The aim of comparative research in soical epidemiology is to determine how risk factors for disease may vary within and between sociocultural and ethnic groups and in relation to outcomes. This aim assumes that measurement equivalence within and between social groups can be established, an assumption usually left untested. A model is presented here for deriving cross-culturally valid measures that are also intraculturally sensitive. Measurements so derived can then be used to compare cross-cultural and intracultural effects in a single analytic model. This approach is illustrated by pooling data on social stressors, social supports, and blood pressure from three studies. (AU)