Eilert Sundts hus
4th floor (map)
Moltke Moesvei 31
Lecturer: Dr. Michael Shalev, Senior Lecturer in Sociology and Political Science,
Department of Sociology, The Hebrew University of Jerusalem, Israel
Main discipline: Political Science, Sociology
Dates: 24 - 28 July 2006
Course Credits: 10 pts (ECTS)
Limitation: 30 participants
Research in comparative political economy seeks to explain differences between advanced capitalist democracies in economic and social policies and outcomes. As is true of sociology and political science generally, the field is methodologically split by the quantitative/qualitative divide. Single and comparative-historical case studies stand on one side of this divide and ever more sophisticated quantitative studies on the other. With few exceptions (notably QCA and other methods developed by Charles Ragin), quantitative comparative research typically utilize multiple regression. Because of the small number of cases being compared, regressions often pool cross-sectional observations at multiple time points.
This course will highlight incompatibilities between the causal model underlying standard regression analysis and core elements of comparative research. These tensions are especially great for pooled regression analysis. Through close study of key examples it will be shown that the precision of high-tech analysis is often illusory, while low-tech methods of exploratory analysis (tables and charts) may yield deeper insights into causal processes and more credible assessments of evidence.
Not all multi-country comparative research seeks broad, context-free generalizations about causal relationships between variables. Esping-Andersen's work on worlds of welfare represents a more contingent, historically conscious research strategy based on a typology that identifies qualitatively different "regimes". The course will introduce non-regression approaches to systematically verifying the existence and determinants of welfare regimes. Here too it will be shown that quantitative research need not be synonymous with regression and that data analysis can be carried out without losing sight of cases or of causal complexity. We will also consider two alternative approaches: Ragin's methods and combining regression with case studies.
Day 1 (Lectures 1 and 2): The core methodological issues.
Introduction to the central question of the course: what kind of quantitative analysis is appropriate to comparative research? Review of the core methodological challenges posed by cross-national studies: small-n, causal heterogeneity and interaction (Ragin 1994:105-130).
The logic and the strengths of the primary quantitative tool used by comparativists: multiple regression (MR). Disadvantages of the MR approach for socio-political analysis generally (Abbott 1988) and comparative political economy specifically (Hall 2002). Two different usages of MR (descriptive and explanatory). Outline of non-regression quantitative approaches: tables and visualization, Ragin's methods and biplots.
Day 2 (Lectures 3 and 4): The mainstream solution - cross-sectional regressions pooled over time.
Non-technical introduction to pooled regression. Illustration from Garrett's Partisan Politics in the Global Economy (1998:Chapter 3). Observations on two examples of "best practice": Huber and Stephens' (2001) Development and Crisis of the Welfare State (Chapter 3), and Hall and Franzese's (1998) study of central banks. Two theoretically interesting approaches to pooling: (1)Multilevel models explaining dynamic variation by country-level institutional differences (Western 1998). (2)Pooled analysis within regimes - class discussion of Rueda and Pontusson (2000).
Day 3 (Lectures 5 and 6): Quantitative alternatives to regression: exploratory analysis and QCA.
Exploratory analysis implies the use of tables and charts to help identify patterns in compact cross-national datasets, including elective affinities (clustering of causes), interaction effects and deviant cases. Illustration using Rothstein's (1990) study of union membership variation.
QCA and "fuzzy set analysis" are methods developed by Ragin that use boolean algebra to identify causal configurations. Evaluation of applications of these methods by Ragin (2000:286-308) and Katz, vom Hau and Maloney (2005).
Day 4 (Lectures 7 and 8): The typological approach to explanation and how to test it.
Esping-Andersen's distinction between three welfare state regimes will be our model (Esping-Andersen 1990:Chapters 2 & 3). Empirical and methodological problems of "Three Worlds" (see the survey by Arts and Gelissen 2002).
An alternative to either regression or exploratory methods is factor analysis or PCA of multiple indicators and graphic visualization of results vis-à-vis country cases. The biplot (Jacoby 1998:73-85) extends this approach.
Day 5 (Lectures 9 and 10): Where do we go from here?
What practical solutions are there to the methodological problems of small-n cross-national research, other than regression? Ragin's methods are one alternative. This course will have presented two others: exploratory and typological analysis. The issues and alternatives are summarized by Esping-Andersen (1993) and Shalev (forthcoming).
Particular attention will be paid to the scope for combining different methods. The use of multiple quantitative methods is illustrated by comparative studies of union membership by Ebbinghaus and Visser (1999) and Oskarsson (2003). The option of combining case study and quantitative analysis is illustrated by Huber and Stephens' study of welfare states (2001:compare pp. 181-201 with Chap. 3) and explicated by Rueschemeyer (1991).
Course Reading List (total approx. 550 pages)
Michael Shalev is a sociologist specializing in comparative welfare state studies and methodology in comparative research. His work has been divided between studies of the political economy of Israel and comparative political economy. Currently he is engaged in research on linkages between class and gender inequality in 17 OECD countries.