Journal of Official Statistics, Vol.28, No.2, 2012. pp. 173–198
Sources of Error in Survey and Administrative Data: The Importance of Reporting Procedures
Jeffrey A. Groen
Abstract:With increased attention to administrative data for statistical purposes, analyses of the quality of administrative data and comparisons to survey data are greatly needed. This article presents a methodology for identifying sources of error in administrative and survey data and for identifying sources of differences between administrative and survey estimates. The first part of the methodology is a statistical decomposition of the difference between administrative and survey estimates. The second part investigates the causes of measurement error and the factors associated with differences between what the same respondents report in the survey and the administrative records. I illustrate this methodology using a case study of the monthly employment figures gathered from U.S. businesses. This analysis demonstrates that both administrative data and survey data may contain errors and that reporting procedures are likely to differ between the two types of data. The article also identifies practical ways to assess data quality.
Keywords:Measurement error, data collection, establishment survey, Current Employment Statistics (CES) survey, Quarterly Census of Employment and Wages (QCEW)
Copyright © Statistics Sweden 1996-2017. Open AccessISSN 0282-423XCreated and Maintained by OKS Group