Unexpected Correlations and the Cross-Cultural Method

Abstract

Cross-cultural studies, in which the cases are societies rather than individuals, may be subject to errors due to individual differences within societies, cultural differences in behavior variables, and the implicit theories of the observers. In addition, antecedent variables may correlate more highly with unexpected consequences than with dependent variables predicted from theory. Ratings provided by Whiting and Child (1953) were correlated and the resulting matrix of correlations between all antecedent and all dependent variables was examined. Off-diagonal correlations, e.g., that between oral socialization anxiety and anal explanations of illness, were often found to be as high as correlations predicted from psychoanalytic theory. It was concluded that the caution suggested by Campbell and Fiske (1959) for validity studies should be applied to crosscultural studies.