A science of campaigning?

Axelrod and Rove’s Masterclass on campaigns and elections

One of my birthday gifts this year was a subscription to Masterclass which provides online classes taught by experts/celebrities. I was most interested in Penn & Teller on magic and Kasparov on chess. Both were okay, but not as systematic or detailed as I had hoped. 

I noticed that Masterclass also offers a course on campaigns and elections which is close enough to some things I study that I thought I would take a look. I’ve also been thinking about the political science major which I help to administer. It occurred to me that the major should probably provide students with more practical skills. Running a campaign, the heart of this course, could be one of those skills. I was curious whether this course provides a model for the sort of thing that we could offer.1

The course is taught by David Axelrod and Karl Rove. Axelrod became famous working on Obama’s campaigns and Rove gained infamy as the bare-knuckled advisor to George W. Bush, though both have long histories in the business. Axelrod explains the odd-couple pairing as a reflection of his belief that polarization has gone too far in America. There is a personal reason as well -  he bonded with Rove over the fact that both had parents who took their own lives.

Like most of the courses on Masterclass, this one is structured around a couple of dozen 10-15 minute lessons. Many if not most of the lessons take the viewer through the details of a campaign from doing research on one’s own candidate and one’s opponent to preparing a calendar and budget to the basics of polling and focus groups to get out the vote (GOTV) efforts.

My bottom-line impression is that the course does a good job of laying out the basic elements of a modern campaign and much of the accumulated conventional wisdom about what leads to success and failure. I am less sure that the more general claims that the presenters make are necessarily true or backed up by good research in political science. For those reasons, I’d be hesitant to teach something like this as a course, though it could be useful background material for a more analytical class.

The central lesson of the course and the one that binds it all together is that the “most critical” part of a successful campaign is having a strong and consistent argument for why voters should choose this candidate. This is the message of the campaign. According to Axelrod, this argument must be “authentic, relevant, and connecting” to voters. 

Bad campaigns are ones that have a weak argument - missing one or more of these elements - or flit from one argument to another. Negative examples include Hilary Clinton’s 2016 campaign whose central argument was breaking the glass ceiling, in their view a weak argument, and which cycled through different messages. Obama’s argument for change and W’s for restoring dignity were better arguments. In turn, everything else about the campaign - from its launch to the messaging to responses to attacks to the logos and slogans - should relate back to this main argument.

This seems commonsensical, though I can’t immediately think of any political science research that backs it up. Axelrod and Rove don’t really describe the types of arguments that candidates might use or provide tools for assessing which ones are good. They imply that the goodness of an argument depends on the time and place, the specific candidate connecting with the local zeitgeist (which is always changing). More generally, they suggest that a good argument is about doing something, not being something. They sometimes invoke the idea of change versus status quo elections and candidates (Axelrod uses the terms replicas and remedies) though they don’t really formalize this advice - that is, how to recognize these kinds of elections and profit from them. They even manage to explain Trump’s victory despite a chaotic campaign with their messaging theory - in their view, he stayed on message during the homestretch.

This sort of advice makes sense coming from two advisors who focus on messaging, but I have my doubts. I think the main worry is that the theory quickly becomes tautological or post hoc. Winning campaigns must have had authentic, relevant, and connecting arguments because they won. It is hard to identify these things a priori. Clinton may have lacked them, but she also very nearly won.

Most helpfully for newbies, they provide a basic description and skeleton of a modern campaign. This includes the typical functions and offices on the campaign team, different types of polls and focus groups (but not the more controversial parts like push polls), different ways of raising funds and recruiting volunteers, different kinds of messaging (introduction/biography, issue, contrast, attack). And all of this is interspersed with anecdotes from campaigns they have worked on, especially Obama and Bush. One false note is hearing Rove opine on the ethics of oppo research given his history and reputation.

In a couple of places Axelrod and Rove make arguments that intersect with political science research. Here are a few that I picked out and my assessment of them:

  • Both Axelrod and Rove talk frequently about get-out-the-vote efforts and various ways of doing this from knocking on doors (walking in Rove’s parlance) to phone banking to direct mail, but they don’t mention or address the extensive research on the cost efficacy of these different efforts. A summary of some recent research can be found here.

  • Similarly, the classes emphasize the distinction between mobilizing your supporters to vote and persuading non-supporters to be your supporters. But as with mobilization, they ignore most of the research on persuasion. To be sure, the research on persuasion is weaker than that on turnout, but arguably we do know some things. See, for example, this recent volume.

  • There isn’t much recognition of the role of fundamentals - mainly the state of the economy - which play a large role in political science explanations of elections. Axelrod and Rove might argue that since the state of the economy is a given for any candidate, it can be ignored. But there is good research on how candidates run different types of campaigns under different circumstances. I suppose this is captured in their advice to tailor the campaign to the local mood, but I think they could have been more specific here.

