This interim report makes recommendations and suggests further research focused on the statistical methods used to produce estimates from the American Community Survey (ACS) that include group quarters residents. During the course of the next year, the panel will continue to examine these topics in the broader context of data user needs and the role and mission of the ACS. The overall question to consider is whether the current and potential uses of the ACS data justify the costs associated with including group quarters (GQs) in the sample and whether data user needs could in fact be better met with alternative approaches to producing the estimates of interest.
The panel will consider the implications of dropping some types of group quarters from the sample, as well as alternatives to the current approach of producing estimates that may be necessary to users. The alternatives include a possible redesign of the data collection approach to combine the household and GQ data collection in a way that potentially increases efficiency and reduces some of the challenges related to maintaining two separate sampling frames. Better integration of the household and GQ address lists into one unified Master Address File may be an important aspect of this. Other possibilities involve the use of alternative sources for the GQ data and statistical techniques, such as small-area model-based estimates to replace unreliable direct estimates. The panel will also continue to examine the sampling approach, including the sample allocation and subsampling rates.
The discussion so far has focused on the problem of coverage error. Another area of interest is the quality and usability of the data obtained from group quarters when an interview is completed. The ACS survey instrument was developed with the household population in mind, and even though collecting some of the data on the questionnaire from GQ residents is necessary, many of the questions on the present questionnaire are not applicable to all GQ populations. Customizing the instrument by GQ type may not be cost-effective, but a short-form questionnaire that includes only the items that are absolutely necessary and applicable to group quarters could be considered. This is another topic of interest for the final report.