ABSTRACT
Medical decisions are almost invariably based on samples. Samples of blood, urine, sputum, stool, and biopsies are everyday occurrences. Sampling is the only feasible method in this situation since the complete material of these biological substances for anybody cannot be studied.
The thrust in this chapter is on samples from individuals. Whether the research is descriptive or analytical, it is conducted on a sample of subjects. Sample is the statistical term for the group of subjects included in the study, which almost invariably would be a fraction of the target population.
There are two dimensions of adequacy of a sample. First, it should represent the full spectrum of subjects in the target population. Various methods of sampling are adopted in different situations to meet this objective. These methods are discussed in the first section for descriptive studies and in the second section for analytical studies. Second, the sample size must be reasonably large to give reliable results, without being excessively large. Sample size is discussed in the last section. In between there is a discussion on nonsampling errors in contrast to sampling errors. Controlling for nonsampling errors often assumes more importance because these have the potential to vitiate the results beyond redemption, whereas sampling errors can be handled by choosing a representative sample of adequate size.
