Multilevel models provide an effective means for studying individuals who are clustered into common higher-level organizational contexts (e.g., students clustered in schools, patients clustered in hospitals, children clustered in neighborhoods). One limitation of traditional multilevel models is that they require individuals to be “purely clustered” in higher-level contexts. This requirement is problematic when individuals are clustered into multiple contexts at a given level of a data hierarchy (e.g., students attend middle schools and high schools, but not all students from a given middle school are fed into the same high school). Cross-classified random effects models (CCREMs) and multiple membership random effects models (MMREMs) are flexible extensions of traditional multilevel models that do not require “pure clustering” of individuals in higher-level contexts. This presentation will provide an overview of CCREM and MMREM techniques with a focus on identifying and dealing with
commonly occurring cross-classified and multiple membership data structures.