Quantifying heterogeneity in human behavior: An empirical analysis of forklift operations through multilevel modeling

Human operators will remain to play an essential role in picker-to-parts order picking systems despite increasing digitalization and automation of warehouse processes. While manual order picking is a laborious and cost-intensive task in warehousing, it is extensively examined in the logistics and supply chain management literature. However, the operational and individual performance of forklift operators in warehouse picking operations has received little attention yet. We aspire to close this gap by drawing on sociotechnical systems theory and formulate a multilevel approach to evaluate heterogeneity in human behavior towards differences in picking performance. We use batch execution times as the dependent variable and source level of operation, target level of operation, filling level of the palette, the necessity to correct replenishment quantities, as well as travel distance as independent variables on the first level. For the second level, we utilize forklift operators to quantify whether heterogeneity in human behavior is impacting the performance of forklift operators. We find that 15.1% of the variance among batch execution times results from heterogeneity in human behavior. In a further simulation, we show that this method can be used to assess the performance of order pickers through a multi-dimensional parametric production frontier analysis. Our findings are highly relevant for logistics management when aspiring to forecast the necessary capacity of forklift operators in a warehouse or building bonus systems that are based on more than the existing two-dimensional measures such as process time per operation.