The Origin of Information: Sampling

Abstract

Sampling from medical records is a classical technique in medical and epidemiologic research. Let us look at a fictitious example from the pre-computer age. Suppose Dr. Q. wants to find the frequency of a risk factor A among all patients who were treated in a given ward of his hospital during the period 1950–1959. There had been N = 9850 such patients, and for every one of them a medical record in the form of a file exists where the presence or absence of A is recorded. Dr. Q. could look through the whole (dust covered) set of files and count the occurrence of A in it. Suppose that he finds it 271 times; this would be the “frequency” he is looking for. The so-called “relative frequency”, or “proportion”, of A is more instructive; it is equal to

Sampling is the art of obtaining information on a target population by studying only a suitably chosen subpopulation. The latter is called a sample or a study population . The information includes a description of the error committed by using only a sample. Almost every epidemiologic study involves sampling, but sampling may also be performed regularly within a given information system.