Sampling From Samples: a Comparison of Strategies in Longitudinal Research

In this study, estimates of population values obtained using three sampling techniques (simple random sampling, stratified random sampling, and cluster sampling) and three size samples for each technique (10%, 25% and 50%) were compared with known "popu lation" values. The objective of the study was to determine if second stage sampling was justifiable, and if so, which sampling strategy and what size sample yielded adequate estimates of the known "population" values. The data were the responses to a questionnaire administered to 1481 teachers as part of a long term societal experi ment dealing with the effects of standardized testing upon a society. The results supported the feasibility of sampling from samples and showed that a 25% simple random sample provided quite good esti mates of the "population" values.