that a particular political personality cause unification or polarization among different groups in a party.

The point is that I have yet to see a candidate typology that even tries to solve these three issues. Even fuzzier labels like “radical,” “centrist” or “moderate” won’t make things better. With that, I built my own.

The typology

The idea behind this new candidate typology isn’t difficult: I separate different presidential candidates based on the language they use when delivering the campaign’s message: slogans, stump speeches or which events they appear or comment on. The problem has been that pundits tend to classify a campaign’s message using those mainstays of “radical,” “centrist,” “moderate” and so on.

Amy Klobuchar and Kirsten Gillibrand are both female senators with “moderate” pasts, but if the former says her campaign is about “grit” and the latter about “women and kids,” I think this difference is substantial. If Kamala Harris endorses both Medicare For All and some milquetoast income transfer credits, is she “radical” or “centrist?”

My typology, the way I process all these rhetorical quirks from candidates’ speeches and press releases, is to place each candidate on two axes:

Establishment/Anti-Establishment: The “establishment candidate” has a long record in politics, associates themselves with big donors and party elders and will find themselves endorsed by an array of known figures. The “anti-establishment candidate” won’t have any of these but will be happy that they don’t, instead touting their support from celebrities, small donors, average Joes and so forth.

Vision/Agenda: A less conventional axis, I define an “agenda candidate” as someone who cares about specific policy interests, and will run on a record of lifelong advocacy for these interests. It helps that they surround themselves with like-minded activists. For Donald Trump this is immigration; for Bernie Sanders this is economic inequality, and so forth.

A “vision candidate” isn’t someone lacking an agenda, but someone who talks about leading the country without pet causes. Instead, they talk about how they will bring disparate interest groups together and uniting the country. This is the message conveyed by Barack Obama when he said “there is no liberal or conservative America, only the United States of America.”

In practice these are axes, hence sliding scales, with the point being that contrasting qualities between candidates imply distance over one, or both, axes. Later I go into detail about the difference between Joe Biden and Bernie Sanders as an example: how both of them could be closer to, say, Mike Bloomberg than to each other.

For simplicity I will often refer to the four possible quadrants formed by the interaction of these two axes, so for example the Establishment/Vision (EV) quadrant, the Anti-Establishment/Agenda (AG) quadrant and so on.

Are the two axes meaningfully different?

My typology, I think, already addresses criticism (C) at the start: it is more grounded on what we observe are the different ways a candidate is sold to the public. How does it overcome criticism (B), that different candidates actually vary in where they are on both axes?

For one, the idea of there being two types of candidates — “establishment/vision” versus “anti-establishment/agenda” — already seems intuitive among pundits:

As of this writing, Bernie Sanders just began his campaign and he is labelled along one axis already. This is in contrast to the descriptions of Joe Biden, Cory Booker or former MA governor Deval Patrick, who get called “establishment Democrats,” “centrist Democrats” or other such monickers.

There is also talk about a “vision/agenda” axis without referencing whether a candidate is in the establishment. Consider the following:

Here Robin makes a distinction between “agenda” candidates, Sanders and Warren, and the other bunch of candidates as based on “them, their person.” Nate Silver retorts by saying Klobuchar’s campaign is also based on agenda-setting, unlike whatever a campaign not based on agenda-setting is; that would be a “vision candidate” in the new typology.

In contrast, Dave Weigel identifies Kamala Harris and Booker as “vision candidates” because of their faith in an ideal American social order, in contrast with Warren as the campaign not based on that kind of faith; in the typology those who do not campaign on faith in ideals campaign on the ideas underlying their agendas.

A much trickier question is whether or not the two axes are independent from each other: that there are candidates or voters who think they want an Anti-Establishment/Vision or Establishment/Agenda candidate the most. One poll I found can provide some justification: a Morning Consult poll of Iowan Democrats. What’s crucial about this poll is that it asks its sample questions about desirable qualities they want from a nominee.

I divide the 12 quality questions into four groups, each corresponding to one of the four ends on two axes. For example, “Someone who can heal the racial, ethnic and partisan divide in our country” is a “Vision quality,” while “Someone with a long track record and real government experience and know-how” is an “Establishment quality.”

