Identification Risks of Microdata

In the social sciences and in other fields where data from individual respondents are collected, it is an important concern that information remains confidential and that the identity of individual data providers is not disclosed. For several realistic scenarios, the present article examines the risks of discovering the identity of data providers even in cases where the microdata have been previously anonymized by some procedure. The results show that, even under peculiarly risky conditions, the identification risk is smaller than has been assumed by previous research. The inherent unreliability of measurement is found to be an implicit protection of anonymity and a natural barrier to identification.