Search for: All records

Too much of a good thing can be harmful. Choice overload, a compelling paradox in consumer psychology, exemplifies this notion with the idea that offering more product options could impede rather than improve consumer satisfaction, even when consumers are free to ignore any available option. After attracting intense interest in the past decades from multiple disciplines, research on choice overload has produced voluminous yet paradoxical findings that are widely perceived as inconsistent even at the meta-analytic level. This paper launches an interdisciplinary inquiry to resolve the inconsistencies on both the conceptual and empirical fronts. Specifically, we identified a surprising butrobust pattern among the existing empirical evidence for the choiceoverload effect and demonstrated through mathematical analysis and extensive simulation studies that the pattern would only likely emerge from one specific type of latent mechanism underlying the moderated choiceoverload effect. The paper discusses the research and practical implications of our findings—namely, the broad promise of analytical meta-analysis (an emerging area for the use of data analytics) and machine learning to address the widely recognized inconsistencies in social and behavioral sciences, and the unique and salient role of the information systems community in developing this new era of meta-analysis. 

With the advent of computing technologies, researchers across social science fields are using increasingly complex methods to collect, process, and analyze data in pursuit of scientific evidence. Given the complexity of research methods used, it is important to ensure that the research findings produced by a study are robust instead of being affected significantly by uncertainties associated with the design or implementation of the study. The field of metascience—the use of scientific methodology to study science itself—has examined various aspects of this robustness requirement for research that uses conventional designed studies (e.g., surveys, laboratory experiments) to collect data. Largely missing, however, are efforts to examine the robustness of empirical research using “organic data,” namely, data that are generated without any explicit research design elements and are continuously documented by digital devices (e.g., video captured by ubiquitous sensing devices; content and social interactions extracted from social networking sites, Twitter feeds, and click streams). Given the growing popularity of using organic data in management research, it is essential to understand issues concerning the usage and processing of organic data that may affect the robustness of research findings. This commentary first provides an overview of commonly present issues that threaten the validity of inferences drawn from empirical studies using organic data. This is followed by a discussion on some key considerations and suggestions for making organic data a robust and integral part of future research endeavors in management.