Measuring reciprocity: Double sampling, concordance, and network construction

Abstract Reciprocity—the mutual provisioning of support/goods—is a pervasive feature social life. Directed networks provide way to examine the structure reciprocity in community. However, measuring involves assumptions about what relationships matter and how elicit them, which may impact observed reciprocity. In particular, practice aggregating multiple sources data on same relationship (e.g., “double-sampled” data, where both “giver” “receiver” are asked report their relationship) have pronounced impacts network structure. To investigate these issues, we concordance (ties reported by parties) set directed, double-sampled support networks. We find low people’s responses. Taking either union (including any ties) or intersection only concordant results dramatically higher levels Using multilevel exponential random graph models from 75 villages India, show that changes cannot be fully explained increase number ties produced layer aggregation. Respondents’ tendency name people as givers receivers plays an important role, but this varies across contexts type. argue no single method should necessarily seen “correct” choice for aggregation Methods depend research question, context, question.