The “do what you want” theory of politics

Why embracing “Abolish ICE” and Medicare-for-all won’t doom the Democrats.

An unspoken assumption of most political punditry is that the political positions taken by, and the policies supported and enacted by, politicians play a significant, perhaps decisive role in determining the outcomes of elections.

This is the premise of basically every piece of commentary about, for example, Alexandria Ocasio-Cortez’s surprise victory in the Democratic primary for New York’s 14th District. To fellow democratic socialists, Ocasio-Cortez’s victory is evidence that ideas like Medicare-for-all or a job guarantee aren’t just popular in opinion surveys: They can win elections. Even radical-sounding ideas like abolishing Immigration and Customs Enforcement might fly! To conservatives and squishy moderates, Ocasio-Cortez’s defeat of a 10-term incumbent is proof that Democrats are willing to commit electoral suicide for the sake of ideological purity.

Each of these arguments has specific problems. But all of them share one big issue: They dramatically overestimate how much the actual issue positioning of candidates matters for how people vote.

What I want to propose is a null hypothesis for political punditry: Outside of truly extreme proposals, there’s basically no plausible position a politician or political party can endorse or enact that will have a meaningful impact on their likelihood of retaking political power. The US has for decades had a stable system where liberal and conservative policy coalitions (which have sorted out under the Democratic and Republican parties, respectively) semi-regularly alternate in power, with long periods of divided rule and gridlock in the middle. Dramatic shifts in the ideological makeup of both parties during that same period did not upset that alternation of power. It continued apace.

The upshot of this phenomenon is that parties should be a little less nervous about sticking to their guns and arguing for what they believe, whether or not it polls well. Call it, if you like, the “do what you want” theory of politics.

It’s possible to push the “do what you want” theory to ridiculous extremes. Obviously if the Democratic presidential candidate were to suddenly start calling for dissolving Congress in favor of decentralized rule by workers’ soviets, that would probably hurt them. And on certain issues, particularly race, American voters’ baseline apathy tends to fade. There’s a reason even the most left-wing Democrats don’t tend to emphasize the need to integrate public schools through busing anymore. That’s something that white voters will wake up to stop.

But as a baseline position, I think assuming a null effect is a more reasonable guess than assuming that voter preferences are heavily influenced by candidates’ issue statements. We just have too much evidence that this isn’t how voters really make their decisions.

Instead, we see evidence that Democrats and Republicans exchange power at regular intervals, in spite of massive changes in the beliefs of those parties’ elected representatives. Maybe it’s time to argue that parties should adopt positions by arguing for those positions on the merits, not because they’re electorally useful or mandatory.

What we know about American voters

My basic mental model is that the typical American voter thinks about national politics and elections with roughly the frequency I think about professional football (I’m borrowing a bit here from the political scientist Jonathan Bernstein).

I have a football team that I root for; I maybe check in on standings two or three times throughout a season. I watch the Super Bowl. But I don’t typically watch other games, I couldn’t name many players (even on the Seahawks), and if you asked specific questions about football strategy, about which players should be traded or whether the Seahawks should focus more on developing their wide receivers or their running backs, I wouldn’t really be prepared to give you an answer.

And that’s fine! I have other stuff going on that it turns out I’d prefer to spend my time on. And while the stakes of electoral politics feel startlingly real if you’re a naturalized citizen facing a vociferously anti-immigrant government, or a black family in Flint, Michigan, whose water has been poisoned, or a trans woman forced by a state government to use men’s restrooms, for Americans outside marginalized communities, politics can feel like a game to which you can be indifferent.

Tufts University political scientist Eitan Hersh has described much political activism in 21st-century America as a kind of hobby. “The stakes in political activity can sometimes seem low. In a large republic, an individual’s contribution is almost always non-pivotal. Policy in the U.S. often changes very slowly,” Hersh writes. But that’s what makes it an ideal hobby for some people: “Low stakes are what make hobbies restorative … they are a release from the pressures of work and other obligations.”

A natural corollary of something being treated as a hobby is that other people can simply choose to not have that hobby. This is the camp into which most American voters seem to fall, and if you’re in this category, then you’re not going to be obsessively watching to see if, say, a candidate supports abolishing ICE or Medicare-for-all or what have you. You just don’t care enough. It’s not your hobby.

“Most people have strong feelings on few if any of the issues the government needs to address and would much prefer to spend their time in nonpolitical pursuits,” University of Nebraska political scientists John Hibbing and Elizabeth Theiss-Morse write in their book Stealth Democracy: Americans’ Beliefs About How Government Should Work. “The people as a whole tend to be quite indifferent to policies and therefore are not eager to hold government accountable for the policies it produces.”

