Data with a multilevel structure has become more frequent in various fields of interest, such as health sciences, public health, epidemiological, and health economics research where binary outcomes are common. Multilevel logistic regression models allow you to account for the clustering of subjects within clusters of higher‐level units when estimating the effect of subject and cluster characteristics on subject outcomes.

Unlike linear regression, which predicts a continuous outcome, logistic regression models a binary outcome (e.g., success/failure, yes/no, 0/1), while multilevel logistic regression models can compute intercepts and slopes for nested units. In terms of data structure, multilevel logistic regression models are used when data have a multilevel structure, such as when subjects are clustered within groups. This type of structure allows for rigorous data analysis without being held to a typical randomized clinical trial (RCT) design that uses simple random sampling. In this course, we will introduce the basic random intercept logistic regression model that will be used throughout the course.  

The purpose of multilevel logistic regression is to estimate the odds of an event occurring while accounting for the dependency of data. It is often used in research areas like health services, population health, and epidemiology where binary outcomes are common. Multilevel logistic regression models can account for the clustering of subjects within higher-level units. For instance, a multilevel logistic regression model might consider how the probability of an event depends on both individual-level variables and higher-level variables.  This training equips you with the competencies to interpret model parameters, including fixed effects and variance components, and describe multiple logistic regression analysis, among others.

By the end of this training, you will be able to