Cross-Cultural Research

An accurate and flexible method for measuring cultural diffusion is presented. When the diffusion of pairs of traits is measured, the method constitutes a solution to Galton's problem. Application of the method to data on the association of patrilocal residence with patrilineal descent indicates that the association is based on diffusion rather than function.

The various samples and sampling universes available to the researcher about to undertake a cross-cultural study are systematically reviewed. Discussed are the Human Relations Area Files, the "World Ethnographic Sample," the Ethnographic Atlas (both the complete atlas as well as the summary version), the Standard Cross-Cultural Sample, the Quality Control Sample (or HRAF Probability Sample), and the "Standard Eth nographic Sample." Following an assessment of each sample, recom mendations are made as to the conditions and circumstances under which the samples should be utilized. The discussion concludes with recommen dations for drawing a sample from one of the samples or sampling universes.

Two aspects of Galton's problem are considered here—the tendency for cultural diffusion to inflate coefficients of correlation and also its tendency to inflate the number of independent cases for significance tests. Spatial autocorrelation is seen as a generally useful tool for measuring and control ling the influence of diffusion. The Strauss-Orans cluster reduction method reduces spatial autocorrelation in 2 x 2 matrices. In special cases, cluster reduction may eliminate Galton's problem entirely by eliminating spatial autocorrelation of key variables. Wirsing's second order partials reduce in flated correlation coefficients. The Orcutt-James correction reduces inflated sample sizes. Where none of these approaches is useful, Zucker's cluster difference test may be helpful. HRAF plans to apply these methods to its HRAFLIB computer program library.