POLITICS

Was there really a ‘populist quadrant’ in American politics?

Different methods paint a different picture of its conservatism

Photo by Josh Blanton on Unsplash

You may have seen this plot a few times over the past eight years. It pointed to the existence of a ‘populist quadrant’ (top left) that was liberal on economics and conservative on social issues. One might have called it the Sanders/Trump quadrant.

Source: Voter Study Group (June 2017).

Today, some speculate that RFK, Jr. now calls it home.

I write to share three points:

1. There is a ‘populist bottom’ and not a ‘quadrant.’
2. The social conservatism of its occupants may be overstated.
3. We might use this perspective to think about third-party runs ‘from the top’ as well as those like RFK’s, which might be understood as ‘from the bottom.’

When the graph came out, I wanted to see if a neutral algorithm would produce the same basic picture. Like my colleague Lee Drutman, I was interested in what was happening to the party system and what it might mean for electoral reform. That’s because many electoral reforms have been about containing ‘populism.’

I also had just read a book by Cas Mudde and Cristóbal Rovira Kaltwasser on ‘populism’ and integrated it into my model of politics as follows. Populist attitudes are anti-establishment and can pick up elements of socialist or nationalist ideology. For me, this theory implied a Cartesian ‘space’ with two dimensions: one capturing ‘traditional’ left/right issues, then a second capturing attitudes toward the political system as such.

Did the ‘populist bottom’ exist? In 2022, Josh Dyck and I published some results in Public Opinion Quarterly (open access). Here is what we did:

1. We used survey items that also had been appearing in the emerging ‘populist attitudes’ literature. One group came from the ‘stealth democracy’ tradition and included items on “compromise” and “successful business people,” etc. The other group came from the ‘political discontent’ tradition and included items on ‘external efficacy’ — “people like me have no say,” “elites don’t understand,” “elections don’t matter.”
2. We used other survey items on ‘social’ and a range of liberal-conservative issues. These included the famous items on ‘racial resentment.’ One of our analyses also used some items from a newer racism index.
3. We used a non-metric scaling procedure to see how items and respondents were configured in the ‘space’ of politics. (Hare, Highton, and Jones have just published something interesting with similar methods.)

Here is what we found:

1. American politics in 2016 reflected more of a ‘populist half’ than a ‘quadrant.’ This was very clear for certain items (see below). Others looked more like they had folded into the liberal-conservative divide. That was the case for items that, 20–30 years earlier, had been correlated with support for Ross Perot (see paper).
2. As an added test of populism theory’s spatial implications, we asked if respondents ‘toward the bottom’ tended to have voted for Bernie Sanders or Donald Trump in their respective primaries. They had.
3. The north-south axis used to be about social issues. In 2016, it was more about ‘the system.’

Figure 2 from “The Structure of American Political Discontent” (Santucci & Dyck 2022). D = Clinton vote, R = Trump vote, O = other vote. Dashed lines estimated from respondent coordinates via probit. Solid lines and coordinates generated by ordered optimal classification.

Looking back at the paper two years later, I also notice something interesting about ‘conservative’ responses to the classic racial-resentment items. To the extent that the vertical dimension picks these up, people who agreed were closer to the top of the space than to its bottom. This is the opposite of what one might conclude from the viral graph.

The original graph was right in that it showed an ideologically heterogeneous grouping in its ‘populist quadrant.’ That appears in ours as the range of opinion across the bottom. Either therefore can be used to understand RFK, Jr’s blend of appeals — some of which sound like Bernie, others more like Trump.

But the graph that emerges with the algorithm also might help us understand periodic threats of candidacy ‘from the top.’ I am thinking here about No Labels and the sometimes-overlapping push for nonpartisan elections (i.e., reform).

One more point is worth note. Voters struggle with multi-candidate elections to single-seat offices. If both dimensions end up active in November — as arguably happened in 1924 as well — the whole idea of choosing becomes interesting. Do voters first need to figure out what politics itself is about? If so, do they get confused and just not vote? Could the going flavors of electoral reform bring order to the situation? (Many formal modelers have said no.)

I look forward to extending the ‘populist bottom’ project and will report accordingly.