n.a portion, within a body of records, selected for permanent retention through a process that ensures each file within the series has an equal chance of being included in the retained setLewinson 1957, 303A statistical sample, as has been said, is set up by elaborate technical and mathematical procedures designed to give specific information within strictly measurable limits of reliability.Boles 1981, 126A statistical sample begins also with an examination of the record group to determine the homogeneity of its documents. Rather than just a conclusion that the documents are more or less the same, a statistical sample requires that an attempt be made to specify homogeneity in a mathematical measure of variance. Then the exact degree of accuracy desired is specified, as well as the degree of certainty that this degree of accuracy will be obtained. Armed with these three numbers: variance, the specified degree of accuracy, and the level of confidence that the specified degree of accuracy will be achieved, algebraic formulas designed to yield specific sample size may be used to calculate how much of the material to retain.Lockwood 1990, 400In order to ensure that it successfully identified most of the files with research value, the task force also created computer profiles for each subject classification with varying retention periods de- pending on the informational value of the particular crime, took a statistical sample of all FBI records to preserve evidence of the “typical” file, and developed a list of exceptional cases to be saved.

Notes

True statistical sampling guarantees that any resulting sample will be representative of the whole. By determining the size of the original body of records (the population) and the level of homogeneity within it, an archivist can determine how big the resulting sample must be to be representative. Since random methodology is required to create a statistical sample, a statistical sample is also called a random sample. Occasionally, people categorize both systematic and random samples as subtypes of statistical samples, but systematic samples cannot be truly valid statistically.