CSP 2017 Online Program

The Nexus Between Data Science, Survey Design, and Statistical Practice (303292)

Keywords: Data science, predictive analytics, statistical practice, survey design

The field of data science has served to rapidly expand the knowledge base and decision-making ability through the combination of seemly disparate and diverse sources of information and content, which include survey and administrative data, social, financial and economic micro-data, and content from mobile devices, the internet and social media. Other attributes of data science include data visualization and rapid prototyping; social media analytics and social network analysis; predictive, mathematical and simulation modeling; use of Bayesian methods, machine learning; GIS and geospatial analytics and Big Data technologies. When appropriately harnessed, the resultant outputs have the capacity to serve as catalysts to yield new, actionable insights. Advances in data science also serve to facilitate the effective and efficient utilization of statistical models and procedures in concert with big data applications. Through these synergies, greater accuracy and efficiencies can realized in the application of predictive modeling techniques, the implementation of optimization analytics, and in the development of enhanced data resources that can provide the informatics to guide decisions and target interventions. In this presentation, attention is given to demonstrate the capacity of data science to inform the design of surveys, their operations and associated strategies to reduce survey errors and enhance data quality. Examples are provided as they apply to sample frame development, sample design specifications, operationalizing oversampling strategies and analytic file creation. In each of these settings, the gains in efficiency, accuracy and quality in survey procedures realized through specific applications of data science to ongoing survey efforts are highlighted.