1-014 - The Application of Latent Growth Modeling with Categorical Data

Session Type: Paper Symposium

Integrative Statement

Developmental researchers have used latent growth curve modeling (LGCM) to describe and analyze change over time. Based on the assumption of a normal distribution in continuous repeated data, maximum likelihood (ML) estimation is used in conventional LGCM to estimate growth parameters. However, it is also possible to have repeated measures of non-continuous (i.e., categorical) measures, which are not normally distributed. In such cases, the conventional LGCM can lead to biased growth parameter estimates. Moreover, as researchers model more intricate patterns of change in family attributes to test complex hypotheses derived from developmental and family theories, various extensions of LGCM, such as second-order LGCMs and growth mixture models, are warranted.
Within a SEM framework, a conventional LGCM can be extended not only to first- and second-order growth models but also to growth mixture models. More importantly, all of these LGCM extensions can be applied to categorical data, including dichotomous (binary) and ordinary variables. The first paper of this symposium introduces conventional LGCMs with binary variables. The second paper focuses on growth mixture modeling with binary variables. The third paper introduces two competing higher-order growth curve mixture modeling approaches (Curve-of-Factors and Factor-of-Curves) with binary variables. Using longitudinal panel data from 444 adolescents (mean age = 13.5, 53% female in 1990) who participated in the Iowa Family Transition Project during a 4-year period (1990-1994; PI. Rand D. Conger), this symposium will introduce and illustrate the application of various extensions of latent growth curve modeling and growth mixture modeling for continuous and categorical data.

Sub Unit

• Methods, History, Theory

Chair

• Emilie Smith, University of Georgia, USA

Discussant

• Assaf Oshri, University of Georgia

Individual Presentations

• An Introduction to Latent Growth Curve Modeling and Its Application for Categorical Data - Presenting Author: Catherine Walker O'Neal, The University of Georgia; K.A.S. Wickrama, The University of Georgia

• An Introduction to Growth Mixture Modeling and Its Application for Categorical Data - Presenting Author: K.A.S. Wickrama, The University of Georgia; Tae Kyoung Lee, University of Miami

• High-Order Growth Curves and Mixture Modeling with Categorical Data (Curve-of-Factors vs. Factor-of-Curves) - Presenting Author: Tae Kyoung Lee, University of Miami