Toward an assessment of continuous measurement: A comparison of returns with 1990 census returns for the portland test site

Abstract

Under the auspices of the ``American Community Survey'' (ACS), Multnomah County, Oregon was selected as a test site in 1996 for the ``Continuous Measurement''(CM) option being considered for Census 2000 and beyond. The National Research Council recommended against substituting CM for the long form in the 2000 census because there were too many unanswered questions for which research was needed. This paper is part of the initiation of an empirically-based discussion of CM's capability to provide small area data comparable in quality to that provided by the census long form, the current gold standard for detailed, small area data. We compare mail return rates of the 1996 ACS to the 1990 mail return rates of the census long form for tracts in the test site and find that overall return rates are virtually the same -- 69 percent. However, ACS return rates are higher than those of the 1990 long form in that quartile of the tracts where the return rates for the long form were the lowest and lower in that quartile of the tracts where the long form rates were the highest. Taking into account the effect of ``regression-to-the-mean,'' the ACS return rates are higher than what one would expect under this artifact, an encouraging finding. Further comparisons between the ACS and traditional census data can serve a useful role not only in understanding ACS but also in providing new insights on traditional census products.