Kashy Deborah A, Donnellan M Brent, Burt S Alexandra, McGue Matt
Department of Psychology, Michigan State University, East Lansing, MI 48824, USA.
Dev Psychol. 2008 Mar;44(2):316-29. doi: 10.1037/0012-1649.44.2.316.
Growth modeling is a useful tool for studying change over time, and it is becoming increasingly popular with developmental researchers. There is a considerable methodological literature surrounding growth modeling for individuals; however, far less attention has been focused on growth models for pairs of related individuals (i.e., dyads). In this article, the authors consider dyadic growth models for those cases where there are no relevant variables that can empirically distinguish between dyad members (e.g., same-sex twins or best friends). The authors describe how researchers can estimate growth models for indistinguishable dyads using both multilevel modeling and structural equation modeling. Although both approaches can be used to estimate the same underlying models, the authors focus on practical similarities and differences between the two approaches. They illustrate modeling issues using an overtime study of adolescent twins' conflict with their mothers, a substantively important topic given the enduring interest in parent-child relationships during adolescence.
增长模型是研究随时间变化的一种有用工具，并且在发展研究人员中越来越受欢迎。围绕个体增长模型有大量的方法论文献；然而，对于相关个体对（即二元组）的增长模型关注要少得多。在本文中，作者考虑了二元组增长模型，适用于不存在能够凭经验区分二元组成员的相关变量的情况（如同性双胞胎或最好的朋友）。作者描述了研究人员如何使用多层建模和结构方程建模来估计难以区分的二元组的增长模型。尽管这两种方法都可用于估计相同的潜在模型，但作者重点关注这两种方法在实际中的异同。他们通过一项对青少年双胞胎与其母亲冲突的长期研究来说明建模问题，鉴于对青春期亲子关系的持久兴趣，这是一个具有实质重要性的主题。
