The uses of multilevel modelling in data that occurs at several levels

Multilevel models are increasingly used in the social and health sciences for the analysis of clustered data. There are two main reasons for this. Firstly, failing to properly account for clustering in a single level regression analysis can lead to incorrect inferences. Secondly, the ability of multilevel models to simultaneously analyse data at multiple levels of systems provides a framework for investigating a wider range of types of research question than is available with traditional techniques. This introductory talk describes the basic multilevel model and gives examples of analysis where the quality of inference is improved and new types of research question are addressed because a multilevel modelling approach is adopted.