The lack of geographical information on individuals and households represents an important barrier to population-environment scholarship. Linking people to their proximate environments requires locational data, but such data can breach confidentiality. We contrast different approaches to anonymization through household clustering with a focus on one of 50+ Health and Demographic Surveillance Systems. We first develop three approaches to household clustering, then make use of the clusters to link households with data on proximate natural resource availability. Using the different measures, we estimate models of migration, natural resource use and food security followed by contrasts of results to increase understanding of the influence of these methodological choices. In the longer-term, our aim is to contrast the results with those that would be estimated with actual locational data (not yet in this paper!). Ultimately we aim to develop anonymization techniques of broader applicability within the network of Health and Demographic Surveillance Sites.
Presented in Session 46. Innovative Data and Methods for Population and Environment Research