Product Recall Decisions in Medical Device Supply Chains: A Big Data Analytic Approach to Evaluating Judgment Bias

Abstract

This study investigates judgment bias (under‐reaction or over‐reaction) in product recall decisions by firms when they respond to adverse event reports generated by users of their products. We develop an integrative theoretical framework for identifying the sources of judgment bias in product recall decisions. We analyze user‐generated reports (big and unstructured data) on adverse events related to medical devices, using a combination of econometric and predictive analytic methods. We find that (i) noisy signals in user feedback, that is, high noise‐to‐signal ratio, are associated with under‐reaction likelihood; and (ii) user feedback related to adverse events characterized by high severity is associated with high over‐reaction likelihood. We also identify conditions related to the situated context of managers that are associated with under‐reaction or over‐reaction likelihood. The findings of this study are consequential for firms and government regulatory agencies, in that they shed light on the sources of judgment bias in recall decisions, thereby ensuring that such decisions are made correctly and in a timely manner. Our findings also contribute toward improving the post‐launch market surveillance of products (e.g., medical devices) by making it more evidence‐based and predictive.