Building an inter-disciplinary research community to protype computationally-intensive analysis of large scale educational datasets

Cuny Graduate School University Center, New York NY

Investigators

Abstract

This proposal develops a community of researchers that crosses domains of education, social science, statistics and computer science to expand the research agenda that utilizes large scale, longitudinal educational databases. This research collaborative is engaging in the design and implementation of computationally-intense data analytic methodologies to address limitations experienced by researchers who use these techniques to analyze multiple types of large scale data in education including data from a national transcript study, local education agency longitudinal student data and an administrative dataset collected about students across a large university system. The researchers are bringing together scholars to identify a research agenda for the effective engagement of such methodologies and their use in addressing educational practice and policy questions. The researchers are developing and implementing a seminar on data-intense research methodologies that includes practical analysis of three different educational databases, conducting iterative bi-weekly workshops for researchers, and developing, implementing and documenting a graduate level course for new researchers. Researchers from the City University of New York Graduate Center are collaborating with the Inter-University Doctoral Consortium that includes eight additional institutions in the region. Scholars from across the country are recruited from across the country to participate in a virtual community. The project is operating through three distinct mechanisms. The first is the faculty seminar that brings together researchers from different domains to study the challenges and strategies involved in applying computationally-intensive techniques to the analysis of large longitudinal educational datasets, repeated across the two years of the project. The second is a biweekly workshop in which a working group of researchers from computer science, applied statistics and education design and pilot approaches to data analysis from specific datasets. The third mechanism is a semester-long course for doctoral students that connects them with the researchers who are developing and testing new methodologies for data analysis. The growing number of large scale longitudinal datasets in education present multiple analytic challenges. The first is that many techniques that have been developed for such analyses encounter problems when having to address the complexity of many educational datasets. The second is that the understanding of the fit between different computationally-intensive statistical methodology and the types of questions that need to be answered through these methodologies is not always straightforward. The third challenge is the growing need for researchers who have the background and training to use these multiple analytical techniques.