This project (consisting of two collaborative awards, this and 9807373 to Roberta Baer at the University of South Florida) continues a line of research using the cultural consensus model, a statistical technique for assessing the degree of homogeneity in the underlying models of respondents from a local community or cultural population, in this case of specific illnesses (diabetes, AIDS, the common cold) used by Latin American respondents. The research will focus on the systematic examination of sources of intracultural variation, in particular on the effect of specialized knowledge or personal experience of a specific disease. This research builds on and expands earlier work done on 9204555 by collecting new samples for comparison to previously collected samples. With this new data it will be possible to explore variation in beliefs across diverse groups within and across sites. Three hypotheses will be tested: that differences in beliefs are due to differences in `experience` or exposure to the illness in the general community; that beliefs about illness will vary by whether the individual respondent has had the illness; and that the differences in belief structure between professionally trained and lay respondents will be greater than differences from nationality or culture. The project will collect new data about AIDS and the common cold from six sample populations in Guatemala, Mexico and south Texas, and from diabetics and physicians in Guadalajara, Mexico and south Texas. In addition some simulation work will be done to assess the effects of response bias and skewed answer keys on consensus estimates. The consensus model brings rigor to the description of community beliefs by assessing the concordance or homogeneity across informant' responses to a series of related questions and proving estimates of normative answers, similar to that which a psychometrician might apply to analyzing the reliability of test items.