Chalk Talk: Cross-Classified and Multi-membership Multilevel Models

Hedeker, Donald. (2014, April 15). Cross-Classified and Multi-membership Multilevel Models. Institute for Health Research and Policy, University of Illinois at Chicago.

Cross-classified multilevel models apply when, say, a student is clustered within both a grammar and high school, but not all students from a given grammar school go the same high school. In this situation, the two types of clusters — grammar schools and high schools — do not represent hierarchical factors, but crossed factors. Multi-membership models are used in studies in which, for example, an individual patient may see multiple health care providers, or when a person is observed in multiple neighborhoods. In this case, the subject is not nested within a single cluster, but is a member of multiple clusters. For both of these types of multilevel models, Dr. Hedeker presented applications and describe software syntax (SAS, SPSS, STATA) to estimate the model parameters.

See a PDF of Dr. Hedeker's presentation (in six-frame pages). Save this zip file with data, syntax examples, and output to your desktop. Then open it.

Learn more about Dr. Hedeker.