Multi-level Modelling of Voting Behaviour

Political Systems and Electoral Behavior: a Review of Internationally Comparative Multilevel Research

“…Internationally comparative electoral research requires a wider conception of context. To address not only the "social logic" but also what might be called the "institutional logic" of politics, the notion of context must be conceived in a more abstract way, as encompassing social and spatial arenas of experience into which individuals are embedded and that, by structuring opportunities and constraints, create expectations and incentives for these individuals' political behavior (Anderson 2007b(Anderson , 2009Dalton and Anderson 2011b;Friedrichs and Nonnenmacher 2014;Lubbers and Sipma 2017;Schoen et al 2017). As highlighted by Anderson (2009, p. 323), this understanding nicely accommodates the notion of political institutions as formally (or informally, for that matter) imposed constraints on human behavior.…”

“…The concomitant development of sophisticated procedures for analyzing data that are hierarchically nested into two or more levels of observation created the tools for efficiently and adequately dealing with the statistical challenges posed by such data. Although two-step approaches (Jusko and Shively 2005) and pooled interactive regression analysis (Franzese 2005) found a fair number of applications, hierarchical modeling (Steenbergen and Jones 2002;Lubbers and Sipma 2017;Schmidt-Catran 2019) has, over time, become the most widely used technique for cross-national multilevel investigations of electoral behavior (see also Spies and Franzmann 2019). This method is uniquely suited for dealing with two kinds of research questions: whether and how properties of higher-level units, such as countries, influence the means and distributions of individual-level dependent variables, over and beyond the effects of persons' own attributes (random intercept models in the language of hierarchical modeling), and whether and how such higher-order phenomena af-K…”