Structural Equivalence in Multilevel Research

Abstract

In cross-cultural research, there is an increasing interest in the comparison of constructs at different levels of aggregation, such as the use of individualism—collectivism at the individual and country levels. A procedure is described for establishing structural equivalence (i.e., similarity of psychological meaning) at various levels of aggregation based on exploratory factor analysis. A construct shows structural equivalence across aggregation levels if its factor structure is invariant across levels. The procedure was applied to the Postmaterialism scale of the World Values Survey. The similarity of postmaterialism at the individual and country levels could not be unambiguously demonstrated; a likely reason is that the concept does not have an identical meaning in countries with low and high gross national products.

Joint Impact of Interdependence and Group Diversity on Innovation

“…Given the hierarchical structure of our data, with individuals nested within groups, the factorial structure of the constructs under study was examined at both the individual and group level of analysis using exploratory multilevel factor analysis ( Van de Vijver & Poortinga, 2002). When constructs are structurally equivalent at two levels of analysis, this forms important evidence that they have the same psychological meaning across these levels.…”

“…If there is no close agreement between individual-and group-level factor structures, different constructs are needed to describe individual and group-level differences (Muthén, 1991). In the process of exploratory multilevel factor analysis, seperate factor analyses are carried out on the matrix of the sample means of the various items, and on the matrix of the individual deviation scores from the group mean ( Van de Vijver & Poortinga, 2002). The factorial agreement of the factor structures is evaluated after target rotation has been carried out by means of Pearson correlation coefficients.…”

Personality in Cultural Context: Methodological Issues

“…They aim at comparing the structure of a construct at individual and cultural levels. Muthén (1991Muthén ( , 1994 has developed a procedure for multilevel confirmatory factor analysis, whereas Van de Vijver and Poortinga (2000) developed a procedure for exploratory factor analysis.…”

Are We Drawing the Right Conclusions? The Dangers of Answer Format Effects in Empirical Tourism Research

“…Possible techniques include the investigation of the invariance of factorial structures (Van de Vijver and Poortinga, 2002;Van Herk, Poortinga and Verhallen, 2004) underlying the responses of different (cultural) subgroups in order to investigate structural equivalence of the constructs measured as well as investigations of item response patterns across all questions in the questionnaire (Byrne and Campbell, 1999). If the number of countries in the data set is low, Van de Vijver and Poortinga (2002) recommend regression procedures to investigate the associations of constructs with context variables. Greenleaf (1992a and b) suggests the selection of a specific subset of extreme response style variables that can subsequently be used to determine -across a variety of constructs -the information and bias proportions of answers, which enables corrections for the distortions before data analysis.…”

“…The second stream of research is methodological in nature and introduces techniques to determine the existence of response styles in cross-cultural studies and correct for the bias (Cheung and Rensvold, 2000;Byrne and Campbell, 1999;Greenleaf, 1992a and1992b;Van de Vijver and Poortinga, 2002;Welkenhuysen-Gybels, Billiet and Cambre, 2003) with recommendations ranging from very simple approaches, such as investigating if systematic response patterns can be detected for the same cultural group, to modeling approaches to extract the extreme response and acquiescence bias from the actual information content and then correct the data accordingly.…”