ABSTRACT
This chapter discusses further some of the issues surrounding measurement error in panel surveys which Rose discussed in Chapter 1 and to which Kalton and Citro in Chapter 2 and Duncan in Chapter 3 also referred. Non-response and measurement error may be viewed as the two complementary sources of non-sampling error in panel surveys. Survey estimates are functions of the values of variables for the set of responding units. These estimates may differ from the population values they estimate either because the set of units is ‘unrepresentative’, resulting in sampling error or non-response error, or because the values of the variables are erroneously recorded, resulting in measurement error. The problem of non-response has been addressed in the previous chapters. In this chapter, the problem of measurement error is considered.
