This is an archive post from my off-the-record “Weekend Reading” newsletter, edited by Emily Crockett and with research assistance from Andrea Evans.
Earlier this year, in what now constitutes a pre-buttal to popularism, I began a series called [Mis]understanding Elections, which methodically debunks popularism’s essential assumptions. Ironically, the as yet unwritten Part IV was set to lay out why reasoning from those assumptions holds us back from realizing our democratic aspirations.
Since the publication of the Ezra Klein piece about popularism and its most public spokesperson, David Shor, many have published strong critiques. I have struggled with writing something that feels comprehensive. On the one hand, the underlying issues here can fill, and have filled, entire books. On the other, the continually and rapidly evolving nature of the public conversation about popularism means that any extended statement about it risks feeling somewhat outdated the next day. But based on where we are right now in this debate, this longer-than-usual Weekend Reading lays out what is most important about it. What follows also benefits from the tireless efforts of Emily Crockett and Andrea Evans to apply much of what I’ve already written to the current debate.
First, before delving into the nitty-gritty deconstruction of popularism, I will start by refocusing the conversation on affirmative points that I consider more important than just debunking a bad idea (even if that bad idea is very bad and has disturbingly wide currency). Then I will explain what popularism is (a less straightforward endeavor than you might think) and what the biggest problems with its claims are. Then I will discuss what the popularism debate says about our current politics, observe how popularists sabotage the party they say they want to help win, and discuss why we keep getting fooled by poorly reasoned but smart-seeming arguments like this. I’ll conclude with some thoughts on what the real path forward for Democrats in 2022 and beyond should look like.
I want to foreground that while there’s some pretty geeky stuff ahead, I am dead serious when I say that the popularist project can do concrete damage to Democratic goals and priorities. Popularism leans on common but false assumptions about how politics works, assumptions that not only limit our imaginations for a better future but also discourage us from pursuing the proven electoral strategies that we know can help us win. Popularism accepts the stereotypes of Democrats that underlie Republican caricature, which makes it harder to counter those stereotypes and connect with voters. The way popularists make their case can’t help but divide us at the precise moment when it’s most important for us to be united against the MAGA Faction. The Republican Party is now dominated by a faction so committed to restoring the cultural dominance of its base that it is quite literally willing to destroy American democracy in the process.
Popularists say that failure to follow their advice amounts to strategic malpractice that will cause Democrats to lose. They claim that their prescriptions are only a means to an end, and that the best way to respond to Republican dog whistles is by ignoring them. This, they say, will help Democrats win elections and gain the power to not only make the racial justice reforms they care about, but beat back the specter of fascism and minority rule. However, candidates who insist that their most important issues are left unsaid rarely keep their promises about such commitments after they are elected. Jamelle Bouie explains this in a recent New York Times piece, and there is no particular reason to believe it would go better today. Precisely because the stakes are so high, Democrats cannot afford to be distracted by the false promises of popularism.
Refocusing the Popularism Conversation
Popularism is to politics as neoliberalism is to economics. Popularism assumes that politicians respond to voters as corporations do to consumers, and both center rational actors and consumer choice as the sovereigns in determining outcomes. Neither theory corresponds to reality.
The most important point I had planned to make for Part IV of “[Mis]understanding Elections” has been made by others who understand that everything in American politics is about race, except race, which is about power.
Thomas Piketty writes that:
Every human society must justify its inequalities: unless reasons for them are found, the whole political and social edifice stands in danger of collapse. Every epoch therefore develops a range of contradictory discourses and ideologies for the purpose of legitimizing the inequality that already exists or that people believe should exist. From these discourses emerge certain economic, social, and political rules, which people then use to make sense of the ambient social structure. Out of the clash of contradictory discourses — a clash that is at once economic, social, and political — comes a dominant narrative or narratives, which bolster the existing inequality regime (Capital and Ideology.1)
Neoliberalism and popularism are both system justification narratives in that sense.2 They begin by asking us to believe that as voters and consumers, we are making the world we live in. That inequalities are the result of the unstoppable natural forces popularists have “discovered,” rather than the realized intentions of the powerful; that the only way to resist is to act individually as voters or consumers choosing from the selections elites put on offer rather than joining together for something different.
For this reason, we have two parties that debate whether to tax billionaires, and none that question whether we should have billionaires in the first place. Popularists who argue that Democratic campaign decisions are being made by out of touch college elites are correct in the sense that many elites in both parties can’t see past the market-driven logic they learned in college. Yet the popularists themselves demonstrate that they are out of touch with the dominating realities of the voters they claim to speak for—namely, increasingly precarious circumstances and shrinking limits to their hopes for themselves and for their families’ futures. Popularists advise us on how to appeal to the working class without ever talking about work.
