The representation of a longitudinal process for many individuals where the components of growth are hypothesized to be unobserved or latent scores and the errors of measurement at each occasion are independent. It is also possible to add a latent shape to this model so the changes represented by the latent slopes are not equal over equal intervals of time. Other advantages of the technique are that both group and individual levels of either linear or curvilinear change can be estimated, measurement occasions do not need to be equally spaced, measurement errors can be accounted for, as can multiple predictors or correlates of change, including change in multivariate latent factors. The technique, also known as latent growth curve analysis, is part of the structural equation modeling approach to repeated measures data, and thus has similarities with multilevel modeling.
See Latent factor/score/variable, Latent trait, Linear growth model, Longitudinal design, Longitudinal studies, Multilevel modeling, Multivariate analysis of variance (MANOVA), Repeated measures analysis of variance, Structural equation modeling (SEM)