Answers are grounded in this paper and cite their sources.
Methodological responses to the issues associated with multilevel data quickened in recent years. Multilevel models, as a generalization of regression techniques, perform at least as good or better than regression models (Gelman, 2006; Gelman & Hill, 2007); at the same time, beyond explicitly modeling the multilevel nature of a given data structure, multilevel models exhibit differences that practitioners should observe. Recent accounts recommend the “estimated dependent variable” approach where the number of first level variable is high (often individuals) and the number of second-level variables is low (often countries, interviewers, election periods, etc.) (Franzese, 2005; Huber, Kernell, & Leoni, 2005; Jusko & Shively, 2005; Lewis & Linzer, 2005). Given the prevalence of such a data structure in international public opinion research, this article sets out to make the estimated dependent variable more accessible by examining the effect of contextual effects on union members’ preferences over active labor market policies. The note proceeds as follows: the first section outlines the estimated dependent variable approach; the next section clarifies the theoretical puzzle of trade union member support for active labor market policies as well as the data; the third section includes the analysis; the final section concludes.
No takes yet. Share an insight, caveat, or question.
Mark Nelson (2009) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: