Podcast: Drawing on big data to better understand school improvement strategies

August 29, 2019

While education researchers possess good tools for determining whether or not a particular policy or program worked, unlocking the theories of change used by leaders and policymakers to support successful implementation in schools and districts has proven a much greater challenge to do at scale. 

Drawing on novel text data in thousands of school improvement planning and implementation reports from Washington state, new research led by University of Washington College of Education Associate Professor Min Sun offers practitioners greater insight into the mechanisms by which complex interventions achieve their effects.

In a new podcast, Sun discusses her recently published paper in the journal Educational Evaluation and Policy Analysis, "Using a Text-as-Data Approach to Understand Reform Processes: A Deep Exploration of School Improvement Strategies."

“Most importantly, this study serves as proof of concept that detailed text data, particularly when linked to administrative data, offer a promising opportunity for us to explore those key processes of educational change,” said Sun, director of the UW's Education Policy Analytics Lab

Sun and her co-authors used computer-assisted techniques to extract measures of school improvement processes from reports submitted by approximately 300 schools across Washington, identifying 15 coherent reform strategies that varied greatly across schools and over time. Their analysis illustrated the predictive relations of these reform strategy measures by showing that several measures were significantly associated with the reductions in student chronic absenteeism and the improvements in student achievement.

While schools, districts and state education agencies have for decades generated a large amount of text data to document the policy implementation process, and will continue to generate more under the Every Student Succeeds Act, Sun said researchers haven’t been able to make full use of that data due primarily to the time-consuming process of hand coding it. 

“Text-as-data is a method that offers a new approach to the issue and provides a much more efficient way of analyzing this large volume of data,” Sun said. “The information generated can really help us develop more evidence-based policies as well as to support the enacting of more positive changes in schools.”

Subscribe to the UW College of Education on SoundCloud to listen to more interviews with researchers, practitioners, community leaders and policymakers who are working to transform inequitable systems of education and make learning come alive for all students.


Min Sun, Associate Professor of Education
206-221-1625, misun@uw.edu

Dustin Wunderlich, Director of Marketing and Communications
206-543-1035, dwunder@uw.edu