We study the distributional effects of remote learning using a novel approach that combines preference data from a conjoint survey experiment with administrative student records. The experimentally ...