Mixed Methods Approach: Reconstructing Local Identities in Context of Local Referenda

In this study, the authors take a theoretical framework of identity of groups and extend an explorative approach of Qualitative Discourse Analysis by using different methods of Network Analysis, Structured Topic Modeling and Sentiment Analysis to find a model to enlarge the amount of texts that can be taken as a corpus for discourse analysis. The authors define a model of the relation between group- identities and observe these within an instituional framework of local referenda in Germany. Hence, the authors took all local referendum cases in three German cities, Dresden, Bonn and Potsdam, from the database local referenda in Germany (http://datenbank-buergerbegehren.info/) and built a corpus of newspapers' (online) reporting on each referendum. With the help of an explorative data science approach (networks, structured topic modeling, sentiment analysis) the authors detect different topics and subsume these topics with regard to the theoretical framework to inclusive and exclusive patterns of local identity. The authors speak up to expand inductive methods of qualitative research by the chosen methods of text analysis.