Regional Patterning in Illness Theories: Analyses With Different Types of Optimal Scaling

Abstract

Optimal scaling techniques provide holocultural researchers with interesting ways to explore relationships in multivariate data. Carmella Moore's analysis of regional patterning in Murdock's theories of illness data was one of the first uses of these techniques with cross-cultural data. This article shows the flexibility of optimal scaling approaches to holocultural data. It also illustrates how different ways of conceptualizing and transforming data may affect the conclusions drawn from these techniques. The author uses dual scaling techniques to produce two alternatives to Moore's analysis. The first treats Murdock's codes as successive categories data and uses a clustering technique to search for regional groupings. The second identifies regional patterning with forced classification. Both techniques identify the Circum-Mediterranean region as possessing a unique configuration of illness theories and highlight a distinction between theories of witchcraft and sorcery.