Cross-cultural generalization: Using meta-analysis to test hypotheses about cultural variability

Ones, D. S., Dilchert, S., Deller, J., Albrecht, A.-G., Duehr, E. E., & Paulus, F. M.
In A. M. Ryan, F. T. L. Leong, & F. L. Oswald (Eds.),
Conducting multinational research projects in organizational psychology: Challenges and opportunities (pp. 91-122).
Washington, DC: American Psychological Association.
(2012)

When differences are observed across samples in cross-cultural research, true cultural differences (and sample representativeness) are not the only explanations that ought to be considered. A main thesis of this chapter is that when differences are observed, findings can be due to chance (sampling error) as well as other statistical artifacts (see Hunter & Schmidt, 2004). Testing whether cross-cultural variability in findings is due to real effects of culture or such statistical artifacts is an important step that is essential in cross-cultural research. Addressing the biasing influences of statistical artifacts may help reveal cross-cultural universals. The magnitude of cultural variation in results can be empirically examined using approaches of psychometric meta-analysis. A major contribution that meta-analytic techniques can make to cross-cultural research is to enable researchers to test the cross-cultural generalizability of relationships. In the remainder of this chapter, we review, illustrate, and discuss three unique applications of meta-analysis to examine cross-cultural effects. First, we review and discuss pooling findings across intracultural studies to examine questions of cross-cultural generalizability. Second, we illustrate the value of applying meta-analysis to carefully conducted intercultural studies to examine the same question. Third, we demonstrate the use of meta-analysis to examine transcultural variability using primary data collected from different cultural settings. For each type of application, we offer a brief background, review and present illustrative findings, and discuss contributions and potential limitations.