Designed to handle hierarchical and clustered data, it is a statistical technique developed by Harvey Goldstein and colleagues to analyse longitudinal data in which there are variables at several levels, with the lower levels being nested in the higher ones. While it is a form hierarchical regression analysis, analogous to multivariate analysis of variance for repeated measures and related to structural equation modelling, MLM has the advantage of the inclusion of a powerful algorithm for dealing with missing data, an ever-present problem in longitudinal studies. Involves the estimation of parameters by means of an iterative model fitting to the data, which enables the inclusion of incomplete records. The statistics associated with main and interaction effects follow a chi-square distribution.
See Attrition, Bock’s profile analysis, Deviance statistic, Hierarchical data structure, Latent growth model, Longitudinal studies, Missing at random (MAR), Mixed-effects models, Multivariate analysis of variance (MANOVA), Person-specific variance, Structural equation modelling (SEM)