There’s tons of research reaffirming this finding. The University of Michigan’s Donald Kinder and Louisiana State’s Nathan Kalmoe show in their book Neither Liberal nor Conservative: Ideological Innocence in the American Public that most Americans don’t really have stable ideologies in a way that matters. This isn’t an original insight of theirs; they view themselves as replicating the work of Philip Converse, who laid out a similar argument in 1964.

American voters aren’t down-the-line liberals or conservatives the way the people they elect are. Astonishingly, what a respondent said their ideology was — liberal, conservative, moderate, etc. — had “little influence over opinion on immigration, affirmative action, capital punishment, gun control, Social Security, health insurance, the deficit, foreign aid, tax reform, and the war on terrorism.” Ideology seemed to matter on LGBTQ rights and abortion, but even that went away after they controlled for religion.

Kinder and Kalmoe looked at a study that asked the same people questions about politics in 2000 and 2002. The finding was even more astonishing. “If you asked an average voter in 2000 whether they were liberal, moderate, conservative, or none of the above, their answer was only 63 percent predictive of what they’d tell you two years later,” my colleague Ezra Klein summarizes. That’s wild. That’s a two-year period — if people have durable political beliefs, you should expect 90-plus percent of them to say they’re liberal in 2002 if they said they were liberal in 2000. That’s not what happens.

Berkeley political scientists Sean Freeder, Gabriel Lenz, and Shad Turney conducted a related study measuring stability in views on individual policy questions, and examining whether voters were able to correctly match policies with politicians. They find that only 20 to 40 percent of Americans “hold stable preferences on salient economic public policies.” In one mid-’90s survey they review, only 19 percent of respondents could correctly answer five simple questions about where the parties stand on abortion, defense spending, government services/spending, guaranteed jobs, and whether the parties were liberal or conservative. Another 18 percent got four of the five right. Most respondents, however, were fairly ignorant.

This perspective is not unanimous among political scientists; UChicago’s Anthony Fowler has issued a forceful paper arguing that policy voting — people holding coherent opinions about what policies they want and voting based on those opinions — is more common than the above research indicates. But the weight of the evidence suggests, to me, that voter ignorance is the norm and that relatively few voters have the kind of stable policy views you’d need to have to vote on the basis of candidates’ issue statements and voting record.

“Numerous studies have demonstrated that most residents of democratic countries have little interest in politics and do not follow news of public affairs beyond browsing the headlines,” Vanderbilt’s Larry Bartels and Princeton’s Christopher Achen conclude in their 2016 book Democracy for Realists. “They do not know the details of even salient policy debates, they do not have a firm understanding of what the political parties stand for, and they often vote for parties whose long-standing issue positions are at odds with their own.”

The best reason to think issue positions matter (and why I disagree)

The finding that most voters don’t have stable opinions on policy questions is not, on its own, enough to prove that politicians can hold whatever policy opinions they want without any electoral consequences. It could be the case that the 20 to 40 percent of people who do have stable opinions on economic issues are swing voters who use those preferences to determine the outcomes of many elections.

But that doesn’t seem to be what’s happening. Michigan State professor Corwin Smidt has shown that the share of “floating voters” — voters who switch their party allegiance from one election to another — has plummeted in recent decades. In 2012, only 5.2 percent of Americans voted for a different major-party presidential nominee than they had in 2008; from 1952 to 1980, the average rate was 12 percent, indicating that swing voting has fallen by more than half (Smidt’s study predates the 2016 election, when there appears to have been a greater number of swing voters). Smidt doesn’t find that rates of floating voters were higher among the politically ignorant, but it’s still a quite small group, and the share of it that is highly informed and has stable preferences is smaller still.

The best evidence I’ve seen that a critical mass of policy-sensitive voters exists, such that parties would be making a huge mistake by ignoring their candidates’ policy positions, comes from Stanford political scientist Andrew Hall. In two papers, one solo-authored and the other with Stanford’s Dan Thompson, Hall looked at a number of different closely contested primaries for US House elections, in the first paper from 1980 to 2010 and the second from 2006 to 2014. He specifically analyzed “coin flip” elections, where the moderate candidate barely defeated the extremist or vice versa.

His big finding was that a party picking the extreme candidate hurts that party — a lot. In the 1980-2010 study, he found that nominating an extremist cost the party about 9 to 13 percentage points of the vote, and reduces the odds of victory by 35 to 54 percentage points. “These,” Hall writes with almost hilarious understatement, “are large effects.” His study with Thompson clarifies that this seems to happen not because swing voters are turned off by a candidate’s extreme positions, but because that candidate’s presence mobilizes the other party’s base more than it mobilizes that of the extreme candidate.

