Evaluating Multilevel Models in Cross-Cultural Research

Abstract

To assess how culture influences the behavior of people, multilevel models are an immediate choice for modeling the relationship at the levels of the individual and culture. The authors propose structural equation modeling (SEM) to test the universality of psychological processes at the individual and culture levels. Specifically, the structural equivalence of the measurement (where the instrument is measuring the same construct across countries) is first tested with meta-analytic SEM. If the measurement is structurally equivalent, cross-level equivalence (where the instrument is measuring similar constructs at different levels) will then be tested with multilevel SEM. A large data set on social axioms with 7,590 university students from 40 cultural groups was used to illustrate the procedures. The results showed that the structural equivalence of the social axioms was well supported at the individual level across 40 cultural groups, whereas the crosslevel equivalence was partially supported. The superiority of the SEM approach and the theoretical meaning of its solution are discussed.Culture influences how people think, behave, and communicate. Hofstede (1980) defined culture as "the collective programming of the mind which distinguishes the members of one human group from another" (p. 25). This definition suggests that culture is shared among individuals. It has long been recognized that people within the same culture share similar languages, beliefs, values, and other psychological attributes when compared with people in other cultures. From a statistical point of view, people and culture are considered on two distinct levels, but people are nested within culture.One of the key research questions in cross-cultural research is whether psychological processes are universal across cultures. To answer this question, the dominant approaches are the individual-level analysis and the culture-level analysis (Leung, 1989). For example, based on more than 60,000 individuals from 40 nations or regions, Hofstede (1980) factor analyzed societal means (the average scores of people in each society) and identified four Downloaded from Cheung et al. / MULTILEVEL MODELS 523 cultural dimensions. These cultural dimensions, especially individualism-collectivism, have been frequently used to explain cross-cultural similarities and differences.Although Hofstede's approach was based on the culture-level analysis, Schwartz and Bilsky (1987) proposed a value-based model (the Schwartz value survey) presented at the individual level. Schwartz and Bilsky identified 56 values that could be grouped into 10 types. Further research has showed that these value types can be applied to explain the psychology of people from a diverse range of cultures (Schwartz, 1992).As international collaborations have increased, it has become popular to conduct research using large, cross-cultural data sets (e.g.

Urban land expansion under economic transition in China: A multi-level modeling analysis

“…First, in order to evaluate the proposed models, this study introduces the Bayesian Information Criterion (BIC) as a goodness-of-fit indices. Smaller values on BIC suggest better models in terms of model fit and parsimony (Cheung, Leung, & Au, 2006). Based on BIC, which quantify the degree to which the two-level model represents an improvement over the one-level models, we find that there are consistent reductions in deviances from the one-level model to the two-level model.…”

Do people’s world views matter? The why and how.

“…Using multilevel analyses, Cheung, Leung, and Au (2006) tested the factor structure of social axioms using meta-analytic structural equation modeling, and supported the five-factor model at the individual level and adopted a two-factor model at the culture level (labeled Dynamic Externality and Societal Cynicism, see Bond et al, 2004b). This pan-cultural structure provides core etic dimensions to study the functions, antecedents, and consequences of domain-general, context-independent world views in relation to global self-views.…”

Cross-cultural measurement equivalence of generalized trust. Evidence from the European Social Survey (2002 and 2004)

“…This implies that we want to investigate the cross-cultural equivalence of the three-item generalized trust measurement scale in ESS (Mullen 1995;Johnson 1998;Cheung and Rensvold 2000;Harkness et al 2003;King et al 2004;Cheung et al 2006). Following the literature on this topic, we can distinguish five different levels of measurement equivalence, or measurement invariance (Steenkamp and Baumgartner 1998, 80-81).…”