Education Polarization and the Lost Cause
Popularists use “education polarization” (more on this later) as an incantation to avoid reckoning with the fact that, since the enactment of the Civil Rights and Voting Rights Acts, the Republican Party has campaigned in every election by showing its solidarity with those voters unwilling to accept full equality for Black Americans (and others). Especially when cultural threat is stoked, politics quickly becomes a matter of identity. That’s why politicians campaign by saying they are on “your” side against “them.” For Republicans, “your” has generally referred to white people and “them” has referred to Black people. For Democrats, “your” has meant working people and “them” has meant the corporations or the rich.
All but the most racist Republicans claim that their objection is not to racial equality in principle, but to outsiders telling them how to achieve it. That claim loudly echoes the Confederate Lost Cause narrative—that the Civil War was not about slavery, and that Reconstruction failed because of carpetbaggers and scalawags rather than a sustained just-born KKK’s terrorist campaign to extend the Civil War as a guerrila war. The popularist account of the last 55 years accepts that Lost Cause mentality when it blames those seeking racial justice, rather than those resisting it, for how divided we are now. This explains why popularists use “culture” to obscure the fact that the active ingredient in MAGA is the resistance to racial and gender equality which is the “great again” they insist America returns to.
An Intramural Battle
Popularism is not about what Democrats should say, but about who Democrats should listen to. Like the Republicans writing the 2013 autopsy, popularists are using the language of demographics and data to wage an intramural fight. It’s worth giving that autopsy a reread and seeing how similar its arguments are to today’s popularist gripes. In both instances, the advice-givers claim their party’s problem is an echo chamber of addled activists, and that electoral success depends on saying popular things and never saying the quiet part out loud. Today’s popularists do very little to help Democrats win or seek to persuade those Democrats they disagree with, and spend more time heckling their copartisans while rejecting invitations to discuss their arguments or acknowledge substantive criticisms.
Suckers for “Science”
It’s important to call out what this episode should remind us about how and to whom we assign credibility. Popularists are no more than pundits who successfully use their superior facility with and access to data to name themselves “data scientists.” Popularism makes claims like that there is “physics of elections” or that “polling is the only way to see outside ourselves” to make its conclusions sound objective and inevitable and make its critics sound like they won’t face facts. In reality, the popularists are the ones engaging in motivated reasoning, cherry-picking data (when most of the cherries were grown in their own gated orchards), and giving unsound advice based on their own intuitions, policy preferences, ideological priors and commercial interests.
Even though they burn us almost every time, we are often suckers for people offering mathematical-seeming answers that promise more certainty than is possible. This is how we swallow whole McNamara telling us that we were winning the Vietnam War and just another few hundred thousand troops would do the job, economists insisting that trickle-down and tax cuts would “lift all boats,” financial analysts and experts to tell us that deregulating finance would bring nothing but prosperity, and political forecasters to convince us that Trump couldn’t win the nomination or election, or that Biden would win by more than he did. Forecasters are closer to shamans than scientists, though they are treated more like the latter.
Those who know me (or been reading Weekend Reading), know that I am hardly uncomfortable with data. On the contrary, I am very proud of the data related infrastructure I’ve been fortunate enough to play a role in building. But, I hope it’s also been clear that I see data as a way to develop testable hypotheses and to attempt to validate strategic propositions. There is no unified field theory of politics. Data should make us humble because it allows us to test ourselves rather than be arrogant when we insist that data and poll-driven paracosms create a tractable reality. Popularists are to politics what Madden video game players are to professional football.
James Baldwin said, “Not every problem that is faced can be changed, but nothing can be changed if it is not faced.” While we don’t know if we can make the changes our country and world desperately need, we can be assured of failure if we refuse to acknowledge the true nature of the problems we face. By creating more excuses to delay that reckoning, the Sturm und Drang over popularism can only be cause for despair.
Popularism, Briefly Explained
Before laying out the summary of popularism that will inform this piece, it’s important to make clear that popularists make it nearly impossible to refute their arguments directly since there is no concrete popularist manifesto—just a collection of interviews and tweets, along with references to data pulled from their own proprietary polling which they never share. Thus, popularists routinely move the goalposts to suit the conversation and context they’re in. It makes their claims inherently unfalsifiable.
