In some areas the measurement issues are complex, for example in health and education. Ensure you familiarise yourself with the particular measurement properties of the variables you intend to use. For example, many of the educational assessment studies use an item response model which means that any one student only answers a subset of the total test. This is to ensure a broad range of topics can be covered. The design of these studies is to estimate the variable of interest at population rather than individual level although the files contain estimates for each respondent. The attainment variable is then modelled to reflect how that student would have done if they had answered all the questions in the test based on the how similar students performed on the questions not answered. Many of them also use a Plausible Values (PV) methodology so that different estimates of attainment variables multiple values for an individual are provided. The spread of these plausible values can be quite wide. Read the recommendations of the survey team for which method to use for selecting plausible values for your analysis.

In health there are complex measurement issues in physical and biochemical measurement. For example, the DHS takes blood spots for testing for anaemia and for HIV. Again, familiarise yourself with the methodological and process issues involved.