This study analyzed assessment data from over 17,000 humanities and science students to estimate their learning gains over time and compare results between disciplines and universities. Multilevel modeling found significant variance between departments and students in learning gains for both universities studied. While overall positive gains were observed, aggregate estimates can misrepresent gains at specific discipline or degree levels. The results suggest learning analytics may help identify areas for targeted intervention and guide learning design by visualizing grade trajectories across programs and institutions.