Blackwell Reference Online

Extract

Longitudinal methods permit the systematic study of stability and change over time and thus are a powerful tool for examining processes underlying social phenomena and the causal relation between different constructs. There are many different statistical methods to analyze longitudinal data (for an overview, see Crowder & Hand 1996 ; Menard 2002 ). Most of these techniques are related to applications of the general linear model to continuously distributed repeated measures (e.g., repeated t tests, ANOVA, multiple regressions, autoregressive models). A major shortcoming of these methods is that they are fixed-effects models assuming that there are no interindividual differences in change. One prominent and powerful method to overcome this shortcoming are latent growth curve models (LGMs; synonymous terms used in literature are “latent curve models” or “latent trajectory models”) as one variant of structural equation modeling ( Bollen & Curran 2006 ). With LGM it is possible to analyze individual trajectories and related processes of change over time. Meredith and Tisak (1990) first proposed LGM based on the seminal work of Tucker and Rao in 1958. LGMs are based on the assumption that a set of observed repeated measures taken on a given individual over time can be used to estimate an unobserved underlying trajectory. This assumption can be formally expressed as y it ... log in or subscribe to read full text