ABSTRACT
This chapter explains a model that can be used to explain why specific indicators function differently across cultural contexts. It proposes a multilevel confirmatory factor analysis model with random factor loadings to capture cross-group variation in the strength of indicators. The chapter explores the Multilevel confirmatory factor analysis (MLCFA) with random loadings in an analysis of the items measuring ethnic citizenship conceptions included in the 2003 round of the International Social Survey Programme. The MLCFA model conversely conceptualizes cross-group inequivalence as random effects. MLCFA is a more parsimonious model that is especially useful when large numbers of groups are being compared. The chapter utilizes a Bayesian estimation procedure that is better suited to deal with the relatively small sample size at the country level. It presents the results of a Monte Carlo simulation study evaluating the reliability and validity of the procedure.
