The results were revealing. Although direct labor cost appears to be a driver of overhead cost, it really doesn't explain a lot of the variation. I then searched for other driversparticularly non-unit driversthat might offer more insight into overhead cost behavior. Every time a batch is produced, material movement occurs, regardless of the size of the batch. The number of moves seemed like a more logical driver. I was able to gather only 10 months of data for this. (Our information system doesn't provide the number of moves, so I had to build the data set by interviewing production personnel.) This information is provided next: Materials-Handling Cost Number of Moves $80,000 1,500 60,000 1,000 70,000 1,250 72,000 1,300 65,000 1,100 85,000 1,700 67,000 1,200 73,500 1,350 83,000 1,400 84,000 1,700 The regression results were impressive. There is no question in my mind that the number of moves is a good driver of materials-handling costs. Using the number of moves to assign materials-handling costs to products would likely be better than the cost assignment using direct labor cost. Furthermore, since small batches use the same number of moves as large batches, we have some evidence that we may be overcosting our