A group-based modeling approach to estimating longitudinal trajectories of Korean adolescents' on-line game time

Abstract

The purpose of this study was to identify the number and the pattern of latent classes based on online game time and to test the significant determinants of classes among the variables, such as sex, self-control, self-esteem, aggressiveness, emotion regulation, academic stress, academic motivation and social skills. In order to identify the latent classes, we applied Nagin's semiparametric, group-based approach to the four waves of longitudinal data from the Korean Youth Panel Study using SAS Macro. As a result, four latent classes emerged according to different patterns of change in game play time: the low group, the rising group, the declining group and the chronic group. As for the influential factors, the results of conducting a multinomial logit analysis indicated that all factors, except social skills, proved to be significant. (C) 2013 Elsevier Ltd. All rights reserved.