Sam Whitt, Alixandra B. Yanus, Brian McDonald, John Graeber, Mark Setzler, Gordon Ballingrud, and Martin Kifer
Over the past two decades, researchers have turned to behavioral experiments to examine “other regarding preferences” across varying social identity treatments (Fershtman and Gneezy 2001; Jacquemet and l’Haridon. 2018; Bardsley et al 2020). Increasingly, scholars have also found utility in behavioral experiments to study partisanship (e.g., Carlin and Love 2013, 2018; Fowler and Kam 2007; Iyengar and Westwood 2015). Researchers have consistently found a partisan gap where co-partisans are treated more favorably than opposing partisans, which, in the United States, in particular, has been attributed to underlying affective polarization (Iyengar et al. 2019). Our main contribution to this emerging literature has been to update prior findings from the Obama era to the Trump era. During this period, we find that affective polarization in partisan behavior has only increased.
Our research was conducted using behavioral experiments embedded in a representative online survey of 1210 Americans conducted between May 24-28, 2019. Our experiments utilized standard tools of behavioral economics—dictator, trust, and public good games—to detect partisan identity-based preferences. The dictator game is a simple tool for measuring altruistic preferences toward others in terms of how much money people are willing to give to co-partisans compared to opposing partisans. The trust game measures trust preferences between co-partisans and opposing partisans when incentivizing trusting behavior in sending and receiving money in two-person exchanges. Finally, the public good game incentivizes group cooperation in larger 10-person groups, but we randomly assign members of those groups to different partisan majority compositions. In all of our experiments, allocations and financial rewards are hypothetical, but we nevertheless find a consistent bias in terms of subjects favoring co-partisans over opposing partisans to greater levels than observed in prior studies. In other words, despite incentives to signal cooperation with opposing partisans through non-costly financial gestures, most subjects responded to experimental treatments in ways that signal increased partisan polarization. This behavior was not conditional on demographic characteristics in the sample or regional effects. However, true independents, as opposed to independent leaners, did not respond to treatments as clearly as partisan respondents.
While we cannot identify the affective drivers of partisan polarization, we do find that people who feel less socially distant from and have more contact with opposing partisans are more trusting, cooperative, and altruistic toward others in our experiments. This is in line with the contact hypothesis literature; contact works to facilitate better information, reduce fears and promote empathy with out-groups (Pettigrew and Tropp 2008). Unfortunately, because partisan sorting appears to be increasing in the United States, the contact mechanism necessary to reduce micro-level polarization may be in short supply (Brown and Enos 2021).
In our ongoing research, we use panel survey data to explore how affective polarization has been impacted by the Covid-19 pandemic, the outcome of the 2020 election, and intersections between partisan biases and ethnic and racial biases in experimental decision-making. We look forward to reporting those results in future manuscripts.