Abstract
A principal purpose of work sampling is to determine the percentage of time a specified activity is occurring within a predetermined tolerance at a specified statistical risk. Traditional practices are based on a large number of observations taken at random times. In the analysis this sampling is considered a Bernoulli process and the normal (Gaussian) approximation to the binomial distribution is typically used. An alternative Bayesian procedure is suggested, developed, and demonstrated in this paper. Several advantages are available through this Bayesian approach in contrast to the traditional procedure including greater efficiency (especially with a small number of samples) and managerial flexibility. These advantages are illustrated through a numerical example.