First, I consider the percentage of those polled who treat a quality as “critical.” To test if people are mostly either EV or AG voters, I check whether there is a difference in the percentage of people who think establishment qualities are critical and those who think vision qualities are critical, and vice versa. The reasoning is that fewer people would find both an establishment and antiestablishment quality absolutely critical, ensuring I am measuring people on one end or the other of the axes.

If, say, more people think vision qualities are critical than they do establishment qualities, that suggests there’s a mass of voters who are anti-establishment or do not care about that axis, but who do care about if a candidate offers a vision.

I also check, across groups, the ratio of people who think an issue is critical and people who think the issue is at least “somewhat important.” This supplements the first test I run by validating that the sample spreads out along both axes, so my typology does separate different groups of Democrats.

Group-wise means of two statistics in the Morning Consult poll: (A) percent of Democrats treating a quality as critical and (B) Democrats treating a quality as critical as a percentage of all those thinking the quality is important. Different background colors indicate two separate axes. Vision qualities correspond to Q 26, 33; establishment qualities to Q 27, 34; antiestablishment qualities to Q 28, 31; all other Qs are agenda qualities.

The result may be underpowered, but it is suggestive of my two hypotheses. Observe that maybe a quarter of those surveyed would be placed nearer to each end of the E/A axis, or half of those surveyed are placed closer to the center of this axis. A much larger mass of people care about vision qualities crucially compared to those treating establishment qualities crucially, and vice versa. This implies some “vision voters” are not “establishment voters,” and analogously for “agenda voters” who are not “anti-establishment voters.”

The results’ statistical validity is questionable. The most scientific way to measure where voters are on the two axes is to design a set of candidate quality questions equally spread over all four ends of the two axes. Individual data should show that, for example, people with a strong approval of establishment qualities also strongly disapprove of anti-establishment qualities. Then people can be placed on a two-dimension matrix according to their E/A and V/G preferences, from which we can see the mass of people in each of the four quadrants.

Where are the candidates placed on the axes?

How does the typology address criticism (A), that there is a credible and intuitive way to place different candidates across the axes?

We could assign placement scientifically by using similar processes to how we would assign placement of primary voters. Questionnaires can be sent out to the campaigns or pundits to identify which candidate qualities the campaigns identify with. Someone clever can take the speeches and campaigning material pushed out by the campaigns, mine frequently appearing phrases, assign the phrases to positions on the axes and then assign candidates accordingly. The problem is that neither of those things have been done, even suggestively.

Without any data analysis, I still tried to place candidates following my own intuition. I compare all the stories I have heard about various candidates, like how Biden defends his record of bipartisanship with Republican Senators after he was paid to speak at a Republican congressman’s fundraiser; how Bloomberg poured money into gun violence and global warming campaigns and defends his past support for the NYPD; how Kamala Harris is willing to reinvent her past into that of a “progressive prosecutor”; how Bernie Sanders continues to hammer Trump after the President’s addresses for dawdling on economic inequality; and so forth. The result is below.

A subjective evaluation of presidential candidates across the two axes. Candidates higher up in the chart are “more establishment,” and candidates to the chart’s right are “more agenda-based.” Larger candidate bubble sizes also represent subjective evaluations of higher influence in the primary (due to name recognition, etc.)

Consider Kamala Harris. She calls herself the “progressive prosecutor:” at the same time, others wonder if she is “liberal enough” and if she’s “hard to define ideologically.” But stepping aside from ideology, we know she was close to San Francisco Mayor Willie Brown and that she earned the Bay Area Democratic Machine’s approval, if not from national politicians. She started her campaign with a call for unity and is talking up her Jamaican roots, though she answers questions about her prosecutorial record if she gets them.

Even if it’s hard to tell which liberal policies she cares about, I feel confident calling her an establishment/vision type right now, albeit one hedging closer to the center.

You may object here that the way pundits usually see Harris’s campaign is that it operates on identity politics, emphasizing her role as a successful black female public servant, so is this not a case where the two axes can’t capture a candidate’s distinctiveness? This criticism is valid, but I think the identity politics angle is getting overblown because it’s early in the game.

