Statistics play a central role in constructing social problems and in policy debates. Claimsmakers' estimates typically exaggerate the extent of these problems; research statistics are considered more accurate. Yet the National Incidence Studies of Missing, Abducted, Runaway, and Throwaway Children (NISMART) produced an estimate for family abductions far higher than the figures offered by missing-children activists. Extrapolations for family abductions based on official statistics from eight statewide missing-children clearinghouses are much lower than the NISMART estimate. In explaining the anomaly, this paper notes that NISMART's methods defined family abductions broadly, thereby producing a large figure, while police probably underreported cases to the state clearinghouses. Statistics, especially statistics gathered through social research, play an important role in policy debates because they tend to be regarded as indisputable facts. The operational definitions, coding practices, and other methodological choices that underlie every social statistic tend to be ignored or overlooked. Unquestioning acceptance of statistical data by the press and by policymakers illustrates society's limitations for critically assessing social statistics. The article includes a discussion of the politics of statistical research and the implications of a statistical anomaly such as the one examined here. Notes, tables, references