The design of samples to monitor the quality of the data being collected (data quality monitoring) and to evaluate properties of measurement errors for the survey (measurement error evaluation) in the context of large-scale longitudinal surveys is discussed. Longitudinal surveys provide historical information that allows for more complex sample designs than simple ad hoc approaches used in one-time surveys. The properties and feasibility of several classes of probability sample designs for data quality monitoring is investigated. The problem of assessing the properties of measurement error is addressed with emphasis on designing a subsample to estimate the measurement error contribution to the variance of a sample estimator. The investigation is motivated by potential applications to the United States Department of Agriculture\u27s National Resources Inventory.
展开▼
