This study investigates the influence of monetary incentive and type of instructions given to operators on the spread of actual data points around two possible learning models; the linear and the power models. The investigation is based on a laboratory study in which subjects were asked to perform a specific task in different conditions of learning: with or without incentive, and with or without exact specifications according to which the task had to be performed. The value of the spread for these four cases was measured by the coefficient of variation and other similar indicators. Analysis of the spread points to the following major conclusions: ( a) with the introduction of an incentive system the spread reduces by approximately 20%; (6) when an exact work method is specified, the spread reduces; (c) the coefficient of variation reached an average value as high as 0-28. Finally, a numerical example shows how the coefficient of variation can be used to improve the estimate of batch production time.