Predictive modeling to study lifestyle politics with Facebook likes

Abstract “Lifestyle politics” suggests that political and ideological opinions are strongly connected to our consumption choices, music food taste, cultural preferences, other aspects of daily lives. With the growing polarization this idea has become all more relevant a wide range social scientists. Empirical research in domain, however, is confronted with an impractical challenge; type detailed information on people’s lifestyle very difficult operationalize, extremely time consuming costly query survey. A potential valuable alternative data source capture these values choices media data. In study, we explore value Facebook “like” complement traditional survey study politics. We collect unique dataset likes than 6500 participants Belgium, fragmented multi-party system. Based both types data, infer preference respondents. The results indicate non-political indicative useful describe voters terms common interests, features. This shows can be