Left Populism in the Democratic Primary
By Jon Green (@_Jon_Green)
Ever since Bernie Sanders and Donald Trump both launched campaigns for their respective party nominations in 2016, their appeals have invited comparisons invoking the populist style in American politics. In more recent years, such comparisons have extended to other insurgent progressive politicians such as Alexandria Ocasio-Cortez. However, many of these comparisons seem unclear about what exactly populism is -- many pundits seem to simply know it when they see it -- making it hard to tell how valuable such blanket comparisons actually are. What constitutes populism in general, and what separates distinct populist appeals from each other?
One thing that populism definitely is an appeal to a specific notion of the popular will. As the late political theorist Ernesto Laclau argued in On Populist Reason, populism works in part through what he termed an “empty signifier” -- a demand that is more about the creation of an overarching identity (say, “the people”) than its specific policy content. The populist identifies “the people” as virtuous and deserving, in a struggle against a politically constructed out-group that subverts their true will -- and whose rightful claim to political power the populist seeks to realize. The push to create an overarching and singular identity can further lead populist appeals to be some combination of idealistic, nostalgic, and nationalist -- with a tendency to simplify complex political issues to fit into the frame of “the people” struggling against their antagonists.
Left there, one can see how this framework could be fairly applied -- with a caveat here or there -- to Trump, Sanders, Ocasio-Cortez, and other politicians who seek to pit a virtuous and politically-constructed people against a nefarious and politically-constructed out-group (Josh Hawley comes to mind). However, some 50,000-foot similarities can lead political observers to gloss over stark differences between different populists based on who they mean by “the people” and how they construct their out-groups. Donald Trump and Bernie Sanders may both be populists, but they offer very different visions of who they mean by “the people” and which out-groups should be the target of their anger. One step past the simple fact that they offer a clear “we” and a clear “they”, and the comparisons start to break down. The two populists have very different politics. This suggests that there isn’t one “populism,” and different strands of populism require their own measurement strategies.
We took a preliminary attempt at one such strategy by embedding a series of questions designed to tap into distinctly “left” populism in our recent Data for Progress / YouGov Blue survey of the Democratic primary electorate. We started with six items we thought reflected a particular combination of left-wing economic appeals, frustration with the pace of political change, and the fundamental struggle between labor and capital that invoke the left populist style in American politics. After fielding the survey, we found that four of these items loaded reliably onto one dimension, and so we extract their first principal component to give all of our respondents a left populism score. The other two did not scale onto this dimension, and are excluded from this analysis.
Left Populism Battery:
Please indicate the extent to which you agree with the following statements:
Scaled items:
INEQUALITY: Economic inequality is the root cause of most of society’s problems.
SOCIETY: It is wrong for a society to have billionaires while there are people living in poverty.
ALLEGIANCE: US companies building factories overseas shows that they care about profits and not American workers.
CLASS: The US economic system has made it impossible for people in the working class to have a comfortable life.
Dropped based on alpha:
PROCESS: Our system of government has too many rules that stand in the way of necessary change.
DETAILS: The best politicians are pragmatic and detail-oriented.
As one might expect, most likely Democratic primary voters agree with most of these left-populist sentiments -- as shown by the fact that the distribution of scores on the left populism scale is heavily weighted toward the high end. In the chart, 0 represents the mean observed value of left populism, the range covers all possible values, and each number on the x-axis represents standard deviations from that mean. That the mean falls on the higher end of possible values, and that the lowest possible value is farther from the mean than the highest possible value, indicates that the average respondent in the survey has a relatively high score on the measure. This tracks with other survey items designed to measure general populism in the full electorate, such as whether the respondent prefers politicians they feel like they could get to know as a person. Populism constitutes appeals to “the people.” Those appeals are generally popular.
Additionally, the degree to which Democratic primary voters endorse left populist sentiments is associated with other political values we might expect to vary along this dimension. Below, we plot average levels of left populism across the three response options to another item on the survey -- which asked respondents which of the following statements was closest to their view:
The government should guarantee that no one lives in poverty.
The government should guarantee that no one who works full time lives in poverty.
