Heterogeneous Networks: A Shrinking Safeguard against Outparty Hostility?

• In Event: Networks: US Politics and Theoretical Advances

Abstract

Do heterogeneous political networks inoculate against outparty hostility? And does partisan animosity affect social relations? These two dynamics are intertwined: exposure to heterogeneous social networks may prevent outgroup hate, however, especially in a highly polarized political climate, individuals might proactively avoid interactions with outparty members, thus reducing their exposure to alternative viewpoints. In its extreme form, a self-sustaining dynamic could arise in which polarized citizens avoid one another, their networks become more homogeneous, their animosity towards outpartisans increases, etc.

Hypothesis 1: Heterogeneous social networks inoculate against outparty hate
Heterogeneous political discussion networks are related to more moderate political attitudes (Zuckerman, 2005) and network composition influences political attitudes and behavior (Sinclair 2012, Rolfe 2012, Klofstad et 2013). More recent studies have also found that heterogeneous networks are associated with lower levels of affective polarization (Amsalem et al., 2022; Rossiter, 2021). We therefore expect that social network heterogeneity could reduce outgroup animosity.

Hypothesis 2: Partisanship affects social network selection
There is plenty of evidence that many social contexts in the US are becoming more politically homogeneous (Rodden 2019, Brown et al., 2022; Brown & Enos, 2021; Huber & Malhotra, 2017), and some evidence of slightly more homogeneous political discussion networks (Lee and Bearman 2021, Lee 2021). There is less certainty on whether partisanship per se is driving this development, or whether it is a by-product of social selection based on other characteristics that correlate with partisanship (Baldassarri and Page 2021). In fact, political homophily has always existed. However, historically, political homophily was the by-product of homophily along sociodemographic dimensions, like class, ethnicity, religiosity, age, gender etc. – which are correlated with partisanship, but do not coincide with it. In contrast, some recent scholarship seems to suggest that political partisanship directly informs patterns of relationship. For instance, when asked to rate dating profiles (Huber & Malhotra, 2017) or roommate profiles (Shafranek, 2021) in conjoint experiments, respondents seem to prefer copartisans over non-copartisans. It is, however, unclear whether partisanship is used as a proxy of other characteristics, or people really care about it. We also do not know if partisan selectivity is related to network homogeneity.

Design and Results from wave 1
We run a two-wave survey coupled with several experiments on a representative sample of 3000 US citizens using the market research company Bovitz Inc. The study has been preregistered and received IRB approval. The first wave of data collection was conducted in the week before the 2022 Congressional Election, when inter-party animosity was high (Hernández et al., 2021; Sood & Iyengar, 2016). The second wave will be fielded in the next month (February 2023), in a period when politics is more likely to recede to the background. Ideally, this design allows for a within person test of the effects heterogeneous social networks on partisan animosity, by exploiting the exogeneous variation in partisan animosity induced by the election.

In addition to classical sociodemographic information, measures of partisanship and affective polarization, the survey measures heterogeneity in political networks using both a name generator, to capture close relationships, and an estimate of the number of Republican and Democrat acquaintances, to capture more casual relationships. We assess the relevance of partisanship in the selection of future interactant partners using an open-ended question, and three experiments: a pairwise comparison, a conjoint experiment and an incentivized competence game.

Results from the first wave clearly demonstrate that having heterogeneous social networks is associated to lower levels of animosity toward the outparty. In the second wave we aim at causally testing this relationship by manipulating the saliency of specific (homophilious vs heterophilious ties). Moreover, results across different experimental scenarios consistently show that there is minimal social network selection based on partisanship. When asked which characteristics they are looking for in an officemate, respondents almost never mention partisanship. They also consider partisanship among the least relevant aspects in pairwise comparisons with 2 dozen other traits. Conversely, when presented with a choice in the conjoint and the competence game, respondents strongly select on partisanship. Respondents reporting homogeneous political networks and high levels of outparty animosity seem to be those more influenced by partisanship in their selection of social relationships.

Authors

• Delia Baldassarri, New York University

• Jona de Jong, European University Institute