With that in mind, here’s a rough overview of the popularist paracosm.3
Popularists begin their argument with two observations that are completely uncontroversial, but that they suggest are original contributions to the strategic discussion:
For Democrats to win a majority in the Senate and the Electoral College, they must win in states that are more Republican than the nation as a whole; and
The near extinction of ticket splitting means that hostility to national Democrats sets the ceiling for Democrats running in those states.
The arguments that come next are the problem.
For popularists, the story of American politics centers around education polarization: the fact that over the last several decades, college-educated voters, especially white college-educated voters, have been increasingly voting for Democrats, while voters without a college degree are increasingly voting for Republicans. This trend has supposedly led the Democratic Party to cater too much of its platform to the policy preferences of college-educated people, alienating people without four-year degrees who have different values.
According to popularists, the particular kind of voter we need to reach is the “median voter” who swings elections and holds positions that are more conservative than those of most Democrats. The best way for Democrats to win this coveted median voter—who popularists suggest is a 50-something white man without a degree living outside a city—is by emphasizing the “popular” policy positions that he agrees with and avoiding being associated with any “unpopular” ideas that he disdains. These “unpopular” ideas are usually associated with “cultural” issues, which reflect the values of the woke white elites who are said to control Democratic messaging but also trigger racial anxiety in the median white voter.
The example of an unpopular issue that popularists use most often is “defund the police.” However, both Biden and Sanders opposed this idea strenuously. That means popularists are describing a predicament in which no matter how aligned national Democratic Party leaders are with the median white voter, Democrats will always be doomed as long as activists use slogans that Republicans can exploit to trigger “cultural” anxieties.
If we acknowledge that what popularism deems as unpopular tends to overlap with what is likely to invite backlash from white voters, we can see this as a rhetorical sleight of hand—pretending that Democrats’ commitment to racial justice is being driven by over-educated white elites instead of by authentic grassroots activists, in particular activists of color who are fighting the police brutality and systemic racism that threaten their lives and communities on a daily basis.
The Real Dividing Line in American Politics
The popularist analysis has many basic and fatal flaws. Education polarization is a red herring; the “median voter” doesn’t exist; and the “popularity” of an issue is neither fixed nor terribly relevant to how voters actually behave. We’ll explore all of those specific points in a moment.
But first, it’s important to discuss a basic, crucial truth about American politics that the popularist narrative leaves out: The real story, as political scientists have underscored, is about status threat. There are libraries of books and peer-reviewed work that conclude as much as anything can be concluded in social science that stoking status grievance—particularly by restoring a sense of cultural dominance for those who ascribe to a ethnocentric hierarchical view of society—has been the core Republican message since 1968. When we consider the narrative from a historical context, it is obvious that education polarization is actually a proxy for status threat—a racialized component of cultural identity.
Popularists conclude that much of partisan sorting since 2008 is a result of education polarization, driven by out of touch, ultra-liberal white consultants. They ignore the obvious: In 2008, America elected its first Black president. In 2016, Hillary Clinton ran to become its first woman president. Her opponent was a virulent racist and misogynist who saw the kettle of racial resentment that the Tea Party brought to a boil while Obama was in office, and then poured it into a campaign steeped in ethnic conflict. Racism and misogyny were front and center in the election Trump won—issues that were made salient for ethno-centric voters by first Obama and then Clinton.
The base of the MAGA Faction, those of Trump’s supporters who are particularly drawn to his reactionary messaging and the call for America “again” to be a nation devoted to White Christian Nationalism, fear losing their position in the cultural hegemony. It is this fear, not a lack of education or wealth, that draws them to the seductive lure of authoritarianism and makes them willing to forgo free and fair elections altogether to maintain power.
What Popularism Gets Wrong
Education Polarization: The Numbers Don’t Add Up
As any data scientist understands, relationships do not indicate causality. Before we accept that education level is what is polarizing voters, we would have to account for why the majority of white men with a 4-year degree voted for Trump in 2016, or why a three-quarters majority of Black, Asian, and Latino voters without a college degree vote Democrat. There is a plethora of valid explanations for each of these trends, but the important point is that when factors like race, gender, and religion are included, education is no longer a predictor. This exposes a central flaw in the education polarization premise: There is no substantiated theory for why education polarizes people, and thus popularism cannot account for why, despite contradictory data, education polarization should be accepted as an explanation for how people vote. We only find it persuasive because of our white-centric stereotypes of college and non-college Americans.
Popularists urge us to look at the data. If we do that, we find that no academic statistical model for partisan choice finds educational attainment to be a significant factor after controlling for religion, urbanity, and region.
