Assessing Cultural Bundles in Sweden: A Study of Lifestyle Polarization
Author(s): Elida Izani Ibrahim, Miriam Hurtado Bodell, Anastasiia Menshikova, Felix Lennert
Thursday 15 | 11:20-11:40
Room: TP55
Session: Sociality in the Digital Age: Interactions, Emotions, Lifestyles, and Norms
Political polarization is seemingly structuring not only people’s voting patterns and political action, but also their everyday social life. When everyday preferences, from the choice to follow certain kinds of social media influencers to government agencies, become politically coded, social cohesion is at risk creating less space for common ground. Partisan leaning predicts who you connect to, what kind of news you consume, and what political attitudes you hold with increasing accuracy. The literature has also begun to determine different patterns of cultural consumption (“lifestyle bundles”) based on political leaning. However, further scrutiny into whether or not it is politics or sociodemographic factors are driving these lifestyle enclaves is needed to better understand political cleavages in society. This study builds on this growing literature by investigating political and gender identity in structuring lifestyle bundles in the Swedish political context, as well as discerning which identity is more relevant for certain lifestyles. Our study contributes to understanding lifestyle bundles and the constitutive roles of politics and gender. Our primary goal is to determine the extent to which clusters of lifestyle preferences appear online among politically active users. Secondarily, we analyze what predominantly clusters these disparate topics together: politics or gender.
Using data from Swedish Twitter, this research employs seeded Latent Dirichlet Allocation (LDA) models to measure political preferences and cultural consumption patterns for a sample of politically active Swedish Twitter users (N=12,230). We assess the strength of users’ connections to the eight parties in the Swedish national parliament and the thirty-two discovered lifestyle clusters, such as soccer, environment, trade unions, and cultural institutions. We combine data from the Chapel Hill Expert Survey with our Twitter data to gain a more granular measurement of political preferences. We also use decision trees to evaluate the contribution of gender to the observed polarization patterns.
The findings reveal distinctive corresponding patterns between political views and nonpolitical spheres. Partisan belonging is a major driver of lifestyle polarization when it comes to topics related to media, cultural institutions, religion, politics, and humor. In contrast, the connection between political affiliations and clusters such as sports, military, tabloid press, health-related accounts, and accounts related to IT and technology disappears after adjusting for gender.
These insights into lifestyle-based political clustering in Sweden enhances our understanding of polarization dynamics as it zooms into gender as potential confounder. With this approach to measuring lifestyle polarization, we complement previous survey-based work and discover lifestyle polarization structures using behavioral digital traces which account for relationships between individuals and reveal their cultural preferences.