A method, derived from wildlife biology, of estimating the size of a study population that makes use of overlapping, incomplete, but intersecting sets of data, to derive reasonably accurate numerators and denominators for epidemiological study. If two independent sources are available, a is the number of cases found in both sources, b is the number found only in the first source, and g is the number found only in the second source, the maximum likelihood population estimate is the product of the total found in each source divided by the total found in both sources, that is, (a+b)¥(a+g)a. If three or more sources are available, log linear methods are used to model the degree of dependency among the sources. The method is especially useful in studies of elusive and ill-defined populations, such as sex workers, homeless people, and migratory agricultural laborers.
Subjects: Public Health and Epidemiology.