This paper addresses issues concerning the use of records to evaluate the quality of survey estimates and describes a specific application to the Survey of Income and Program Participation (SIPP) in the United States. Matching administrative records to survey observations on a case-by-case basis, which we call a "record check," provides useful information to survey users and designers. A record check enables the analyst to make a full range of measurement error parameter estimates for evaluation purposes. These estimates, in turn, facilitate two basic kinds of activities: 

(1) adjusting subject-matter estimates such as means, proportions, correlation coefficients, and mu1tivariate regression coefficients to correct for the measurement errors; and 

(2) deriving more efficient survey designs that directly address, for example, the trade-offs between measurement qua1ity and costs.