Dr. Chung's research interests pertain to mixture modeling in particular, latent class and latent transition models. A latent class model explains the relationships among categorical variables by assuming the existence of an unseen classifier. Latent transition models extend this idea to longitudinal settings by allowing subjects to change their latent statuses over time. Dr. Chung extended these models by allowing the class membership to depend on covariates, developing and implementing computational algorithms for modeling and inference. Through this research, he has studied difficult issues in likelihood-based and Bayesian inference for finite mixtures such as the common label switching problem.

Dr. Chung's current research focuses on the application of latent class models in prevention science. In particular, he is investigating latent class models for historical shifts in marijuana use and attitudes from 1975 to the present using data from the Monitoring the Future Study. Dr. Chung has also been studying the impact of pubertal  development on adolescent substance-use behavior.