ABSTRACT
This chapter discusses methodology to assess and to evaluate the impact of inaccurate parameters on the performance of designs. The robustness of ASDs strongly depends on the estimation strategy for survey design parameters. The chapter also discuses the strategies to evaluate the sensitivity of designs to inaccuracy in survey design parameter estimators. Survey design parameters are group response propensities, potentially decomposed to group contact and group participation propensities, and costs per sample unit in each group. The chapter explains Dutch LFS to demonstrate how robustness can be analyzed. In a Bayesian analysis, the regression coefficients are assigned prior distributions that are derived from historic data and/or expert knowledge. In 2015, a 3-year international network, Bayesian adaptive survey design network (BADEN) started to explore the potential of Bayesian analyses in ASD. The Bayesian analysis may be extended with a model for the main survey variables, especially when a survey has a relatively small set of topics and key statistics.