  • One claim that I was uncertain about was Rove’s argument that candidates should not tack to the right or left in primaries and then move back to the center for the general election. He claims that this assumes nobody is paying attention to the primary. I understand that there are reputational costs in these moves, but it is not clear to me whether these costs outweigh the benefits in aligning more closely with the electorate in the primary and general elections. I could not track down research exactly on this point as opposed to research showing that candidates tend to align with the primary electorate or that moderation has benefits in the general. Perhaps I am missing something.

  • On the plus side, Rove notes that members of congress tend not to change their positions in exchange for contributions. Instead, they mostly support the interests based in their district. Similarly, special interests tend to target supporters of their positions rather than opponents. This corresponds to Hall and Deardorff’s theory of lobbying as legislative subsidy plus a raft of other research on the ideological consistency of members or on the relations between members and their districts. 

  • They also point out that many voters pay only sporadic attention to politics and need to be addressed anew at the end of the campaign which is when they are finally paying attention. Again, political science backs this up, though it also casts doubt on what is the ur-premise of the whole course that campaigns matter, a point to which I will now turn.

In general, Axelrod and Rove seem to ascribe considerable efficacy to most particular elements of the campaign - for example, a good launch, a good debate, a good slogan/logo - not to mention the campaign as a whole. At one point, they even suggest that the 1980, 1984, and 1988 presidential elections were won or lost based on single lines at debates. (Try to guess which lines - I’ve put the answers below.) And Axelrod has a fun anecdote about the effort the Obama campaign put into having an outdoor launch to its campaign in the Illinois winter. Political scientists, by contrast, have had difficulty finding large effects for campaigns and even less for specific parts of campaigns like debates (for a balanced view, see this piece).

There are reasons why campaign effects tend not to be large. On the one hand, they may cancel out as both sides do the best they can. If one side deviated from a high-quality campaign, then we might see stronger effects. On the other, campaigns may simply help voters to recognize the fundamentals. Vavreck points out how in presidential elections candidates who correctly tie themselves to the economy (incumbents when the economy is good and challengers when it is bad) are more successful. She calls these clarifying campaigns as opposed to insurgent campaigns which emphasize other issues and tend not to be as successful.

One could interpret Axelrod and Rove’s advice as using the campaign to help realize the fundamentals or as trying to minimize the shocks to these fundamentals. Many of the practices they describe are intended to head off problems like a retweet from a disreputable source or to create problems for their opponent like turning up a record of non-voting. Most research would attribute fleeting effects to such efforts; they would mainly be useful towards the end of a campaign when the effects don’t have time to decay.

In terms of learning to be a campaign operative, it is noteworthy that in the view of this Masterclass campaigning is mainly something that one learns on the job. It is an apprenticeship where the important thing is to jump in and start contributing, something they recommend several times. There doesn’t seem to be much of a role for academic knowledge or even technical skills (one hires experts for these sorts of things). I think this was likely true in the past, but I believe that it is changing. Consider this website devoted to training in “progressive analytics”, which discusses a variety of programming and statistical skills that volunteers can learn.

Axelrod and Rove do acknowledge some of these skills. They mention the idea of microtargeting and the use of randomization to test the efficacy of direct mail appeals, but they mostly seem to be operating in the pre-Money Ball era (at least for the purpose of this course). There is thus more of what seems like homespun wisdom: Make sure to write personal thank you notes to donors. The counterpunch is mostly stronger than the attack. Stay on message to win.

The class concludes with a conversation on the success of Trump and the current state of American politics. They agree that American life has become too tribal (though again this is a bit rich coming from Rove). They attribute Trump’s election to the fact that it was a change election and Trump represented change. (They also note that Clinton was probably the only candidate that Trump could beat.) As mentioned, they agree that Clinton ran a poor campaign and blame her loss mainly on the campaign. They disagree on the power of Fox News (where Rove is a contributor) and whether its power is asymmetric (Rove argues that MSNBC is equivalent). They are short on solutions except for being nicer, relying on the cyclical nature of our system, and believing that successful politicians are more about adding new voters than riling up the base.

Overall, the course is a reasonably entertaining introduction to the main elements of a modern campaign. It provides the basic architecture and a sense of most of the key moving parts. Axelrod hopes that emphasizing the difficulty of a campaign might create more respect for politics. I am not sure about that, but it shows how easily things can go wrong. Whether those mistakes matter in the big picture is an open question. I’m pretty sure that this course wouldn’t substitute for an academic class on the subject - there is too much post hoc propter hoc reasoning and not enough analysis - but it might be a useful supplement, especially in showing what the profession of campaign management regards as best practices.

Key debate lines:

1980: Reagan - There you go again.

1984: Reagan - I will not make age an issue of this campaign. I am not going to exploit, for political purposes, my opponent's youth and inexperience.

1988: Dukakis - His response to a question on his opposition to the death penalty in the case of his wife being raped and murdered.


I think the kind of class I would really like to see is closest to this one taught by Alexander Fouirnaies at the Harris School.