In statistics, survey sampling is random selection of a sample from a finite population. It is an important part of planning statistical research and design of experiments. Sophisticated sampling techniques that are both economical and scientifically reliable have been developed.
An entire industry of public opinion polling as well as the technical activities of the U.S. Bureau of the Census depends on these techniques.
The most elementary methodology is called simple random sampling. Most of the theory of statistics assumes this kind of sampling unless otherwise noted. It assures that every possible subset of the population which has the desired sample size is given the same probability of selection.
The possibility of very expensive or very atypical samples has lead to a variety of modifications such as stratified sampling, cluster sampling, and multistage sampling. The most experienced center in these techniques outside the Census Bureau is the University of Michigan Survey Research Center.
In public opinion polling by private companies or organizations unable to require response, the resulting sample is self-selected rather than random. Volunteering for the sample may be determined by characteristics such as submissiveness or availability. The samples in such surveys are therefore non-probability samples of the population, and the validity of estimates of parameters based on them is unknown. They are, however, unquestionably random samples of that sizeable subgroup of the population which volunteers for opinion surveys.