In recent years, a huge amount of excitement has surrounded the rise of innovations in data-driven decision making (DDDM). The promise of DDDM rests on ready availability of high-quality data resources that can be fed into systems and provide decision makers (people or algorithms) enhanced decision-making capacity. DDDM systems require pre-packaged data of high integrity — i.e. data that displays completeness, accuracy, consistency, and validity vis a vis the needs at hand. However, the institutional, organizational, and technological ecology in which data are used is constantly changing. In this paper we ask, under what conditions do data resources lose their integrity and how do organizations maintain resources of sufficient integrity over time? Emerging from a multi-sited ethnographic study of multiple healthcare organizations, we find that organizations face multiple conditions that usher in breakdowns in data integrity and organizations engage in a complex process of “data crafting” in order to anticipate and manage the breakdown of data integrity. Taking a critical IS perspective and applying a practice theoretic lens, we show that data resources do not have pre-existing characteristics that render them low or high quality. Rather, data quality is always relational to a local configuration of DDDM tools and processes, institutional mandates, and an organization’s goals. Thus, becoming a DDDM organization requires new forms of work. Organizations need to develop and maintain robust data crafting practices in order to begin to realize the promise of DDDM.
Email us at firstname.lastname@example.org for information on attending this seminar.