Clear delineation of the universe and basic survey units is critical in surveys and censuses. Clear concepts and definitions are needed by survey administrators developing and updating frames and by respondents applying the definitions to identify the appropriate reporting and population units. Developing standardized, understandable definitions of units is especially problematic in surveys of facilities, businesses, and other establishments, due to wide variations in size, organizational complexity, functions, and population characteristics (Cox and Chinappa 1995). This paper presents the results of research to improve the definitions and methods for delineating the basic target statistical units in a census of residential juvenile facilities. The concepts defining the target statistical unit in this census–"facility" and "facility type"–are discussed and the questions eliciting this information on the mailout form are shown. Six factors affecting how respondents interpret these concepts and apply them to their own organizations to determine their reporting units are identified. The interaction of these factors is illustrated in general diagrams of facility layouts and arrangements. The implications of these interactions for the fit between the target statistical units intended in this census and the units respondents choose to report on–agency, facility, or program–are discussed. The implications of these interactions for frame coverage, data quality, data collection methodology, respondent burden and questionnaire revisions are identified. Recommendations for revisions to census methods and questions are proposed. The fit between a new definition of "facility" offered here and the concepts of "group quarters" and "establishment" used in other non-household surveys is assessed. The applicability of the findings to other group quarter and business surveys is addressed.
