ABSTRACT

D. D. Stockford and Page used two-phase sampling to estimate the percentage of Vietnam-era veterans in United States Veterans Administration (VA) hospitals who actually served in Vietnam. The 1982 VA Annual Patient Census included a random sample of 20% of the patients in VA hospitals. Two-phase sampling is often used in forestry surveys. A general framework for two-phase sampling is given in C. E. Sarndal and B. Swensson and J. C. Legg and W. A. Fuller. Two-phase sample designs require all the considerations of one-phase samples, plus additional decision of how many resources to devote to each phase. W. E. Deming discusses issues to be considered when deciding whether to use a two-phase sample. J. Armstrong et al. give an example of two-phase sampling for tax records. Two-phase sampling can save time and money if the auxiliary information is relatively inexpensive to obtain and if having that auxiliary information can increase the precision of the estimates for quantities of interest.