Improving Productivity Estimates by Work Sampling

Abstract

Theoretical aspects are presented to evaluate the adequacy of work sampling as a surrogate productivity measure. A 10‐week study is described in which work sampling data were gathered simultaneously with earned value information for a 10‐man, small‐bore pipefitter crew. Two forms of work sampling data were gathered that differed principally in the definition of direct work. The direct work percentages are statistically correlated to the ratio of earned to actual man‐hours, and it is shown that the Pearson product‐moment correlation coefficient could be improved by as much as 86% if the definition of direct work is restricted. The probability of a Type I statistical error is also greatly reduced. Data summarized according to 5‐ or 7‐day moving averages is found to be more reliable than daily or weekly averages. It is shown how a prediction equation could be developed. The slope of this equation suggests that rather significant changes in the earned to actual man‐hour ratio are reflected by rather small changes in direct work.