New research advances methods for estimating the risk of re-identification in nonprobability samples, including those drawn from registers and opt-in panels. The findings were published in the Journal of Official Statistics, with Westat’s Minsun Riddles, PhD, and Tom Krenzke, MS, among the coauthors.
Building on earlier probabilistic approaches developed for representative government surveys and later extended to subpopulation registers (i.e., data sources tracking specific subgroups), the study addresses cases with unknown register membership and sampling mechanisms. The authors demonstrate that a probability-based reference sample can be used to infer population parameters and assess disclosure risk in a probabilistic modeling framework.
The approach is illustrated through simulations and an application using the Survey of Doctorate Recipients (SDR), drawn from a register of PhD recipients from accredited U.S. institutions.
Learn More
Measuring Risk of Re-Identification for a Nonprobability Sample Using a General Reference Sample
Natalie Shlomo, PhD; Minsun Riddles, PhD; and Tom Krenzke, MS