This section contains readings which discuss broad methodological issues related to measurement in cross-national and cross-cultural research, including definitions of types of measurement equivalence, reasons for lack of equivalence and study designs. The section exludes texts which discuss the statistical modelling and analysis of such questions; these are listed in the sections below.
Petersen and Novick (1976): Considers the selection of individuals (e.g. in education and hiring) based on test scores, and what it means to carry out “culture-fair” selection of applicants from different groups. The formulas are based on a situation where the selection rule is derived from an association between test score and a (gold standard) criterion of ability, but many of the concepts apply also to selection based on latent variable models (including rules based on assumptions of equivalence or non-equivalence).
Hui and Triandis (1985): A review of the notion of measurement equivalence in cross-cultural psychology. Discusses different types of equivalence (of different levels of abstraction) and different broad approaches to assessing equivalence.
Meredith (1993): Gives particular definitions of measurement invariance (with respect to a subset selection based on covariates) and various related terms, including “structural bias” (measurement models are of the same form, but with different parameter values), “weak measurement invariance” (conditional means and covariance matrices of items are invariant), and, in the context of confirmatory factor analysis models, “strong” (loadings and intercepts are invariant) and “strict” (residual variances are also invariant) factorial invariance and “configural invariance”. Gives conditions for these to hold, and their connections. Discusses briefly the binary (probit) latent trait model, where measurement invariance requires strict factorial invariance for the underlying continuous-index model.
Johnson (1998): First, a comprehensive review of the many different terms for and types of “equivalence” that have appeared in literature on cross-cultural and corss-national survey research. Groups these into “interpretive” and “procedural equivalence” - roughly, equivalence across cultures of concepts being measured and of their measures respectively. Second, reviews methods of establishing or assessing equivalence at all these levels, grouped under methods in (i) question development, (ii) questionnaire pretesting, (iii) data collection, and (iv) data analysis.
Hampleton et al. (2005): Considers issues related to adapting educational and psychological tests into multiple languages and cultures. Starting point is a set of practical guidelines for adapting such tests prepared by the International Test Commission (listed in Chapter 1). Other chapter consider different methodological issues and give extended examples. Several of the chapters are summarised separately in this bibliography.
van de Vijver and Poortinga (2005): An overview of conceptual and methodological issues in adapting educational and psychological tests into multiple languages and cultures. Summarises different types of bias and equivalence, and then different approaches to adapting or translating tests. (A chapter in Hampleton et al. (2005).)
Sireci (2005) In the context of adaptation of psychological or educational tests into a new language, discusses research designs which make use of bilingual respondents. (A chapter in Hampleton et al. (2005).)
Cook and Schmitt-Cascallar (2005): In the context of adaptation of psychological or educational tests into a new language, discusses methods for deriving comparable scores from tests given in different languages (and typically containing many non-overlapping items). (A chapter in Hampleton et al. (2005).)
Mills et al. (2006): Introduction to a special issue on comparative research in the journal International Sociology. Briefly discusses some central methodological challenges, including construct validity.
__________________________________________________________________