I am not denying, with this typology, that there will be competition to be the “black voters’ candidate” or “the millennials’ candidate.” What I am denying is that, once a candidate has captured a core constituency of black or millennial activists, they will identify themselves as “the X voters’ candidate” to the primaries’ end. When only a few candidates remain in the Democratic primary, I think candidates’ messages will broaden beyond one demographic base. To win a nationwide race, candidates need to develop a personality that will appeal beyond a base and to the entire electorate.

Another perspective is that the sheer number of candidates means that there is competition to be the black voters’ candidate or the millennials’ candidate, unlike past primaries where only one candidate emphasized their ties to each constituency. Harris is competing with Cory Booker for black votes, and we already see a difference in tone between the two: Booker’s relentless optimism, support of bipartisan legislation and Wall Street ties make him much more of an EV-type than Harris. Hence, even within these core constituencies, candidates will differentiate themselves by separating their campaign image across the two axes.

Now we can go back to that three-way comparison between Biden, Bloomberg and Sanders earlier. In the schematic, the three candidates are placed in different quadrants: EV, EG and AG respectively. Notably, Bloomberg is “establishment” because of his extensive donations to existing Democratic leadership, but he is also not backing down on his previous advocacy for certain hot-button issues. His network with the party leadership is something shared by Biden; his refusal to back down on his previous agenda is something shared by Sanders.

There are narratives out there about how different Biden and Sanders’s political careers have been. Biden hashed out many deals as he led two separate Senate committees over six terms of service, while Sanders before 2016 was a gadfly character. Biden has a network of retired politicians and high-spending fundraisers while Sanders fired up a grassroots fundraising organization, and so on. (These are ways in which Sanders’s campaign and message overlaps with Donald Trump’s, even if they are opposites in ideology.)

What the typology predicts is that the two candidates will use sharply different language to describe their respective campaigns and the future with them as POTUS. Another way to look at is that while many Democrats have favorable opinions of both of them, the major financial backers, the staff and the spin doctors for the two campaigns will fight each other tooth and nail.

Consistency and pivoting

It is important to stress that, when I wrote this article, most Democrats

1. Approved of most of the presidential frontrunners, if not all, judging from them having +40-+60 favorability ratings, but
2. When not given a prompt on who they want to be the nominee, a majority of Democrats cannot name the candidate they support the most.

I publish this article with a high degree of belief in the typology. I believe that, though people’s opinions over candidates do not vary greatly, they do vary in what qualities they’re looking for the most in a candidate. I predict that, by 2020, people’s preferences over the four quadrants of politicians implied by the typology will harden; volunteers hoping for an AV candidate will equally bash EV and AG candidates, for example.

If voters’ preferences harden, then there will be an opportunity for a candidate to establish themselves as the leading candidate in one of the four quadrants. You can see from the schematic that I think there are no serious candidates occupying the AV or EG quadrants right now. Therefore, I predict any pivoting in rhetoric by candidates will be EV/AG candidates trying to dominate those other quadrants.

For example, maybe Kamala Harris will shift away from her agenda to unify the country and become focused on a few policy proposals on civil rights, law and order and welfare policy. This would solidify her move into an EG candidate, as well as place her too close to comfort to replicating another EG campaign: Hillary Clinton’s 2016 campaign.

Or Elizabeth Warren, seeing that Bernie Sanders can talk about Medicare For All more than she could, will move towards a campaign message where she vows to speak for the downtrodden, focusing less on her policy planks as instead showing her empathy for the middle class. That would be a pivot by her from AG to AV.

Candidates on the extremes of scales, like Sanders, Biden or Trump, cannot pivot so easily lest they burn their bridges with their donors and supporters. An inability to pivot is a double-edged sword. If it turns out you are the candidate of a minority quadrant in the party there’s not much else you can do. That said, if you commit to not pivoting, you also do not run the risk of losing your base in a pivot while gaining little new followers.

This is the first of (hopefully) a series of articles using a two-axis typology to analyze the 2020 Democratic primaries as well as the 2020 presidential campaign. The goal is that this typology makes more sense than classifying candidates based on “ideology” or “appeal to party segments,” and using it means we can more scientifically track changes in public opinion and campaign strategies as the election comes closer.

In another article, I use this new typology on a case study: how successful would a Beto O’Rourke presidential campaign be? I really do think he could maintain a pole position like Trump did in 2016 quite quickly, and I offer a new argument for why I could be right.