Assessing Admissions Data for Non-Cognitive Indicators of Success

After decades of research into how to predict student success based on pre-enrollment data, even the most sophisticated models offer limited value in admissions. Statistical analysis of traditional success indicators like high school GPA, class rank, and standardized tests can be a useful part of an admissions office’s toolbox, but most pre-enrollment models still fail to explain most of the variance in student success.

Non-cognitive variables like grit seem to offer admissions offices new insight in evaluating applicants, but dedicated, purpose-built non-cognitive assessments are either too new to judge their effectiveness, or produce ambiguous results in practice. The best chance for realizing medium-term value from non-cognitive measures may actually lie closer than we thought: in existing application data.

This webconference, the second in our series for directors of admissions, will review several early but promising experiments that draw non-cognitive "data" from a new reading of existing application information or from a student's application behavior. Each of these approaches, though still works in progress, have realized tangible ROI or increased the predictiveness of pre-enrollment models of success.