Mismanaging Diagnostic Accuracy Under Congestion

Diagnostic processes are difficult to manage because they require the decision maker (DM) to dynamically balance the benefit of acquiring more diagnostic information against the cost of doing so. When additional and unattended diagnostic tasks build up over time, making this tradeoff becomes especially challenging. In their study “Mismanaging Diagnostic Accuracy Under Congestion,” Kremer and de Véricourt uncover different biases to which DMs are subject when making diagnostic decisions while unattended diagnostic tasks accumulate over time. The authors find that, in their experiments, DMs are overall insufficiently sensitive to congestion. As a result, DMs acquire too little information at low congestion levels, but too much at high levels, compared with an optimal normative benchmark. This in fact increases both the diagnostic errors and congestion levels in the system. The authors disentangle the underlying mechanisms for these effects and suggests different approaches to debias the DMs.