Abstract
Continuous variables such as income or turnover are dominated by outliers, and those outlying units are precisely the records most at risk of re-identification. This paper introduces disclosure-risk measures for continuous microdata masking that use robust statistics — notably robust Mahalanobis distances — to detect which records are unusual and therefore most exposed. By targeting protection at high-risk outliers rather than treating all records alike, the measures support more efficient masking that protects the vulnerable without over-distorting the rest. Presented at Privacy in Statistical Databases, the work links robust statistics to disclosure-risk assessment.
statistical disclosure control disclosure risk microdata masking robust statistics