If popularists are correct, then we should find the greatest movement to Democrats in congressional districts where college education voters are the greatest share of the electorate. The following table arranges congressional districts by percent of voters with a college degree and counts how many districts saw the Democratic vote increase in each quintile. Of course there is a relationship between college education and Democratic vote share, but it’s important to see that between 2012 and 2020, Democrats did better in half the CD’s in the lowest quintile, and in 66 percent in the top quintile. Overall, 56 percent of CD’s were consistent with the popularist assertion (greater than average college education corresponding to an increase in Democratic vote). Again, remember that popularists claim that education polarization is the most important thing we should be paying attention to. But even this simple take shows that even if education polarization were a driver (and not a covariate) it would still leave substantial variation unexplained.
Demographic Labels are Sticky, but also Sloppy
When you step back, it should be obvious and sensible that factors like religion and urban/rural culture, as well as the legacy of partisanship in an area, can explain much more than whether or not someone has gone to college. Still, like too many political pundits, popularists tell their story through the lens of demographics. Thus, nearly every explanation begins and ends with voters defined by their race/ethnicity and/or educational attainment. (Occasionally, age and gender are brought into the story as well, but that doesn’t change my point.)
The demographic model is so appealing for two reasons:
Easy comparisons over time. From a practical point of view, demographic categories are about the only ones that enable comparisons across long periods of time. They are stable, defining an attribute that remains unchanged over time. (From a statistical perspective, categories like gender mean the same thing to the Census Bureau that they did in 1800, and very few people can change their designation.) And they are available—tracked by the Census and/or easily asked in surveys. Counts of Americans in each of the universally chosen demographics are the most accurate data we have, which means demographic categories enable the most accurate apples to apples comparisons we can make.
Archetypal thinking. We love to think in terms of archetypes, imagining individual exemplars of each broader demographic category. This often causes us to conceptualize the electorate as consisting of just a few types of voters. For example, when someone invokes the white non-college voter, nearly all of us think of the burly white guy in an upper Midwestern diner. Understanding the electorate in terms of those exemplars makes politics seem tractable. Additionally, archetypes serve a useful purpose for system one thinking and heuristic bias, making the use of exemplars an effective tool for politicians who wish to harness the power of prejudice. (Obama was deliberately used as an exemplar by the right.)
But the demographic model doesn’t work, because rarely do demographic categories provide substantial explanations for voting behavior. Klein writes that both he and Shor think of education polarization as “a crude measure of class polarization,” because “education tracks facets that paychecks miss” when it comes to how class works in America. While talking about “class polarization” does point to something real, they mention it as a rhetorical strategy to take advantage of our preconceptions about class to give their claims more credibility, they ignore those aspects of class when they use “educational.”
To believe that broad categorical designations such as education play a greater role in geographical divides than the specific culture of the region misses the two distinct political cultures that have been a constant throughout our nation’s history, transcending partisanship. Further examinations of demographics, coupled with an understanding of historical political cleavages, show us that knowing which side a state was on in the Civil War and knowing the percentage of white Christians in the state provide nearly a third again as much explanatory power as knowing the shares of white college, white non-college voters in the state. Joe Biden won the Union states by 13 points while losing the Confederate states by 7 points, despite there being more white non college voters in the former and more people of color in the latter.
White Evangelicals vote for Republicans 4:1, and Black Americans vote for Democrats 9:1. By contrast, “white and college-educated” describes about a quarter of the electorate that, even after recent shifts, splits close to 60-40 between Democrats and Republicans. Meanwhile, adding just one other category to white college/non-college identity—religious affiliation—reveals that white non-Christians, regardless of education, have been trending Democratic and are indeed a Democratic base group.
Even if we do want to ascribe critical importance to broad demographic categories, education just isn’t as important as other elements when looking for partisan correlations. For instance, gender is notably missing from the popularist story. Men’s share of Obama 2012 voters dropped by 2.2 points in 2016 and rebounded by an identical 2.2 points in 2020—more likely explained by the fact that their 2016 nominee was a woman than by some elitist messaging misfire.
Popularism Is Wrong About How Voters Make Decisions
Popularist David Shor tweeted that while it’s “true that voters are not policy wonks,” the “main tool politicians have to communicate their values is by issue taking.”
“Issue taking” is not an established political or political science term, but Shor seems to be referring to candidates taking positions on issues and communicating those positions. I’m not aware of any political scholar, successful candidate, or campaign manager who believes that this is the “main tool politicians have to communicate their values.” Simple observation should be enough to see that’s not close to being true. Policy communication is not what made Trump popular.
