Methodologies for Adjustments to Infant Mortality Estimates

Dennis J. Donahue, U.S. Census Bureau

This paper evaluates and compares various approaches to measuring and estimating infant mortality. Infant mortality is a key indicator of social development and general population health, but differences in definitions, reporting standards, and methodological differences preclude a comprehensive understanding of the issue. Direct measures, from vital registration systems, censuses, and/or surveys, are compared briefly with indirect estimation techniques (e.g., growth balance, child survivorship, etc.), as well as with methods developed to adjust reported infant mortality to account for known errors and underreporting (e.g., Hogan method, Kingkade and Sawyer method). Major survey programs have provided valuable data on infant mortality around the world (in addition to contributing many other important indicators), but problems with comparability (across countries, between surveys, and over time), and measurement error (i.e., biases inherent in retrospective recall of death events, measurement error due to small sample sizes, etc.) remain.