Detecting Appropriate Trajectories of Growth in Latent Growth Models: The Performance of Information-Based Criteria

• In Event: 46.079 - Appropriate Modeling of Latent Growth and Relationships

Abstract

This paper assessed the performance of six information-based criteria (i.e., AIC, AICC, CAIC, BIC, nBIC, and HQIC) used to select among a set of five competing latent growth models with different growth trajectories, including linear, quadratic, cubic, piecewise, and unspecified. Sample size and the number of measurement occasions were also varied. The linear model was differentiated the most accurately by the criteria whereas the unspecified model was differentiated the least accurately by the criteria. The accuracy of the criteria tended to improve as sample size increased. The quadratic and cubic growth models were more accurately differentiated as the number of measurement occasions increased. Certain criteria performed more optimally under different conditions. Implications of the findings are discussed.

Authors

• Tiffany Ann Whittaker, The University of Texas - Austin

• Jam Khojasteh, Oklahoma State University - Tulsa