The final 2000 Census state response rates: myths and realities

Abstract

The decennial United States Census of Population determines not only political representation, but also how more than $185 billion of federal funds are allocated. Hence, there is intense interest in who responds, or does not respond, to Census queries. Both the popular media and the Bureau of the Census officials have developed a set of informal hypotheses which speak to who responds. Nine such hypotheses are identified and tested. For example, it is alleged that the elderly do not respond to the Census as often as others. Also, it is asserted that Republican politicians and talk show hosts diminished 2000 Census participation. No evidence is found to support these propositions. Empirical support is found for the media/Census view of the world in only two of nine cases and partial support in one. In two cases, the evidence neither supports nor rejects the hypotheses. In two cases, the evidence partially rejects the hypothesis, and in two cases, more strongly rejects the hypotheses. The results are sufficiently strong that they should inspire increased caution in media reporting and in the statements of Bureau of the Census officials.

Notes

1 Separate from this is the issue whether the Bureau should use statistical sampling techniques to adjust its basic, raw count of the population. In October 2001, the Bureau declined to do so. The financial and political implications were large. The City of Los Angeles, for example, might have gained an additional $325 to $375 million annually if sampling had been used (Schmitt, 2001).

2 A significant literature has arisen exploring the conditions under which individuals refuse to participate in the Census. A recent survey of this work is found in Presser et al. (2000).