The Use of and Selection Biases Associated with Nonstatistical Sampling in Auditing.

Abstract This study investigated the extent to which nonstatistical sampling procedures are used by practicing auditors and tested the most popular form of nonstatistical sampling for evidence of selection biases. Respondents to a survey of practicing auditors reported that nonstatistical procedures are used in approximately 85 percent of all audit sampling applications. Within the ,domain of nonstatistical sampling, approximately 90 percent of the applications utilize a form of nonstatistical sampling known as haphazard selection. In a laboratory experiment, audit samples chosen via haphazard selection were tested for the existence of selection biases. These samples were found to exhibit selection biases in favor of larger, more brightly colored, and more conveniently located population elements, as well as elements with fewer adjacent neighbors. In extreme cases, selection likelihoods for certain population elements were approximately eight times larger than likelihoods for other elements. Study results suggest that organizations using haphazard selection procedures arid standard-setting bodies in the accounting profession should reexamine the acceptability of haphazard selection and further develop guidelines for its use.