Cross-national Comparability of Survey Attitude Measures

Measurement Equivalence Across Subnational Groups: An Analysis of the Conception of Nationhood in Switzerland

“…The initial structure was chosen as a function of the EFA results ( Table 1): Having ancestry, being born, having lived in Switzerland for most one's life, being a Christian, and having Swiss citizenship were assigned to a first ethnic dimension. Although previous cross-national studies on the ISSP items tapping the conception of nationhood (Heath et al, 2009;Reeskens & Hooghe, 2010) revealed strong cross-loadings for the ''having the national citizenship'' item, in the Swiss sample, this item was clearly a part of the ethnic dimension. Respecting Swiss political institutions and laws and being able to speak the language were assigned to a second civic dimension.…”

“…Although the cross-national comparability of these scales has received considerable methodological attention (Heath, Martin, & Spreckelsen, 2009;Medina, Smith, & Long, 2009;Reeskens & Hooghe, 2010), no study, to our knowledge, has so far examined whether they can be validly compared or pooled in countries with distinct subnational groups. Although some common causes of nonequivalence are less likely to be problematic within a country (e.g., data collection and sampling methods; Heath et al, 2009), distinctive features of the subnational groups, in particular different languages (Davidov & De Beuckelaer, 2010), may jeopardize measurement equivalence.…”

“…Although multigroup confirmatory factor analyses (MGCFA) are used to test whether configural (and higher forms of) equivalence is reached, it is advised (by Byrne, 2010) to study the factor structure within each group before proceeding with MGCFA to detect whether the pattern of salient and nonsalient factor loadings is similar across the groups. For instance, Heath et al (2009) performed exploratory factor analyses (EFAs) on the ISSP conception of nationhood scale for each country separately. Results showed that the ''having the national citizenship'' item hampered configural equivalence (Reeskens & Hooghe, 2010), either by loading on the ethnic factor (i.e., salient loading on the ethnic factor and nonsalient loading on the civic factor) or cross-loading (i.e., salient loading on both factors).…”

The Plausibility of Cross-National Comparisons of Internet Use Types

“…In fact research in diverse areas has consistently failed to find invariance across countries. Examples include studies on social values (Davidov, Meuleman, Billiet, & Schmidt, 2008;Davidov, Schmidt, & Schwartz, 2008;Heyder & Schmidt, 2003), engagement with technology and science (Stares, 2009), national identity (Heath, Martin & Spreckelsen, 2009;Medina, Smith & Long, 2009) and consumer behavior (Benedict, Steenkamp & Baumgartner, 1998). In all these studies, constraining measurement models to be equal across countries resulted in models with poor fit, suggesting that the same items do not measure the same constructs across different countries.…”

“…Even where researchers have done their utmost to make individual items comparable, comparisons should be done with caution and full awareness that there might be validity issues. Second, Heath et al (2009) recommend dropping of countries from the analysis for which there is no measurement equivalence. The problem with this approach is that the resulting studies are likely to end up considering homogenous sets of countries.…”