AN ANALYSIS OF NET IMMIGRATION IN CENSUS COVERAGE EVALUATION

National surveys monitored growth in the foreign-born population for t he 1980s, especially net undocumented migration's continuing role, but the 1990 census portrayed an even larger foreign-born population than these surveys. Undercoverage in 1990 could have been higher than init ially presented because preliminary studies may have insufficiently ac counted for decadal net immigration. Assumptions intended to maintain a high undocumented undercount performed poorly when census counts of foreign-born residents became known. Any point estimate for net undocu mented migration, calculated as a residual, is likely to be biased by assumptions and data gaps for components of calculating net legal immi gration, especially in the direction of underestimation. A reasonable statement is that at least 2.1-2.4 million undocumented residents were enumerated in the 1990 census. The number of unenumerated undocumente d residents may easily have ranged between 0.5 million and 3.0 million , and a narrower range of 1 million to 2 million is plausible. Despite the importance of undocumented migration measurement for census evalu ation and policy purposes, differences among various undocumented esti mates are more likely to stem from discrepancies in universe, referenc e dates, or individual judgment, rather than analytic refinement. Bett er measurement of the foreign-born population or its census coverage w ould aid in setting upper limits on net undocumented migration.