Skip navigation.

How to Measure and Compare Context Across-Countries? CFA and Measurement Invariance Tests of Political Culture

TitleHow to Measure and Compare Context Across-Countries? CFA and Measurement Invariance Tests of Political Culture
Publication TypeConference Paper
Year of Publication2013
AuthorsMonica, Poletti, and Kotnarowski Michal
Conference Name2nd European Conference on Comparative Electoral Research
Keywordsmeasurement equivalence, multigroup confirmatory factor analysis, political culture

Plenty of studies have been focusing on political culture, but not many have taken an in-depth look to issues of operationalization and measurement validity of this concept, particularly important in comparative perspective. In this paper we would like to make a contribution in the direction of filling this gap, testing the cross-country comparability of synthetic measures of five dimensions (latent factors) of political culture: (1) social capital; (2) moral values; (3) confidence in institutions; (4) image of democracy; and (5) left-right conflict representation. This measurement invariance test will serve as a base for conducting cross-cultural analyses on the effects of political culture on party choice and turnout in one of the data book chapter of the True European Voter project. Choosing appropriate measures for the analysis of cross-cultural data is, however, a crucial topic to a much broader group of scholars.
Our final aim is to construct a dataset with valid macro-level context variables, using countries in particular points of time as unit of observation. Since the set of indicators for the chosen dimensions of political culture can be measured only at the individual level, we need to transform them into synthetic variables. Before to do that, however, we need a preliminary data validation. Assuming that a latent factor exists for each of the dimensions outlined, run confirmatory factor analysis for each country-time unit using the last edition of the European Values Study (EVS). Then, in order to check their comparability across country/time, we assessed the measurement invariance by putting restrictions on the measurement model, using the procedure of multigroup confirmatory factor analysis (MGCFA). We tested the three forms of invariance that are important for cross-national comparative research: configural invariance, which implies that the measurement model hold across country/time but that comparisons of the measures are still not meaningful; metric invariance, which implies that configural invariance holds and that comparison of relationships between unstandardized measures become meaningful; and scalar invariance, which implies that metric invariance holds and it also becomes meaningful to compare the means of the measures (Meredith, 1993; Davidov, Schmidt and Billiet, 2011). We were able to check whether the difference in dimensions of political culture across country/time can simply be attributed to error scores or they instead reflect true scores in the real world. Results are mixed but points in a promising direction for political culture cross-country analysis.

Poletti,Kotnarowski_HowToMeasureAndCompareContextAcross-Countries_2ndTEVConf_Rhodes.pdf910.58 KB