When conducting the 2000 census, the Census Bureau utilized the "hot deck imputation" method to resolve gaps and conflicts in their data by inferring that an address or unit that returned unclear or conflicting information had the same population characteristics as their geographically closest neighbor of the same type (i.e., single family dwelling, apartment, etc.) Using this method led to a 0.4% increase in the total population overall, but because those increases in population were not even across all the states, it resulted in North Carolina gaining the additional congressional seat that Utah would have received had the imputation method not been used. In 2001, Utah filed a federal lawsuit against the Secretary of Commerce and Acting Director of the Census Bureau, alleging that the hot deck imputation method violated the U.S. Constitution's Enumeration Clause and the Census Act's prohibition on "statistical sampling" when determining population, and seeking a court order requiring the Secretary of Commerce to recalculate the numbers and recertify the official census results.
In 2002, the U.S. Supreme Court ruled in favor of the Census officials, holding that the use of the "hot deck imputation" method was permissible under the Census Act and the Enumeration Clause. The Court's ruling hinged on its finding that "hot deck imputation" was wholly distinct, in form and in function, from the type of "statistical sampling" that the Court had previously ruled was prohibited by the Census Act in Department of Commerce v. U.S. House of Representatives. Sampling, as that term is used in the Census Act, involves an attempt to extrapolate information about a larger population by examining a smaller one and is normally motivated by cost saving concerns. Imputation, on the other hand, involves the filling in of missing data as part of a larger effort to count all individuals one by one, with the "hot deck" method being used to assure that an individual unit used in this process is "chosen nonrandomly" and will resemble other individuals who were selected by "the fortuitous unavailability of data." These fundamental differences, along with the fact that the Bureau has utilized this implementation method for decades without any opposition from Congress, led the Court to conclude that imputation is not covered by the Census Act's prohibition on statistical sampling.
Significance: The Census Act does not prohibit the Census Bureau from utilizing a method of imputation to resolve conflicts and gaps in census data because, unlike statistical sampling, it does not use smaller population sets to make estimates about a larger population's qualities.
U.S. Supreme Court - 536 U.S. 452 (2002)
- Opinion - 6/20/02