Linking Individuals to Areas: Protecting Confidentiality While Preserving Research Utility

Abstract

Modern computational capabilities have brought about concerns about risks associated with the level of information disclosed in public datasets. A tension exists between making data available that protects the confidentiality of individuals while containing sufficiently detailed geographic information to underpin the utility of research. Our aim is to inform data collectors and suppliers about geographic choices for confidentiality protection and to balance this with reassurance to the research community that data will still be fit-for-purpose. We test this using simple logistic regression models, by investigating the interplay between two geographical entities (points for the observations and polygons for area attributes) at a variety of scales, using a synthetic population of 22,000 people. In an England and Wales setting, we do this for individuals located by postcodes and by postal sector and postal district centroids and link these to a variety of census geographies. We also ‘jitter’ postcode coordinates to test the effect of moving people away from their original location. We find a smoothing of relationships up the geographical hierarchy. However, if postal sector centroids are used to locate individuals, linkages to Lower/Medium Super Output Area scales and subsequent results are very similar to the more detailed unit postcodes. Postcode locations jittered by 500–750 m in any direction are likely to allow the same conclusions to be drawn as for the original locations. Within these geographic scenarios, there is likely to be a sufficient level of confidentiality protection while statistical relationships are very similar to those obtained using the most detailed geographic locators.