Multilevel Models

Abstract

Scholars who study political processes often must consider how these processes are embedded within sociopolitical contexts determined by geographical units, organizations, or other higher-level units. Often, this is reflected in the development of multilevel data sets that permit these scholars to analyze social processes that operate across different levels of analysis. Increasingly, the focus of study in such research is on contextual or cross-level variables as opposed to variables at lower levels of analysis (Blalock 1984). For instance, scholars interested in determinants of voting for Democratic or Republican presidential candidates will certainly consider individual-level characteristics of voters such as age, race, gender, education, income, and political ideology. But they might also want to understand how characteristics of electoral districts, such as the unemployment rate, racial composition, percent of workers belonging to labor unions, the length of political incumbency of Democratic or Republican Congressional representatives, or the number of campaign appearances by the candidates shape electoral outcomes.