Computation of Survey Weights

The basic theoretical foundation for weighting survey data is not controversial. Weights account for each respondent’s selection or inclusion probability, and as such, weights say how many people each person who responds to the survey represents in the population. However, in practice weighting is done ad hoc, with little transparency or consistency. This chapter discusses the limitations of ad hoc weighting, advocates greater transparency, accessibility, and replicability, and describes methodological developments to make the computation of survey weights easier for more people to do using transparent and replicable methods.