For a long time, electoral geography has focused on variations over space in voting behaviors as a means to make sense of political opinions. This approach relies mostly on two assumptions. First, political behavior is linked to social, cultural or functional attributes of voters. Second, those attributes tend to concentrate within given spaces such as neighborhoods, areas and regions of the country. Together, those two assumptions form the basis of the Compositional theory of collective political behavior: collective vote outcomes result from dominant attributes across populations. In other words, the Compositional theory reckons that change in a population’s political attitude results from changes in the attributes of the population. Within smaller timeframes, this change associates mostly with demographic transitions due to migrations. In order to test whether changes in the characteristics of local populations affect their political attitude, this study looks at 650,000 municipal outcomes over federal popular votes called in Switzerland between 1981 and 2015. In a first time, we use Geometric Data Analysis (GDA) in order to reduce the votes we consider (n=300) to political components that structure municipal support over four sectional periods. From this data, we look at how local support for those political components evolved in relation to populations’ (n=2,291) demographic changes. We are especially attentive to transitions associated with changes in popular attitudes such as population growth and decrease, aging, socioeconomic status and proportion of foreign-born voters. To perform such analysis, we rely on Dynamic Network Modeling which enables us to identify what characteristics in the population is likely to affect its support for each of the political components.

The study is conducted by Dr. Shin Alexandre Koseki and Prof. Dr. Christoph Stadtfeld at the Chair of Social Networks of the ETH Zurich.


Shin Alexandre Koseki