Simple analytic procedures for rapid microcomputer-assisted cluster surveys in developing countries.

Abstract

Surveys are often deemed necessary in developing countries when routine sources of data are not considered adequate to answer important policy-related questions. Although field work often goes smoothly, many surveys become bogged down in the analysis stage. With the availability of microcomputers and contemporary software, investigators in developing countries can use rapid survey methodology (RSM) to process, analyze, and report survey findings more quickly than ever before. Presented in this paper are three simple analytic procedures for planning and doing two-stage, rapid cluster surveys. All were successfully used in three rapid surveys in rural regions of Burma and Thailand. By use of a spreadsheet and graphics software package, the three procedures (a) derive the first-stage selection of 30 cluster sites with probability proportionate to size, (b) calculate variance estimates and confidence limits for the parameters of interest and graphically present the findings as 90, 95, and 99 percent confidence intervals, and (c) estimate the necessary sample size for planning two-stage, rapid cluster surveys. The procedures can be used both in the field and in teaching workshops or courses on survey methods. Examples are given from three rapid surveys conducted in Hlegu Township, Burma, and Sisaket Province, Thailand. In both countries, local health professionals were first taught the methods in a 1-week workshop before they used the procedure for conducting the rapid computer-assisted surveys.