APA Dictionary of Psychology

n. in categorical data analysis, a situation in which obtained observations display more variation (dispersion) than is predicted by a model. Overdispersion distorts standard error and confidence interval estimates and is often associated with outliers or model misspecification.