Commentary on "A methodology for identifying married couples in Medicare data".

Authors of an article appearing recently in Demography propose a retrospective method to identify husband-wife pairs in administrative records. Although their premise is correct, they over-state the power of their methodology. The error stems from a misunderstanding of how persons entitled to both primary and secondary social security benefits are classified in program data. A retrospective look four years after the reference data shows that the number of pairs is increased by less than 10%, not by more than 50%, as the authors claim.