Employee ‘work memory’ affects rotation scheduling

Many service and manufacturing industry employers believe that cross-training their employees can help cover during turnover and absenteeism, while at the same time, increase workers’ level of interest in their jobs.

But what seems like a win-win situation may not always be the case, says David Nembhard, an assistant professor of industrial engineering at UW–Madison.

Nembhard studies workplace learning and forgetting behaviors, which he says grow more complex as employees take on additional tasks. “If you’re trying to cross-train at too many things, by the time employees rotate back to the first job after going through a half a dozen jobs, they’re not where they left off anymore. They never really reached their highest sustainable level,” he says.

Nembhard is studying ways of assigning workers to jobs in work systems that employ job rotation and in which learning and forgetting are factors. That scenario frequently occurs in companies such as computer firms or auto manufacturers that regularly develop new technologies and manufacture new products.

“People are constantly having to learn new skills,” he explains. They’re also asked to learn and do more than one job simultaneously.

Nembhard says employers need to understand how people learn and forget in those types of environments, in one sense just to estimate productivity for planning and decision-making purposes.

In a recent study at an auto electronics plant, he collected worker information at an inspection station for car radios, a product that changes annually.

“On that line, there were six models of radios, and the workers had to inspect each radio,” says Nembhard. He gathered worker data through existing data-acquisition systems, which enable the plant to track the computer-controlled part of the inspection and find individual radios by bar code. “Since the bar-code system was already in place, we could capture a wealth of detailed data with minimal additional expense,” he says.

That empirical research has generated some surprising facts about how workers in such situations learn, forget and perform on the job. Among his findings, Nembhard discovered that people who learn more slowly reach a higher performance level in the end, whereas faster learners in the same job often level off at a lower level-but there are advantages to having each type of worker. And he observed that people who have some experience with a job will learn a complex job more rapidly than an inexperienced worker trying to learn the same difficult job.

In the case of his research, however, Nembhard doesn’t single out individuals, but rather studies how groups of workers behave. His data reflect a range of their responses to a situation-and another surprising fact: If employers could trade their variable workforce for a troop of “average” workers, they’d be behind in the long run.

“You’d do worse with a work force made up of all identical average workers than having workers with diverse learning skills,” he says. “They’re actually more productive than having only average workers.”

Now Nembhard’s challenge is to develop ways that help employers make good use of his findings. He has created optimization models to help managers decide how much cross training is best in different scenarios. Once he’s fine-tuned the models, Nembhard’s next step is to conduct computer simulations.

“Simulation throws in all the other pieces and uncertainties that help us predict how our models will perform in the real world,” he says. “And then eventually, we’ll want to validate some of these things in real settings.”

His goal is to develop “rules of thumb” that many companies with similar job-scheduling situations can apply. “If we can get most of the way there by identifying the ‘biggest-bang-for-the-buck’ elements of what an optimal assignment schedule tells us, that’s in many ways just as valuable as an assignment policy that may be optimal but is very difficult in an operational sense to put in place,” explains Nembhard.

And, he says, the outcome will be higher quality and increased productivity: “In some cases, up to 30 percent.”