oners in household counts. The item flagging cases in which all residents are only staying at the address temporarily is retained, though space to write in (a single) usual address has been added on the same page. Previous censuses effectively cut off the census questioning if the all-temporary condition applied, but the 1990 census had respondents continue with the full questionnaire.

The 2000 census form (which was displayed in Figure 2-1 and is also illustrated as part of Box 6-2) was the product of an extensive design overhaul based on several objectives. Principal among them were the user friendliness of the form (with attention to color and graphical cues to aid navigation) and easier automated data capture and optical character recognition. Like the 1990 form, the 2000 form asks a respondent to go through a preprocessing step, providing a count of people in the household in Question 1 before providing information on them on the subsequent pages. The list of include/exclude instructions is streamlined: college students are no longer mentioned in both categories, and the provision to include “foster children, roomers, or house-mates” consolidates multiple points used in previous censuses. The word “usual” does not appear anywhere in the instructions, though the working definition of usual residence as the place where one lives or stays most of the time is embodied in the last bullet point of both the include and exclude lists. Rather than “usual,” one of the bulleted instructions introduces a different concept—persons staying at the home on Census Day are to be counted there if they do not have another “permanent place to stay.” In terms of physical layout, the 2000 form is different from its predecessors in that the largest part of the instructions for Question 1 (the include/exclude lists) are placed after the answer space, not side by side (as in 1990) or before the answer space (1970 and 1980).

Though the “verbal and visual changes” improved the 2000 questionnaire, Iversen et al. (1999:121) criticize the 2000 census questionnaire development, arguing that “far less research attention was devoted to errors in responses provided by those who answer the census, to the factors that contribute to such errors, and to changes that could contribute to the accuracy and completeness of the information provided by respondents.”

Coverage Probes

In addition to the basic residence question (asking for a count or listing of household members), past decennial censuses have made different use of coverage probe questions. These questions, placed slightly later in the questionnaire, serve to jog respondents’ memories and prompt them to reconsider additions or deletions to the list of household members. Even before the advent of self-enumeration in the census, the 1950 census schedule included such a coverage probe as Housing Item 8: “We have listed (number) persons

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement