This chapter presents methods for using SAS® software to analyze data from two-phase samples and capture-recapture studies. Two-phase sampling, sometimes called double sampling, is useful when the key variables of interest are relatively expensive to measure but related variables can be measured fairly easily—for example, when information on an inexpensive, but possibly inaccurate, screening test from a large initial sample is used to define strata for a subsample of persons to be given a more comprehensive diagnostic test. This chapter uses a macro presented in Appendix B to analyze data from simple two-phase sampling designs with jackknife variance estimation.
The size of a population can be estimated by comparing multiple independent samples randomly selected from it. In the simplest form of capture-recapture estimation, two independent simple random samples are taken and the number of population members found in both samples is used to estimate the population size. Code and output for examples in Chapter 13 of Sampling: Design and Analysis, Third Edition illustrate estimating population sizes and associated confidence intervals from two-sample and multiple-sample capture-recapture studies.