This study investigates how faculty, student, and course features are linked to student outcomes in Learning Assistant (LA) supported courses. Over 4,500 students and 17 instructors from 13 LA Alliance member institutions participated in the study. Each participating student completed an online concept inventory at the start (pre) and end (post) of their term. The physics concept inventories included Force and Motion Concept Evaluation (FMCE) and the Brief Electricity and Magnetism Assessment (BEMA). Concepts inventories from the fields of biology and chemistry were also included. Our analyses utilize hierarchical linear models that nest student level data (e.g. pre/post scores and gender) within course level data (e.g. discipline and course enrollment) to build models that examine student outcomes across institutions and disciplines. We report findings on the connections between students' outcomes and their gender, race, and time spent working with LAs as well as instructors' experiences with LAs.