Recommendations and Cautions

Abstract

During the design of a data system, indirect estimators rarely, if ever, are considered for Federal statistical programs when resources to produce direct estimates of adequate precision are available. However, given an existing system, if direct estimation is judged to be inadequate for a domain not specified in the design, indirect estimation may, in some cases, prove to be a valuable alternative. There are a number of reasons that direct estimators are preferable to indirect ones and, if federal statistical agencies are to improve the usefulness of indirect estimates, a number of important issues, including those that follow, should receive additional attention. The following points are developed further in the sections of this chapter.KeywordsDirect EstimateNational Health Interview SurveyEmpirical EvaluationAuxiliary InformationIndirect EstimateThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.