In the Queue
Monday, September 26, 2016
We’ve all been there — stuck in line at a bank, retail store, service center or event. Although numerous studies suggest that single queue lines are faster, new research coauthored by Krannert operations management and economics professor Yaroslav Rosokha finds consequences on server behavior that were previously unexplored, specifically that servers work more slowly in a single queue than when they each have their own queue.
In "Humans are not machines: The behavioral impact of queueing design on service time," forthcoming in Management Science, Rosokha and his coauthors, Julie Niederhoff of Syracuse University and Masha Shunko of the University of Washington, examine human behavior under different queue structure designs.
Their work began with a pilot study conducted during volleyball games at Purdue that explored the behavioral aspects of queue design and visibility. In an effort to measure the relative pace of transactions, the researchers shifted the ticket lines between a single pooled queue with multiple servers and multiple parallel queues with a dedicated server for each.
“The problem was that we could not separate the effects of single versus parallel queues because the single queue was long and essentially invisible to the servers, while parallel queues were more visible,” Rosokha explains. “Two things were changing at once.”
To isolate those effects, the researchers followed up with a controlled experiment in a computer simulated environment to determine the performance of workers by manipulating queue structure and visibility separately.
“Because serving customers in a single queue line is viewed as a shared task, some servers feel less accountability,” Rosokha says. “They may work more slowly and expect others to pick up the pace.”
Another dimension the researchers consider is the saliency of feedback due to poor visibility, which can lead customers to literally being “out of sight and out of mind” to the servers at the other end of the line. The study found that servers who had better visibility of either kind of queue worked more quickly.
“Although many complex variables affect service time and ultimately customer satisfaction,” he says, “managers can improve the speed of service in both single and parallel queues simply by addressing these system design issues.”
An abstract and downloadable PDF of "Humans are not machines: The behavioral impact of queueing design on service time" is available at http://ssrn.com/abstract=2479342
Contributors: Julie Niederhoff, Assistant Professor of Supply Chain Management, Syracuse University; Yaroslav Rosokha, Assistant Professor of Operations Management and Economics, Purdue University Krannert School of Management; Masha Shunko, Assistant Professor of Operations Management, University of Washington Foster School of Business.