Skip to main content
Original Articles

Modelling Behaviour with Multivariate Multilevel Growth Curves

Published Online:https://doi.org/10.1027/1614-1881.1.2.71

Abstract. Developmental data are often both longitudinal and multivariate and can be handled within a multilevel framework. This paper fits a range of multivariate multilevel models to both continuous and binary data to examine the ways in which a set of behavioural measures change together with age. These data were collected by researchers in Montreal, Canada. Methodologically, we find that within and between individual correlations vary only a little according to the ways in which the models are specified. Substantively, we find that measures of aggression and opposition are closely related but both are less closely related to a measure of hyperactivity. Models for the effects of socio-economic status on levels and changes are fitted, as are models that examine change conditional on an initial measure. The findings are compared with those previously obtained using a growth trajectories approach.

References