In politics, the typical analysis of what is “popular” relies on public opinion surveys. However, people are very bad at knowing their own motivations for doing things, much less accurately explaining them to a pollster. Even if they can, answering a poll question gives their views in a vacuum, unmediated by other factors in the real world. Passive approval or disapproval of issue items on a survey just isn’t how real people decide on who to vote for. Their real decision-making process is complicated, often based more on identity and emotion than rational calculations about policy, and heavily mediated by external influences and institutions.
Such cue-taking is one reason that public opinion shifts over time, such as on critical issues of justice like civil rights and marriage equality. The more people hear from trusted sources that these issues matter, the more they accept that they do.
The idea that Democrats can change voters’ fundamental intuitions about them by tweaking their message on some issues and staying silent on others doesn’t comport with a reality where Republicans are constantly lying about what Democrats believe. Remember that right-wingers have engaged in decades of ruthlessly efficient propagandizing to lay the groundwork that makes their current set of lies about Democrats seem so plausible to so many Americans. Untangling and countering that is a massive project, and no serious conversation about Democratic messaging can ignore this reality.
In August 1980, Ronald Reagan, running against Jimmy Carter, a devout Southern Baptist, told the more than twenty-five hundred conservative evangelical pastors from forty-one states gathered in Dallas for the for the National Affairs Briefing Conference, “I know this is non-partisan, so you can’t endorse me, but I want you to know that I endorse you.” And, well, we know how that turned out: The Christian Right joined the Republican electoral coalition.
The “Median Voter” Is a Discredited Rhetorical Trope
Although thoroughly debunked in political science, the notion that there is a median voter who decides elections is tempting. (See for example, Christopher Achen and Larry Bartels’ Democracy for Realists as well as the work of Kenneth Arrow who won a Nobel Prize for, among other things, showing mathematically that the “median voter” couldn’t exist.)
Think about the real world in which voters have to choose between only two parties. In the real world, different voters have different priorities and vote for the candidate with whom they align on what matters most to them, rather than some kind of imaginary weighted average of issues. Thus, you have some voters who went for Trump because of his racism, and others because of the tax cuts. To the former, Trump’s position on tax cuts is not determinative but also not discouraging, and vice versa.
In reality, instead of a single-file line of voters, there are two fixed voter blocks. With only two major-party choices on the ballot, individual voters who disagree with each other about many things routinely vote for the same candidate because their party affiliation is what matters most to them. Even if the parties were more fluid, voters’ preferences are not ranked in the sort of transparent (and stable) uniform ordinate scale that the Median Voter Theorem requires to be useful. Not only is it difficult to infer a voter’s preferences from their position on an issue, but also two voters who have the same preferences might not agree on how important that issue is.
In the case of Trump, some voted for him because they liked his tax policy, and some found appeal in his racist rhetoric, but not necessarily both, and many voted for him simply because he was on the Republican ticket. In the real world, no campaign manager believes that they can win if they can just appeal to a single kind of voter.
Applying the Median Voter Theorem as a strategic framework also fails because even if such a continuum could be constructed, it would be unstable for two more reasons:
Voters also decide whether to vote, so the median voter changes as the number of those eligible to vote decide whether to be on the line at all.
If exiting the electorate is an option to voters, candidate actions to appeal to the median voter could also suppress turnout for those who were already on the candidate’s side of the lineup, offsetting any potential gain from shifting to the center.
As I’ve written before, the voting electorate is never the same river twice. There is a significant difference between those who vote in almost every election, and those who vote at all. Up through 2016, the percentage of those eligible in each election was pretty stable, which made it easy to believe that the country was divided into voters and non-voters. But, broadly speaking, only slightly fewer than 40 percent of those who are eligible to vote in every general election do so, but an additional 35 percent vote in some general elections. Those voting in every election are likely committed to the process, while others come in and out based on circumstances.
And it turns out that the decision whether to vote or not often has a pretty profound impact on the outcomes. The 70 percent of 2020 voters who had also voted in 2016 favored Biden by the same margin Clinton won the popular vote in 2016. But the election was won by the 12 point margin that Biden had among the 46 million who had not voted in 2016. Had we been competing in 2020 for the same fixed number of votes that were cast in 2016, this would be an entirely different argument. The 2020 election was not decided by one side picking off a few margin voters from the other--the losing candidate received 11 million more votes than he did in 2016, a record amount except for that Joe Biden won 15 million more votes than Clintion did, defeating Trump by 7 million more still. Popularists before 2020 argued that Biden couldn’t win by turning out more voters. And now, they are ignoring the importance of turnout in the midterms despite higher turnout having just prevented a second Trump term.4
What This Debate Says About Our Politics
I hope it’s obvious by now, if it wasn’t already, that popularism fails in its diagnosis and prescriptions. So why are so many strategists and analysts spending so much time and energy talking about it?
