Implementing Predictive Analytics in Academic Advising

This paper examines how academic advisers interpret and apply predictive analytic tools, what training is necessary for advisers to best prepare them to use predictive data appropriately, and what institutional and end-user needs should be anticipated and supported when using predictive analytics in advising. We conducted a pilot study involving a small cohort of academic advisers who used an analytic tool developed at Penn State that predicts individual students’ course grades. Findings from focus groups and interviews suggest that the adoption of predictive analytics requires space and time for human users to develop a competent understanding of the data they are working with to successfully operationalize and contextualize its use. Additionally, institutions need to create ongoing training, support, and sustained community among users and avenues for advocacy to address systemic issues. Without this, mental shortcutting and optimization of workflows have the potential to yield behaviors where data is more likely to be used out of context of its original intent or purpose, raising the possibility for flawed or even unethical decision-making when supporting students.