An empirical test of household identification risk in geomasked maps

Abstract

ABSTRACTGeomasking techniques displace point data to new locations in order to protect privacy while maintaining overall spatial distributions. If the end users of geomasked data are unaware that the data are masked, there is a risk that they will incorrectly associate individuals at the new locations with the masked data attributes. The probability of correct and false household identification depends on human understanding of whether maps contain masked coordinates and the spatial relationships of the points to contextual geographic data. Using a map-based experiment, this study finds that confidence in performing a household identification is substantially lowered when masked points are situated equidistantly between residential parcels. Despite initial notifications that data are masked, map users often report confidence in assigning masked points to specific households. Only map users who receive frequent notifications that the points are masked have reduced confidence in associating them with particular households, thereby lowering identification risk.