A multi-tier framework for dynamic data collection, analysis, and visualization

Abstract

This thesis describes a framework for collecting, analyzing, and visualizing dynamic data, particularly data gathered through Web questionnaires. The framework addresses challenges such as promoting user participation, handling missing or invalid data, and streamlining the data interpretation process. Tools in the framework provide an intuitive way to build robust questionnaires on the Web and perform on-the-fly analysis and visualization of results. A novel 2.5-dimensional dynamic response-distribution visualization allows subjects to compare their results against others immediately after they have submitted their response, thereby encouraging active participation in ongoing research studies. Other modules offer the capability to quickly gain insight and discover patterns in user data. The framework has been implemented in a multi-tier architecture within an open-source, Java-based platform. It is incorporated into Risk Psychology Network, a research and educational project at MIT's Laboratory for Financial Engineering.