non-probability sampling The distinction between probability and non-probability sampling is fundamental to discussions of sampling. In the former case, each unit of a known population has a specifiable probability of being included in a sample, drawn according to simple random, stratified random, multi-stage cluster, or some other form of representative sampling technique that permits the use of inferential statistics. Non-probability samples, on the other hand, are generated by a variety of ad hoc techniques (such as snowballing), usually in those circumstances where no suitable sampling frame exists, or the research design does not actually require probability sampling. Studies of business élites offer an example of the former, since no convenient and exhaustive lists are generally available of (let us say) company directors holding multiple directorships in large business enterprises, whereas studies of (for example) members of a religious sect rarely require probability sampling: a selection of the membership (not necessarily statistically representative) is usually considered to be sufficient for most sociological purposes.

sampling, non-probability See NON-PROBABILITY SAMPLING.