These are extremely well-designed studies, and I learned a lot from them. But I’m not sure they prove that extreme positions on specific issues themselves hurt candidates. Hall estimates ideology by using candidates’ donors, assuming that candidates with similar donors have similar viewpoints. This is a clever methodology, pioneered by Stanford’s Adam Bonica, but it has its limits.

As Hall and Thompson discuss, this measure doesn’t correlate perfectly, or even particularly well, with actual roll-call voting in Congress. The donor-based ideology measures might tell you something about which faction of donors are backing a candidate — whether, say, a Dem is getting Wall Street backing rather than union backing, or a Republican is getting Tea Party-linked money or more traditional business interest money — but that doesn’t always tell you precisely how they’re going to vote. That makes the data less than determinative when you’re trying to figure out if taking individual positions, of the kind that show up on roll-call votes, hurts or helps candidates.

“The goal is not to isolate the ‘causal effect’ of candidate positions, themselves,” Hall and Thompson write. “In fact, it is not even clear that there is such a thing as a causal effect of candidate positions.”

Moreover, these are studies about contested primaries. The finding that an extremist narrowly winning endangers a party in a general election jibes with examples like, say, Tea Party activist Christine O’Donnell defeating moderate Rep. Mike Castle in the 2010 Republican primary for Joe Biden’s Senate seat in Delaware. Given polling showing that Castle would almost certainly win if nominated, it makes sense to think that nominating O’Donnell caused the party’s defeat.

But much current discussion concerns, instead, the choice of incumbents such as Sens. Kirsten Gillibrand (D-NY) and Elizabeth Warren (D-MA) to come out for controversial policies like a job guarantee or abolition of ICE or single-payer health care. It’s not clear to me that in the absence of a primary campaign that causes voters to learn to view such positions as “extreme,” such moves would have much of any electoral effect at all. Indeed, the micro evidence on voter behavior makes me guess there’d be little to no impact.

Policymaking in a world of indifferent voters

It’s less precise evidence than that limned above, but arguably the best case for the “do what you want” theory of politics is the mere fact of alternation of power.

“In well-functioning democratic systems, parties that win office are inevitably defeated at a subsequent election,” Bartels and Achen write. “Moreover, voters seem increasingly likely to reject the incumbent party the longer it has held office, reinforcing the tendency for governmental power to change hands.”

There are some exceptions to this rule, in which a party becomes dominant for extremely long periods in otherwise democratic countries — the longstanding dominance of the Liberal Democratic Party in Japan, the Social Democrats’ rule of Sweden from 1936 to 1976, the Conservative Party’s 1979-’97 period of control in the UK — but in the US, no party since World War II has ever held the White House, or even won the popular vote, for more than three terms in a row.

That is despite the fact that from 1945 to the present, the Democratic Party underwent a dramatic shift toward a more egalitarian stance on race issues, shedding its Dixiecrat base in the process; that it largely abandoned traditional labor politics after the 1984 election; that the Republican Party moved right on race and very far right on economic issues toward a more stridently laissez-faire stance; and that the electorate itself has changed its composition dramatically in demographic terms.

All that — the whole history of post-World War II American politics and the grand ideological shifts of parties it included — was not enough to disrupt the basic fact of power alternating hands. Even when the House was firmly under Democratic control, the presence of conservative Democrats meant that Republican presidents like Ronald Reagan could pass their agendas. That implies that parties can undertake rather large ideological shifts without jeopardizing their chance of one day, eventually, taking over again.

In a world where this is the reality, and where the public is largely indifferent to public policy, mass politics and attempts to affect public opinion become a lot less important. Instead, the main lever of influence is lobbying of party elites. Those elites might mistakenly believe that public opinion matters and so public opinion polls become a compelling way to lobby them, but ultimately, if you want to abolish ICE, you don’t need to persuade the American people. You need to persuade Nancy Pelosi and Chuck Schumer. Shifting public opinion might help with that, but it’s only a means to an end.

Conversely, if you want to dramatically cut taxes on corporations, you don’t need the American people on your side — and by God, the Republican Party did not have the American people on their side when they cut corporate taxes in 2017. Insofar as Americans cared, they hated the idea of giving big companies a tax break. But Republicans did it anyway. Convincing Paul Ryan and Mitch McConnell to give it a go was, it turned out, enough for corporate tax cut advocates.

Ryan and McConnell might or might not have thought that the cuts would be popular. They might not have cared at all. They probably knew that majority parties are always doomed in the midterms, and decided to do what they wanted anyway, since their time in power would be brief. Democrats could learn a lot from that example.