The AIR Professional File
Fall 2016, Article 139
Mining Text Data: Making Sense of What Students Tell Us
John Zilvinskis, Greg V. Michalski
Text mining presents an efficient means to access the comprehensive amount of data found in written records by converting words into numbers and using algorithms to detect relevant patterns. This article presents the fundamentals of text mining, including an overview of key concepts, prevalent methodologies in this work, and popular software packages. The utility of text mining is demonstrated through description of two promising practices and presentation of two detailed examples. The two promising practices are (1) using text analytics to understand and minimize course withdrawals, and (2) assessing student understanding and depth of learning in science, technology, engineering and mathematics (STEM) (physics). The two detailed examples are (1) refining survey items on the National Survey of Student Engagement (NSSE), and (2) using text to create a learning analytics system at a community college (City University of New York [CUNY]: the Stella and Charles Guttman Community College, or CUNY Guttman). Results of this study include identification of additional item choices for the survey and discovery of a relationship between e-portfolio content and academic performance. Additional examples of text mining in higher education and ethical considerations pertaining to this technology are also discussed.
John Zilvinskis, Indiana State UniversityGreg V. Michalski, Florida State College at Jacksonville
Copyright © Association for Institutional Research 2016