Measurement Issues in Cross-National Research

Abstract

Substantive inference from cross-national studies have important implications for theory (e.g., because they reveal insights into generalizability and boundary conditions) and managerial practice (e.g., because they offer guidelines to MNC managers). However, few empirical studies attend to measurement issues involved in cross-national research, and still fewer recognize the risk of inferential errors that are likely to occur by overlooking measurement issues. We discuss four measurement issues, namely (1) standardized versus unstandardized coefficients, (2) the impact of measurement error and unequal reliability, (3) the overall error rate and simultaneous analysis and (4) construct equivalence. Using illustrate examples we demonstrate the nature of each of these problems, the likely impact they can have on substantive conclusions, and approaches for tackling these problems. Additionally, we reanalyze a recently published three-nation study by Dubinsky, Michaels, Kotabe, Lim and Moon [1992] to clarify these measurement concerns, highlight a methodological approach, and delineate the extent and severity of inferential errors. Our reanalysis shows that the interactive effects of these measurement issues are pervasive, complex and unpredictable. We close with implications for cross-national research in general.