Methodological Requirements for Conceptually Valid Research Studies on the Behavioral Effects of Malnutrition

摘要: A number of papers and chapters have been addressed to methodological issues in research on the effects malnutrition human behavior (Tuddenham, 1974; Warren, 1973; Elias, 1976; R. E. Klein, Irwin, Townsend, Engle, unpublished manuscript). Most stressed importance designing studies that permit investigator infer with a high degree confidence statistical relationships between nutritional behavioral variables reflect influence nutrition per se, not one or more confounding (i.e., nonnutritional) variables. For example. Warren’s review criticized used nonexperimental designs grounds representing status may correlated social environmental factors themselves could accounted for relationships. Tuddenham reviewed five classes studying described as adequate only those experimental provided random assignment treatments thus avoided problem selection bias. Klein colleagues discussed several problems interpretation retrospective behavior, including possible onset severity duration hospitalization, covariation home conditions history, difficulty separating acute from chronic components history. In short, primary concerns are summed up assertion by Elias (1976) “the issue is non-nutritional factors” (p. 455).