Previous |
Next |
October 26, 2007, 08:48 AM ET
Predicting Student Struggles in College
Seattle—Colleges now collect a wealth of data on their students. And within that data lies a way to keep students from failing a class by finding early signs of trouble, says John P. Campbel, associate vice-president of teaching and learning technologies at Purdue University.
At a session at the Educause technology meeting here, Mr. Campbell noted that course-management systems now track student participation in assignments.
He looked at 27,276 Purdue students enrolled in 597 courses. By putting together this engagement data with their GPA’s and scores on standardized tests, Campbell developed a model that successfully predicted students who were heading for a grade of C or lower 66 percent of the time.
In a freshman biology class, students who fit the failing profiles were first sent e-mail messages or told by the instructor about study sessions and extra help available.
Did it work? Not for the highest risk, least engaged students. But for a middle-risk group, the interventions helped. They went to study sessions, got tutoring, and moved out of the high-risk category in their behavior. —Josh Fischman
Categories: Educause-2007
Add Your Comment
Commenting is closed.