<p>The purpose of this research is to investigate the social antecedents of academic performance over time, as well as understand if a longitudinal view of performance affords greater understanding of graduation and retention. Academic performance across four years was tracked among 1,911 university students. Embeddedness prior to institutional entry was examined alongside high school grade point average and standardized test scores as predictors of performance improvement over time. Embeddedness was also examined as a potentially stronger predictor of performance improvement among ethnic minorities. Results indicate that Black and Hispanic students with greater social ties were more likely to improve their academic performance over time. Additionally, initial academic performance and improvement in performance over time were uniquely related to graduation and retention rates. These results suggest that performance improvement is a novel and valuable conceptualization of academic performance. Further, universities may promote performance improvement by helping newcomers form bonds with current members. Further research may identify the specific supports current members provide to inform structural solutions of such supports. Such research and policy interventions may be of relatively greater benefit for Black and Hispanic students.</p>