Structural Equation Modeling, 2nd Edition

4 Latent growth modeling (LGM) for longitudinal data analysis

4.1 Introduction

In this chapter, we will expand the application of structural equation modeling (SEM) to longitudinal data analysis where subjects are followed up over time with repeated measures of each variable of interest. The distinctive features of longitudinal data include, but are not limited to, the following: (i) there are two sources of heterogeneity – within‐subject or intra‐subject variation and between‐subject or inter‐subject variation; (ii) within‐subject observations usually are not independent; (iii) between‐subject variation may not be constant over time; and (iv) longitudinal data are often incomplete or unbalanced (i.e. the number of repeated measures and time intervals between follow‐ups varies by subject). Various new statistical methods have been developed for longitudinal data analysis, among which multilevel modeling (MLM) (Bryk and Raudenbush 1992; Goldstein 1987, 1995; Mason et al. 1983; Raudenbush and Bryk 2002), generalized estimating equations (GEEs) (Diggle et al. 1998; Diggle et al. 2002), and latent growth modeling (LGM) (Chou et al. 1998; Duncan and Duncan 1994; Duncan et al. 2006; McArdle and Anderson 1990; Meredith and Tisak 1990; Muthén 1991; Willett and Sayer 1994) have gained popularity in longitudinal studies. All these approaches enable us to deal with the special features of longitudinal data. However, compared with MLM and GEEs, LGM is a more generalized approach ...