Growth Mixture Modeling: A method for describing specific class growth trajectory

追踪研究可以展现个体随时间的变化趋势、个体间差异并更好地揭示变量之间的因果关系,日益受到社会科学研究者的青睐,处理追踪数据的统计模型也日新月异。传统的分析模型假设样本同质,然而当此假设不满足时,就需要考虑群体的异质性问题。目前,处理异质性样本增长问题的模型主要有潜类增长模型(或组基增长模型)和增长混合模型。增长混合模型构成一般的形式,在一定条件下传统的潜在增长曲线模型和潜类增长模型均是其特例。沿着增长混合模型的发展路径,本文从应用性角度介绍了上述三个模型各自的特点、优势和Mplus软件实现,并通过一个实例演示了分析过程。最后,对当前研究存在的问题和将来的研究方向做了简要讨论。 

Longitudinal research has attracted more and more attention among social science researchers in recent years, because it could describe the within-person change and between-person differences in change simultaneously, along with special advantages in explaining causality. For conventional latent growth curve modeling (LGCM), it is not to be hold to assume all individuals are drawn from a homogeneous population in many applied research situations. When the population is heterogeneous, the growth mixture modeling (GMM) and latent class growth modeling (LCGM; or group-based growth modeling) are the best choices. In fact, LGCM and LCGM are the special types of GMM as the universal model. The purpose of this paper is to provide a non-technical overview of these growth modeling (LGCM, LCGM and GMM). The modeling procedures are illustrated with a practical example. Finally, current debates and issues in the modeling process, as well as the new directions for future research, are briefly discussed.

欢迎阅读《社会学研究》！您是该文第3970位读者！