The answer is that popularism is the public discourse of an otherwise unarticulated contest for power within the Democratic Party. It is the language of intramural rivalries, not practicable prescriptions for winning general elections.
In this section, I will put that struggle in its historical context, show how popularists’ faulty reasoning serves that agenda, and observe how popularists sabotage the party they say they want to help win.
The Adults in the Room
The debate over popularism is essentially the same argument over how to handle Republican race-baiting that Democrats have been having among themselves since the mid-’60s. In every presidential election since then, Republicans have caricatured Democrats as effete, out of touch, and unreasonably beholden to people of color. And after each electoral defeat, a faction of the Democratic Party gets fixated on avoiding this Republican caricature, and throws out all that we know works to win elections. Everything old is new again—but this time it’s surfacing after Democrats won with record turnouts in 2018 and 2020.
Popularists take a tough-love, “eat your veggies” tone to argue that theirs is the rational, adult approach, and that those who object to its premises merely “don’t want to hear“ these harsh, yet soberly data-derived, truths. But it turns out that many Democratic elites are all too eager to hear these truths. There’s a reason that it’s the same story they’ve been telling themselves for decades. It’s a story that flatters them as the adults in the room, and condescendingly suggests that activists of color are unreasonable children who need to learn patience.
Again, popularism is not really about what Democrats should say; it’s about who Democrats should listen to. The zombie idea of the “median voter” may not make theoretical sense, but it’s politically useful to pretend that it does. Focusing on the median voter allows Democrats to argue that one kind of voter matters more than another without the awkwardness of saying so out loud.
Blaming out-of-touch white elites for the Democrats’ woes is yet another way to center white people in the party. Arguing about who matters more, white elites or the white working class, conveniently turns attention away from the work Black activists have done to push for a more progressive agenda and to turn out voters. And it obscures the tension behind Democrats’ ongoing efforts to have it both ways—maintaining the loyalty of the Black electorate, but reserving the right to keep them at arm’s length to avoid alienating whites. Popularism allows white party elites to attribute Democratic wins, achieved by virtue of enormous margins of support from Black voters, to “practical” white compromise.
There are dangerous consequences to reinforcing the “Democrats are out of touch” trope. It plays into the broader anti-intellectualist narrative that the MAGA Faction uses to sow ethnic conflict. Richard Hofstadter, when coining the term “anti-intellectualism,” pointed out that anti-intellectualist leaders (often well-educated elite men themselves) consider scientific reasoning to be an attempt by those without morals to engineer a system where they are virtuous. They liken intellectualism to movements known to draw broad feelings of resentment, such as atheism, communism, elitism, and the broader project of racial justice. This taps into the culture of grievance that defines those who ascribe to an ethnic hierarchy and draws them into the right-wing party.
Forecasting Is Not a Science
Again, poularists like to use the language of hard science (like “the physics of elections”) to make themselves sound more impressive, but forecasting is not a hard science.
The most important problem with estimates from statistical forecasting models is that even when they are caveated, they are expressed as probabilities. But forecasts are not probabilities. Probabilities are objective: On average, I will flip heads half the time. No one disputes that, or says it will land on heads two-thirds of the time, because landing on heads is a physical property of the coin. If forecasting worked the same way, as with coin flipping, it wouldn’t matter if we were Democrats or Republicans; we would get the same result.
All forecasting estimates are deeply subjective. A statistical model’s estimate seems “objective” because the public accepts conclusions reached with a lot of data and equations as objective. But in fact, subjective judgments must be made about everything from which data to use to the algorithmic methods. (And in a deeper sense, the data itself is the result of the original decisions made about what kind of data should be collected.) The resulting forecast is nothing more than a statement of the model’s own assessment of how often its forecast would have been correct if the same data were applied to all the elections that have already happened. It should be enough for people to see that subjective judgment matters when FiveThirtyEight, the New York Times and the Economist offer different forecasts.
