This paper presents a methodological approach for examining variation and consensus both within and between research settings and for addressing issues of generalizability and replicability. The comparison is based on how individuals diagnosed with diabetes and living in three Canadian Anishinaabe (Ojibway) communities explain diabetes and talk about their responses to the illness. Two kinds of interview format are used--an open-ended explanatory model type interview and a more structured, true-false, interview, amenable to analysis with cultural consensus theory. The responses given in both interviews converge on a set of explanations which can be found in all three communities, although differences occur in how these explanations are framed and emphasized. Implications of these differences and how these accounts relate to how individuals respond to diabetes are discussed. It is argued that the analysis of both interview formats leads to a deeper and more finely nuanced representation of understandings about causes of diabetes across the three communities than could be achieved by using one alone.