When analysing individual trajectories across the life course, the fact that repeated observations of individuals are not independent (i.e. correlated) should be taken into account. Several techniques to do this are available, such as mixed models, MM and latent growth models, LGM. These models can also elegantly incorporate different stages of the life course, including childhood, adolescence and adulthood in the modelling process. MM do so by the inclusion of a ‘time’ variable denoting each stage in the model and LGM can be conducted in a piecewise manner, where each ‘piece’ represents a life course stage. Moreover, both techniques can further be extended to allow for possible heterogeneity in health trajectory (shape), but do so in different ways. MM can include random slopes to account for heterogeneity in growth; LGM can be extended into latent class growth models to allow for the possible revelation of subgroups of individuals determined by the data with distinct health trajectories across the life course.
This chapter will explain and compare the two techniques using existing life course data of the Amsterdam Growth and Health Study cohort and combines a methodological focus with empirical findings.
Copyright 2015, The Author(s).