Importantly, probabilistic models cannot be disproven because you can’t do multiple runs of history the way you can roll dice or flip coins. If the forecast says that there is a 66 percent probability that Clinton will win, and Clinton wins, forecasters claim a win, but if she loses, forecasters scold us for not understanding that Trump was going to win a third of the time. Heads I’m smart; tails you’re stupid. This kind of thing happens a lot. If you can see the difference between that and a pundit telling us that it was “pretty likely” that Clinton would win, let me know.
The greatest challenges to forecasting are the “unknown unknowns.” Most models do not incorporate exogenous variables that have been stable, so those models cannot account for what happens when they change. History is full of twists and turns, but political forecasting is biased toward what has happened continuing to happen. For example, in 1988, George H.W. Bush’s victory was the fifth GOP presidential win out of the previous six, with the exception of Carter, a devout Southern Baptist, barely squeaking by even with Watergate wind at his back. In 1989, the GOP won the Cold War when the Berlin Wall fell. In 1990, Operation Desert Storm won the Persian Gulf War decisively with almost no American casualties. A year before the election, H.W.’s net approval was +20. And Democrats had controlled the House for the previous 46 years. Yet since then, Republicans have won the popular vote only once in the eight elections that followed, but controlled the House for 10 of the next 14 Congresses.
Polling Has Limits
Popularists argue that we should rely on polls to tell us what is popular. But one anecdote in Klein’s article, where Shor used data to argue Democrats need to talk about climate change less, is unexpectedly revealing:
Shor showed me, as an example, a set of environmental talking points he’d tested, in which the ones that mentioned climate change performed worst. “Very liberal white people care way more about climate change than anyone else,” he said. “So when you talk about climate change, you sound like a weird, very liberal white person. This is why policy issues matter more than people realize. It’s not that voters have these very specific policy preferences. It’s that the policies you choose to talk about paints a picture of what kind of person you are.
If popularism’s prescribed approach was just neutrally following polling data, one might notice that a two-thirds majority of Americans, and an overwhelming 90% majority of Democrats, agree that the federal government is doing too little to reduce the effects of global climate change. Yet Shor chooses to focus on a modest difference of 2.1 percentage points in a controlled message-testing environment to make a nebulous argument about cultural signaling—one that happens to fit a prior assumption that winning white non-college voters is essential to winning elections.
There is no empirical evidence that the point estimates in persuasion message testing translate directly into real life, and there is no theoretical basis for imagining they could. There is also no reason to believe that the survey takers are representative of the whole population in terms of their responsiveness to messaging, a variable that no one has ever even tried to develop weights for. Further, surveys like this exaggerate response because taking the survey requires you to engage with the messages, which is obviously not the case for everyone who, say, has the message in their Facebook feed or in a letter they don’t open.
Message testing is an indispensable tool for campaigners when it comes to deciding on how to phrase or frame a TV ad. But such testing can only tell you how people respond to that particular ad; it can’t tell you whether they will actually vote for your candidate, because that decision is informed by much more than just your ad.
How Much Do Popularists Actually Care About Winning?
Popularists say that their project is to help Democrats win campaigns, but their track record suggests that’s not their highest priority.
If popularists wanted to help Democrats win, you’d think they’d be more helpful. David Shor uses his black box “power simulator” model, which regularly simulates every election for the next decade, as the “scientific” basis for what he says. Yet if this simulator was accurate in September of 2020, he apparently decided that it wasn’t worth telling anyone two months before the election that a dozen House races were closer than they thought, or that Democrats might have used elsewhere the extra $100 million raised in South Carolina only to see Lindsey Graham win reelection by 10 points.
When pressed to share, Shor says that his underlying data is proprietary. But since all his clients are Democrats, who presumably care more about winning elections than keeping their data projects secret, hiding behind “proprietary” suggests that popularists like Shor have never really pressed the matter with their clients. Those on this list know that most of us—unions, progressive organizations, and Super PACs—routinely share with each other.
Popularists advise Democrats to avoid “cultural” issues that alienate median voters, but never say specifically what Democrats should do when Republicans attack them on “cultural issues.” Indeed, they have precious little specific advice for how Democrats can put their theories into practice.
Popularists also engage in rhetorical bad faith and show surprising contempt for their copartisans. Popularists often claim that people “don’t want to hear” what they are going to say. This sets a finger trap for critics: If you object to their claims, it’s because you just can’t handle the truth. This dynamic is even more troubling since popularist critiques are about how the party deals with race and gender. This means that when people of color or women object, no matter how strong their argument, they can be dismissed with an eye roll: “There they go again; our point exactly.”
If the popularist project was actually to persuade misguided Democrats, then they would answer any of the many substantive critiques offered since the Klein piece and leap at invitations to debate or even initiate such invitations themselves.
