Research Note—Generating Shareable Statistical Databases for Business Value: Multiple Imputation with Multimodal Perturbation

Abstract

Business organizations are generating growing volumes of data about their employees, customers, and suppliers. Much of these data cannot be exploited for business value due to privacy and confidentiality concerns. National statistical agencies share sensitive data collected from individuals and businesses by modifying the data so individuals and firms cannot be identified but statistical utility is preserved. We build on this literature to develop a hybrid approach to data masking for business organizations. We demonstrate the validity of the hybrid approach, which we call multiple imputation with multimodal perturbation (MIMP), using Monte Carlo simulation and illustrate its application in a specific business context. Results of our analysis open new areas of research for information systems scholarship and new potential revenue sources for business organizations.