Survey data concerning teacher education program graduates were used to demonstrate the advantages of a stratified random sampling approach, with followup, relative to a one-shot mailing to an entire population. Sampling issues involved in such an approach are addressed, particularly with regard to quantifying the effects of nonresponse on the results. Comparisons were made between the two procedures, using the precision of estimates of population characteristics as the criterion. Much narrower confidence bounds were associated with the stratified random sampling approach, even though a smaller number of questionnaires had been returned. It is concluded that a redirection of energies from reaching the entire population to increasing the rate of response in a subset of that population has a significant payoff in terms of the precision of the estimates that can be made from the data. Keeping in mind that this advantage will be even greater in cases in which response variation differs across strata, the arguments are strong for adopting a stratified random sampling approach in situations such as the followup of teacher education program graduates. (Author/SW)