如何描述发展趋势的差异：潜变量混合增长模型

在追踪研究中，研究者不仅关心某一特质随时间的发展趋势，而且关注个体之间发展趋势的差异及其存在差异的原因。在总体发展同质的情形下，多层线性模型和潜变量增长曲线模型为解决这一问题提供了切实有效的方法。但是如果所研究的总体本身不同质，就需要一种能够描述总体中不同质子总体的不同发展特点的方法。该文简要介绍了一种能够描述不同群体不同发展趋势特征的统计模型——潜变量混合增长模型，并通过一个实际例子介绍了这一方法的应用过程，同时说明了潜变量混合增长模型与多层线性模型和潜变量增长曲线模型之间的关系

How to Abstract Developmental Variations: Latent Growth Mixed Model

Developmental research involves the identification of individual differences in change as well as understanding the process of change itself. The contemporary approach to the analysis of change,as Hierarchical Linear Model(HLM) and Latent Growth Curve Model(LGCM) ,has focused on growth curve modeling that explicitly considers both intraindividual change and interindividual differences in such change,but treats the data as if collected from a single population. This assumption of homogeneity in the growth parameters is often unrealistic. If heterogeneity is ignored,statistical analyses and their effects can be seriously biased. This paper presents a procedure that accounts for sample heterogeneity---Latent Growth Mixed Model(LGMM) ---and their application to longitudinal data. In addition,the difference of HLM and LGMM,and the difference of LGCM and LGMM were discussed briefly.