standards that are conceptually very different. It may be that the aggregation of multiple years of ACS data makes individual residence reporting problems offset each other and produces estimates consistent with what would be found using a “usual residence” standard. Yet it may also be that different applications of residence concepts produce highly discrepant estimates for some areas or population subgroups. Given the newness of full-scale ACS collection and—more fundamentally—the lack of collection of both residence locations on the same questionnaire, it is simply unknown how problematic the disconnect between the two programs will be.
Our recommendation for the ACS is directly analogous to our recommendation for the decennial census. We have suggested a program of research to gain further insight on how individual people’s concepts of residence match the decennial census “usual residence” concept, as well as the effectiveness of the census questionnaire in eliciting accurate “usual residence” information. The same line of reasoning holds for the ACS: it is unclear how well the ACS “current residence” concept or “2-month rule” fits with respondents’ own notions, and our review above raises considerable uncertainty as to how well the ACS questionnaire items and instructions match the survey’s own residence concept.
For the decennial census, we recommend that the Census Bureau collect “any residence elsewhere” information. As a starting point, these data should be collected as a major experiment of the 2010 census so that rigorous evaluation and analysis of those data can inform changes for later censuses. Likewise, we believe that the Census Bureau will ultimately be best served by the inclusion of a usual residence question in the ACS questionnaire, asked of each person and not of a whole household. The collection of both types of residence information is essential to measuring discrepancies between the residence standards and for evaluating the residence concepts of both the census and the ACS. As a first step—a means to gather baseline information for evaluation and refinement of a full-scale implementation of the question—the Bureau should include a usual residence question in its ACS experimentation. Current plans for the ACS include a “methods panel,” a subset of the ACS sample that may receive experimental versions of questionnaires or revised wordings of specific items. This methods panel would be an ideal setting for asking respondents whether their current residence is what they consider to be their usual residence and, if not, where their usual residence is.
Recommendation 8.3: The Census Bureau should plan to ask a question on the usual residence of each household member in the ACS questionnaire, in order to evaluate the extent of incongruity of residence standards between the long-form replacement survey and the decennial census. The usual residence question should first be tested using the survey’s experimental