Tract-level, Residence-rule Independent, Coverage-error Adjustment of the American Community Survey (ACS) Using An Administrative Records Match

In order to reduce variance and correct for coverage of ACS estimates, there is a desire for controlling estimates of demographic characteristics at the tract level. Currently, intercensal population controls are based on “usual residence” and are not available at the tract level, while the ACS produces estimates at the tract level that are based on “current residence”. This project proposes a way to use new controls, obtained by matching the ACS sample to an administrative records file, and then controlling sample estimates of administrative record tract counts to their known tract totals. By matching to administrative records addresses, this procedure achieves a consistent residence rule between sample and control. To evaluate the effects of coverage error in the administrative records and matching error with the sample, the procedure is applied to the Census 2000 long-form, where the correct population totals are known from the Census 2000 short-form.