The government should not guarantee that anyone, even those who work full time, lives above the poverty line.
As the below plot shows, the broader the guarantee against poverty a given respondent endorses, the more of a left populist they are likely to be. Respondents who oppose any government intervention to keep even full time workers above the poverty line are, on average, nearly a full standard deviation below the mean of left populism. Respondents who support anti-poverty guarantees for full time workers are closer to (but still below) the mean level level of left populism, on average, while those who think the government should guarantee that no one lives in poverty regardless of their employment status tend to be above average on the left populism scale.
We also find generally expected relationships between left populism and candidate support. Below, we show average levels of left populism of each candidate’s first-choice supporters. Respondents who indicated that, if their primary or caucus were held that day, they would vote for Bernie Sanders or Elizabeth Warren tend to have higher-than-average scores on the left populism scale, while respondents whose first choice is Joe Biden, Amy Klobuchar, Tulsi Gabbard, or John Hickenlooper tend to have lower-than-average left populism scores. The other candidates have first-choice supporters who, on average, are not significantly above or below the overall average in left populism. Of these, the only possibly strange result is for Tulsi Gabbard, who may not personally oppose left populist sentiments to the same extent that, say, Joe Biden does. However, as we show below, this may be a function of the fact that relatively few people have her as their first choice. Other likely voters who are higher in left populism are considering her, but would vote for someone else if the election were being held now.
To show this, we specify logistic regressions to model the relationship between left populism and the likelihood of considering voting for each candidate -- controlling for other factors such as age, gender, education, partisan identification, and racial and gender attitudes (while subsetting by race such that each model is specified separately for white and non-white respondents). As the below plot shows, left populism is positively associated with the likelihood of considering Sanders and Warren and negatively associated with the likelihood of considering Biden (among both white and non-white respondents) as well as Hickenlooper, Klobuchar, Beto O’Rourke, Cory Booker, and Pete Buttigieg (among white respondents). The other coefficients not being statistically distinguishable from zero indicates that, for those candidates/racial sub-groups, the likelihood of a respondent in that sub-group indicating that they are considering that candidate does not meaningfully vary with left-populism.
Finally, we examine the relationship between other respondent characteristics and the propensity to endorse left populist statements. To do this, we specify a linear regression with left populism as the outcome variable and the same remaining independent variables as in the previous regressions (again modeling racial sub-groups separately). Here, we find that among both white and non-white respondents, racism (as measured by a scale that combines items from the Racial Resentment and FIRE batteries) is negatively associated with left populism while sexism (as measured by the items on the Hostile Sexism battery) is positively associated with the measure -- after controlling for other factors. This would make left populism distinct from populism as it is traditionally conceived, as the latter is typically positively associated with racism. However, the positive association with sexism also highlights how left populism may be distinct from other strands of liberalism in terms of whose demands it prioritizes and, by extension, to whom it is attractive.
We also find that, among white respondents, holding a college degree and being a man are associated with lower levels of left populism after controlling for other factors. We also find that likely Democratic primary voters who identify as liberal tend to be higher in left populism than those who identify as conservative. And, interestingly, we find that whites who identify as weak Democrats tend to be higher in left populism than whites who identify as Strong Democrats -- possibly reflecting the fact that left populist politicians are typically seen as “outsiders” relative to the Democratic Party establishment in ways that may turn off strong partisans.
To be clear, this is our first attempt at constructing a battery of survey questions designed to tap into the particularly left-leaning dialect of populism espoused by politicians such as Bernie Sanders and, more recently, Alexandria Ocasio-Cortez. To further develop this measure, we’ll want to test different items, make direct comparisons to other measures of populism, and measure the construct on a nationally representative survey (as opposed to a survey of likely Democratic primary voters, as is the case here). However, this preliminary evidence suggests that left populism can be differentiated from both traditional populism and general liberal ideology, while highlighting meaningful distinctions in how the Democratic electorate thinks about politics and chooses its leaders.
Jon Green (@_Jon_Green) is a co-founder of Data for Progress and a Ph.D. candidate at Ohio State University.