Chapter 9 Measurement issues

Measurement is the contact of reason with nature.

— Henry Margenau

This chapter contains three case studies using real data. The common theme is that all have “problems with the columns.” Section 9.1 illustrates a number of useful steps to take when confronted with a dataset that has an overwhelming number of variables. Section 9.2 continues with the same data, and shows how a simple sensitivity analysis can be done. Section 9.3 illustrates how multiple imputation can be used to estimate overweight prevalence from self-reported data. Section 9.4 shows a way to do a sensible analysis on data that are incomparable.