43; the other 843 people were dormitory residents at Concordia University in the city of Mequon—in the same county, but tallied some 18 miles away from the proper location (Cole, 2001a,b).

Poor Levels of Full, Self-Report Data A higher-than-expected proportion of group quarters enumerations were obtained from administrative or facility records; the means by which group quarters questionnaires were completed in 2000 are described in Table 7-1. Of the 83 percent of enumerations for which enumerators indicated the source of data, 59 percent were filled out from administrative data, 30 percent were filled out by the resident, and 12 percent were filled out by an enumerator interviewing the resident. Types of group quarters with high percentages of enumerations obtained from administrative data included nursing homes, hospitals, group homes, and prisons. These types of group quarters had especially high rates of missing data for long-form items.

Lack of Coverage Measurement and Use of Reported “Usual Home Elsewhere” Addresses The Census Bureau opted not to include group quarters residents in the 2000 Accuracy and Coverage Evaluation program, principally because of low match rates in the 1990 Post-Enumeration Survey for group quarters residents in comparison with the rest of the household population. That is, the Bureau found that it was difficult to match group quarters residents interviewed in a follow-up survey (conducted a few months after the census) to their census records. This low match rate was attributed to high short-term mobility in the group quarters population (e.g., college students returning from their studies and shelter residents moving from one locale to another). Low match rates complicate the creation of population-adjusted census estimates using dual-systems estimation (the core methodology of the Accuracy and Coverage Evaluation Program), and so group quarters were dropped from consideration for the coverage evaluation program in 2000.

Because of the difficulty in matching follow-up survey records with samples from the census, dual-systems estimation is most likely not the best means of assessing group quarters coverage. However, in the absence of estimates of omissions and erroneous enumerations in group quarters settings, the 2000 census plan did not include any systematic or comprehensive review of the coverage in group quarters, whether through rigorous comparison with facility or administrative records or through structured observational studies.

Design choices in the Bureau’s Non-ID Process—the procedure by which the Bureau processed all census forms without a MAF identification number, including “Be Counted” forms and all group quarters forms indicating a “usual home elsewhere”—also led to a major lost opportunity for understanding residence reporting problems. As shown in Box 3-1, the Individual