Patel Chandni C, Fairchild Amanda J, Prinz Ronald J
University of South Carolina, Columbia, SC, USA.
Clin Child Fam Psychol Rev. 2017 Jun;20(2):127-145. doi: 10.1007/s10567-016-0221-2.
Parenting and family interventions have repeatedly shown effectiveness in preventing and treating a range of youth outcomes. Accordingly, investigators in this area have conducted a number of studies using statistical mediation to examine some of the potential mechanisms of action by which these interventions work. This review examined from a methodological perspective in what ways and how well the family-based intervention studies tested statistical mediation. A systematic search identified 73 published outcome studies that tested mediation for family-based interventions across a wide range of child and adolescent outcomes (i.e., externalizing, internalizing, and substance-abuse problems; high-risk sexual activity; and academic achievement), for putative mediators pertaining to positive and negative parenting, family functioning, youth beliefs and coping skills, and peer relationships. Taken as a whole, the studies used designs that adequately addressed temporal precedence. The majority of studies used the product of coefficients approach to mediation, which is preferred, and less limiting than the causal steps approach. Statistical significance testing did not always make use of the most recently developed approaches, which would better accommodate small sample sizes and more complex functions. Specific recommendations are offered for future mediation studies in this area with respect to full longitudinal design, mediation approach, significance testing method, documentation and reporting of statistics, testing of multiple mediators, and control for Type I error.
育儿及家庭干预措施在预防和治疗一系列青少年问题方面已多次显示出成效。因此，该领域的研究人员开展了多项研究，运用统计中介分析来探究这些干预措施发挥作用的一些潜在作用机制。本综述从方法论角度审视了基于家庭的干预研究在何种程度上以及如何有效地检验了统计中介作用。一项系统检索确定了73项已发表的结果研究，这些研究针对广泛的儿童和青少年问题（即外化问题、内化问题和药物滥用问题；高风险性行为；以及学业成绩），对与积极和消极育儿、家庭功能、青少年信念和应对技能以及同伴关系相关的假定中介因素，检验了基于家庭干预措施的中介作用。总体而言，这些研究采用了能够充分解决时间先后顺序问题的设计。大多数研究使用了系数乘积法来进行中介分析，这种方法更受青睐，且比因果步骤法限制更少。统计显著性检验并非总是采用最新开发的方法，而这些方法能更好地适应小样本量和更复杂的函数。针对该领域未来的中介分析研究，在完整纵向设计、中介分析方法、显著性检验方法、统计数据的记录与报告、多个中介因素的检验以及第一类错误的控制等方面提出了具体建议。