The Real Path Forward
Again, Democrats smashed turnout records to win in 2018 and 2020. These are not normal times. Biden has a reservoir of support that no other president has had going into a midterm. Now is not the time for a defensive crouch.
While 2022 could well be another bad midterm—even if Biden’s approval recovers and the economy is doing well—Democrats can still survive it, as I’ve explained. Democratic House candidates received 19.6 million more votes in 2018 than Republican House candidates had received in 2014, far and away more than any president has had to work with going into the midterms.
Because MAGA has only one play in its playbook—stoking racial, gender, and other cultural grievances—Democrats cannot pretend that they won’t have to address race as long as they don’t mention it. The role of racial dog whistling in politics since the passage of the Civil and Voting Rights Act has been well documented in many places, including in Ian Haney-López’s 2015 book Dog Whistle Politics.
An important element in Haney-López’s account is the lengths that Bill Clinton had to go with his own dog whistling to calm white voters, which boxed him into promoting policies that severely harmed people of color. Bouie also discussed this history in his column about why it would be difficult and problematic for Democrats to make the concessions that popularism demands.
Heather McGhee offers important insight on the psychology behind zero-sum thinking on policy issues, which is at the core of the belief that we must shy away from discussing racial justice to win votes. As she illustrates in The Sum of Us, catering to this belief system is only holding us back from the true gains we can make when we view racial equality as a win for society overall.
As Anat Shenker-Osorio has noted:
“Telling Democrats not to talk about race renders them even more vulnerable to right- wing claims about their purported positions on ‘cultural issues’. The instinctive response to keep quiet on race doesn’t make the race conversation go away. It guarantees that the only thing voters hear on the topic is the unrelenting race-baiting of our opposition. And this makes it so our economic promises cannot even get proper hearing—no matter how popular they prove on a poll. When someone has convinced you, for example, that immigrants are both taking your job and, at the same time, living fat off benefits while not working, then our promise of higher wages feels irrelevant.”
More critically, Anat and colleagues at We Make the Future—alongside many others—have shown how to engage our base and persuade the middle amid the siren song of right-wing race baiting, by working with organizers to implement effective messaging.
Luckily, “promote racial justice” versus “increase the Democratic vote share” is a false choice. Unfortunately, it’s a false choice that many elites either find genuinely persuasive, or cynically exploit in order to advance their own agendas.
While we are busy granting unwarranted prime-time hot takes to popularism, we risk losing our moorings in terms of the task before us and what our most recent wins have to teach us about achieving it. We must turn out the voters who surged us to victory and 2018 and 2020, and we must do so in the present-day realities of right-wing race baiting and utterly understandable cynicism from our base.
The irony is that popularists frequently reference his very important graphs to make their case about education polarization, but ignore the implications of almost everything else Piketty has to say in defining it or its meaning.
So much more to be said here, but see John Jost’s A Theory of System Justification for the preeminent psychological articulation of this point.
If you want to read about popularism in the words of its founder, Klein’s piece lays that out well. Here is a key excerpt:
“Shor’s theory of popularism, at its heart, is a critique of the professional staffers, consultants and organizers who shape the Democratic Party’s message, image and strategic choices.
‘I think the core problem with the Democratic Party is that the people who run and staff the Democratic Party are much more educated and ideologically liberal and they live in cities, and ultimately our candidate pool reflects that,’ he said.
Nor is Shor’s ire aimed only at the liberal wing of the party. Popularism isn’t mere moderation. One of the highest-polling policies in Shor’s research is letting Medicare negotiate prescription drug prices, but it’s so-called moderates, like Sinema, who are trying to strike that from the reconciliation bill. To Shor, this is lunacy.
Shor believes the party has become too unrepresentative at its elite levels to continue being representative at the mass level. ‘I don’t think it’s a coincidence that the people we’ve lost are likely to be low-socioeconomic-status people,’ he said. ‘If you look inside the Democratic Party, there are three times more moderate or conservative nonwhite people than very liberal white people, but very liberal white people are infinitely more represented. That’s morally bad, but it also means eventually they’ll leave.’ The only way out of this, he said, is to ‘care more and cater to the preference of our low-socioeconomic-status supporters.’”
Another common popularist against entertaining turnout effects includes the substantial evidence that non-voters are not particularly left leaning, which is something of a non-sequitur. That there is not strong evidence that non-voters are just waiting for a bold left message, has nothing to do with whether turnout changes affect partisan outcomes.