Two-stage designs: how is the second stage performed?

• A. Stage one collects exposure and outcome; stage two collects detailed covariates on a sample
• B. Stage one collects exposure and outcome; stage two collects no covariates
• C. Stage one collects covariates; stage two collects exposures
• D. Stage one uses randomization; stage two observes

Answer: (A)

Two-stage designs balance information gain with cost by collecting essential data first and then gathering more detailed information on a subset. In the first stage, you capture basic information such as exposure and outcome for the entire sample or a large portion. From that pool, you select a subset to obtain more detailed covariate data in the second stage. This setup matches the described approach: stage one collects exposure and outcome, and stage two collects detailed covariates on a sample. The idea is to save resources by limiting expensive covariate collection to a smaller group, while still being able to adjust for confounding with those detailed variables. The other patterns don’t fit this design: collecting no covariates in the second stage removes the purpose of the second phase; swapping covariates and exposures changes the data structure from a sequential two-stage sampling plan; and introducing randomization at the first stage describes a randomized trial rather than a two-stage observational sampling design.