Foundations for Quality Management of Scientific Data Products

Abstract

The costs of making incorrect scientific inferences based on faulty data can be substantial and far-reaching: errors can be subtle, inappropriate conclusions can go unchallenged for years in the literature, and follow-on research may be critically jeopardized. Because most scientific research in the United States is federally funded, the propagation of errors through research studies imbues high costs to taxpayers over time; errors in scientific conclusions, and any technologies based on them, will require rework at some point in the future. Better scientific data quality means more accurate conclusions can be made more quickly, and benefits can be realized by society more readily. To improve scientific data quality, and provide continuous quality assessment and management, the nature of scientific data and the processes that produce it must be articulated.

The purpose of this research is to provide a conceptual foundation for the management of data quality as it applies to scientific data products, specifically those generated by the large-scale instrumentation and facilities that will populate the data centers of the future. Definitions for data product and data quality tailored to the context of scientific decision making are proposed, given two typical scenarios: 1) collecting observational data, and 2) performing archive-based research. Two relevant extensions to the total quality management (TQM) philosophy, total information quality management (TIQM), and total data quality management (TDQM) are then examined to determine if the management of scientific data quality differs from the management of other data or information. Recommendations for planning, assessment/assurance, control, and continuous improvement are proposed, focusing on designing quality into the production process rather than relying on mass inspection.

• data center
• data management
• data quality
• scientific data
• semiotics
• TDQM
• TIQM
• TQM

Additional information

Notes on contributors

Nicole M. Radziwill

Nicole Radziwill is the division head for software development at the National Radio Astronomy Observatory (NRAO) in Green Bank, W.Va. Prior to NRAO, her experience includes managing consulting engagements in customer relationship management and sales force automation for telecommunications clients of Nortel Networks, and working in scientific computation at the National Oceanic and Atmospheric Administration (NOAA) Forecast Systems Laboratory in Boulder, Colo. She has more than a decade of experience managing continuous improvement efforts in business and technology, specializing in software development and process improvements that result from information technology. Radziwill has a degree in meteorology, an MBA, and is currently pursuing a doctorate in technology management and quality systems. She is a frequent speaker at conferences in astronomy and information technology, and is recognized as an ASQ Certified Quality Manager. She can be reached by e-mail at [email protected].