5 Questions to Ask About Predictive Student Data
Look beyond whether a model can identify students at risk. Find out:
How early does a meaningful risk signal emerge? Earlier identification means more time to respond. Additional data may strengthen later predictions.
What data are driving the prediction? Does the model draw from admissions measures, course performance, assessments, engagement or multiple sources?
What does the signal tell us about the student's needs? A general risk designation may be less actionable than information about developing academic knowledge or performance concerns.
How certain is the prediction? A prediction represents risk, not the student’s destiny. Understanding uncertainty can help educators interpret appropriately.
What can faculty do with the information? Identifying risk is the start. Programs also need ways to investigate the signal, determine support and monitor progress.