Examination of a Cluster Effect on Model Specification in Multilevel Latent Transition Analysis

• In Event: 69.046 - Latent Growth/Transition Modeling and Model Selection

Abstract

Latent transition analysis has been used in the wide range of research including education, psychology, public health, medicine, etc. However, the hierarchical structure of data has often been ignored in a LTA modeling, which is most commonly due to the complexity of LTA model specification and its identification issue in estimating parameters in LTA. However, such ignorance may lead to erroneous conclusion due to invalid parameter estimates. In this paper, we derive more general forms in the model specification of a multilevel LTA with random effects than those in Asparouhov & Muthén (2008) and Kaplan, Kim, & Kim (2009) and demonstrate the differences between single level LTAs and multilevel LTAs using children’s behavioral data in ECLS-K:2011.

Author

• Ji Hoon Ryoo, University of Virginia