Neoliberal

Content:

Map Link: https://imgur.com/a/y3zj8HU

Understanding how voters choose a candidate is a fundamental topic in political science. In an ideal democracy, voters would study policy manifestos and pick the slate that most closely matches their views. Reality, however, is messier.

One long-standing explanation of voters’ behaviour is that they support candidates who represent or advocate for the interests of their social group—often defined on the basis of age, class, ethnicity or gender. A more recent school of thought, called “moral foundations theory”, posits that a few broad philosophical values, such as the importance of fairness or authority, shape peoples’ political preferences.

Such principles are harder to measure than demographic data. In the past, proponents of this interpretation have had to rely on surveys, which may not accurately reflect respondents’ true beliefs or voting records. But a recent paper, by Benjamin Enke and Steven Sun of Harvard, Raymond Fisman of Boston University and Luis Mota Freitas of Oxford, managed to assess variation in one such value—“particularism” versus “universalism”—from real-world data. They found that it predicts American voters’ choices more accurately than many oft-cited demographic variables do.

Donations between congressional districts, $
Via DonorsChoose, 2000-16, log scale
Link to Graph: https://imgur.com/J6b8jTi

Distribution of speeches given to 114th Congress
Link to Graph: https://imgur.com/qDTJUM6

The particularism-universalism axis tracks how much people favour those close to them, such as family or neighbours, over those with whom their ties are weaker. This distinction maps neatly onto the philosophical differences between America’s major political parties. Using speeches made by members of Congress, the authors found that Republican lawmakers typically used far more particularist language than Democratic ones did.

To test if the same is true for voters, the authors devised a clever method to measure particularism in each of America’s congressional districts using charitable donations. DonorsChoose, a non-profit, allows people to donate online to classroom-funding requests from public-school teachers. It provided data on 4m donations, including the locations of donors and recipients. The authors argued that the more people in a district prioritised giving to nearby schools, the more particularist they were. They measured proximity both geographically and socially, by analysing the number of Facebook friendships linking pairs of districts. The latter measure reflects the strong ties people who move to cities have with their places of origin. Link to Graph: https://imgur.com/HeavceK

People in every district gave more to nearby causes than far-off ones. But the most particularist districts were also the most Republican. The ten most particularist districts had a Democratic vote share 45 percentage points lower than the ten least. Differences in levels of particularism could predict some 25% of the variation in a district’s vote share, more than the amount explained by some core demographic characteristics, like income (0.2%) and education (2%). Moreover, its effect remained statistically significant even after accounting for the impact of other traits, such as a district’s racial make-up.

These findings do not refute other explanations of voter behaviour: predictions using both demography and particularism were more accurate than those based on either alone. But they do suggest that explanations of elections that exclude philosophical values are probably incomplete.