The conceptual challenges of defining residence among movers overlap with operational concerns:

  • Could the timing of the census mailing—or, for that matter, the placement of Census Day itself on the first day of a calendar month— exacerbate difficulties in response (e.g., if there is a strong tendency for new apartment leases or the like to start at the beginning of a month)?

  • What becomes of census questionnaires that arrive at recently vacated units, and how good will proxy data from neighbors or landlords (collected in follow-up activities) be? Should forwarding or other information about the departed household residents be gathered?

  • How different are the households that occupy a unit before and after the move? In particular, the accuracy of matching makes the treatment of movers a major concern when comparing census returns with independent measures from postenumeration surveys, such as the 2000 Accuracy and Coverage Evaluation.

Work that has been done on the general timing of moves suggests that the March–May time frame of peak census activity is the beginning—but not the peak—of the basic “moving season.” Schachter and Kuenzi (2002) analyzed data from a migration history module conducted as part of the 1996 panel of the SIPP. The module (administered between June and September 1996), included questions on the month and year respondents moved to both their current residence and their immediately previous residence; by subtraction, the data also provide insight on duration of stay at the previous residence. Peak moving activity occurs in the summer months of June, July, and August: those 3 months accounted for one-third of all recorded moves in the data set. The months central to decennial census questionnaire mailout—March and April—are periods where moving activity begins to escalate from the low-move-activity winter months.

Neither the 2000 census nor its past few predecessors defined formal residence rules for the handling of movers; technical documentation for data files for the 2000 census note only that “people who moved around Census Day were counted at the place they considered to be their usual residence” (U.S. Census Bureau, 2001:C-2). However, some census operations have explicitly tried to determine how well movers are handled. Lowry (1987:7–9) describes a limited check on movers that was conducted during the 1970 census; Census Bureau staff were allowed to examine change-of-address cards filed with local post offices in parts of 17 metropolitan areas. If the two addresses were in the same census district, enumerators were deployed to see if the person or family had been counted at one of the addresses. The operation suggested that many movers “escaped enumeration; presumably both their previous and prospective dwellings were reported as vacant.” Bureau analysts concluded that if the



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