Liu Li C, Flay Brian R
Division of Epidemiology and Biostatistics, School of Public Health, University of Illinois at Chicago, Chicago, IL 60612, USA.
Prev Sci. 2009 Sep;10(3):197-207. doi: 10.1007/s11121-009-0125-1.
This study illustrates a method to evaluate mediational mechanisms in a longitudinal prevention trial, the Aban Aya Youth Project (AAYP). In previous studies, interventions of AAYP were found to be effective in reducing the growth of violence, substance use and unsafe sex among African American adolescents. In this article, we hypothesized that the effects of the interventions in reducing the growth of substance use behavior were achieved through their effects in changing intermediate processes such as behavioral intentions, attitudes toward the behavior, estimates of peers' behaviors, best friends' behaviors, and peer group pressure. In evaluating these mediational mechanisms, difficulties arise because the growth trajectories of the substance use outcome variable and some of the mediating variables were curvilinear. In addition, all of the multivariate mediational measures had planned missing data so that a score from the multiple items for a mediator could not be formed easily. In this article, we introduce a latent growth modeling (LGM) approach; namely, a two-domain LGM mediation model, in which the growth curves of the outcome and the mediator are simultaneously modeled and the mediation effects are evaluated. Results showed that the AAYP intervention effects on adolescent drug use were mediated by normative beliefs of prevalence estimates, friends' drug use behavior, perceived friends' encouragement to use, and attitudes toward the behavior.
本研究阐述了一种在纵向预防试验——阿班阿亚青年项目（AAYP）中评估中介机制的方法。在先前的研究中，发现AAYP干预措施在减少非裔美国青少年中的暴力行为、物质使用和不安全性行为增长方面是有效的。在本文中，我们假设干预措施在减少物质使用行为增长方面的效果是通过其对改变诸如行为意图、对行为的态度、对同伴行为的估计、最好朋友的行为以及同伴群体压力等中间过程的影响来实现的。在评估这些中介机制时，会出现困难，因为物质使用结果变量和一些中介变量的增长轨迹是曲线的。此外，所有多变量中介测量都存在计划缺失数据，因此难以轻易形成中介变量多个项目的得分。在本文中，我们引入了一种潜在增长模型（LGM）方法；即一个双域LGM中介模型，其中同时对结果和中介的增长曲线进行建模并评估中介效应。结果表明，AAYP对青少年药物使用的干预效果是由流行率估计的规范信念、朋友的药物使用行为、感知到的朋友对使用的鼓励以及对行为的态度所介导的。
