Stratified sampling is a method in which a population is divided into separate groups, called strata, based upon specific characteristics such as gender or age. This type of sampling method ensures that all proportions of a population are represented within the final sample. It is similar in nature to cluster sampling, but differs in that the divisions are based on characteristics and not all members of each stratum are sampled.
Stratified sampling is a method which begins by structuring the population before taking the sample. This type of technique is very much applicable to human populations in which individuals reside in various cities, countries, types of households and so on. Furthermore, on an even smaller scale, individuals vary in terms of their beliefs, opinions, preferences and other basic characteristics. Thus, deciding how to separate the population being evaluated is a critical first step in this process.
Once all strata have been determined, this sampling method essentially follows the principles of simplified random sampling. All subjects in each stratum are numbered and then these numbers are randomly selected. Initially, a total sample size should be predetermined and then the number of individuals selected per stratum will be dependent upon the sample size desired. Furthermore, to create a non-biased sample, the same number of participants should be selected per stratum.
When performing statistical analyses from stratified sampling, the tests used need to recognize that the sample is comprised of pooled results. For very diverse populations, stratified sampling would potentially be very effective because it recognizes key attributes from various sections of the population. However, in fairly homogeneous populations, stratified sampling may not be as powerful.
Image Credit: ReStore. (July 29, 2011). Population and Sampling. Retrieved from: http://www.restore.ac.uk/srme/www/fac/soc/wie/research-new/srme/modules/mod1/2/