Commentary: Reflecting on Holmbeck (2002): Making Advanced Quantitative Methods Approachable in Pediatric Psychology.
Commentary: Reflecting on Holmbeck (2002): Making Advanced Quantitative Methods Approachable in Pediatric Psychology.
复制标题
评论：反思 Holmbeck (2002)：使先进的定量方法在儿科心理学中变得平易近人。
DOI：
10.1093/jpepsy/jsz035
复制
发表时间：
2019
影响因子：
3.6
通讯作者：
Armstrong,Bridget
中科院分区：
文献类型：
--
作者：
Armstrong,Bridget
在过去的几十年里，儿科心理学领域已经从描述（Elkins＆Roberts，1988）发展到解释现象，询问过程为谁以及如何发生的问题（Karazsia＆Berlin，2018；Karazsia，Berlin，Armstrong，Janicke，＆Darling，2014）。该领域内这种转变的最初迹象可以从早期对理论驱动研究和过程分析的呼吁中看出，这些研究和过程分析考虑了个体水平的差异（Peterson, Farmer, & Mori, 1987; Wallander, 1992）。对过程和以人为本的方法的日益重视导致研究问题日益复杂，从而需要复杂的统计技术。 1997 年，Holmbeck 发表了对调节者和调解者的概述（Holmbeck，1997），并在 2002 年儿科心理学杂志（JPP）出版物中扩展了这项工作，其中详细介绍了检查调节和调解的其他统计步骤；即如何对显着的中介和调节效应进行事后调查（Holmbeck，2002）。 Holmbeck (2002) 中的实用方法策略为为什么应该进行事后探究提供了基本原理，并提供了如何使用相关儿科心理学例子进行这些分析的说明。这种平易近人的方法，与日益强调理论驱动研究的时代精神相结合，有助于解释这些出版物在儿科心理学领域内外的影响（Aylward、Roberts、Colombo 和 Steele，2007）。近二十年后，尽管技术进步减少了计算工作并促进了更新技术的采用，但 Holmbeck（2002）关于探索复杂问题重要性的核心信息仍然存在（Hayes，2013）。研究人员提出的日益复杂的理论在过去十年中推动了方法论的大量进步（Preacher，2015）。根据 Holmbeck 自己提出的监测并持续改进儿科心理学研究和统计程序质量的建议（Holmbeck，2008），该领域似乎正在拥抱调解方法论的进步（例如，Psihogios 等，2017），通过使用引导法、蒙特卡罗方法或贝叶斯置信区间来测试调解显着性（Preacher，2015）。尽管Baron和Kenny（1986）的“建立调解的标准”逻辑仍然可以在现代研究中找到（O’Rourke＆MacKinnon，2018），但它的受欢迎程度正在减弱，并且不再受到调解方法学家的推荐（Hayes＆Rockwood，2017）。这种偏离显着 X→ Y 关系（Baron 和 Kenny 的步骤 1）的要求与儿科干预研究特别相关，即使在总体上没有显着干预效果的情况下，调查调解也可能提供有价值的信息（Hinshaw，2007；O’Rourke & MacKinnon，2018）。更新的调节探索技术的示例也可以在儿科心理学中找到。 Okado、Long 和 Phipps（2014）研究了生活压力事件如何缓解父母与孩子的痛苦之间的关系。作者使用 Johnson-Neyman 技术（Lorah & Wong，2018）来确定这种调节效应的显着性极限，并发现父母和孩子的痛苦只是
Over the past several decades, the field of pediatric psychology has progressed from describing (Elkins & Roberts, 1988) to explaining phenomena, asking questions about for whom and how processes occur (Karazsia & Berlin, 2018; Karazsia, Berlin, Armstrong, Janicke, & Darling, 2014). Initial signs of this shift within the field can be seen in early calls for theory driven research and process analyses that account for individual-level differences (Peterson, Farmer, & Mori, 1987; Wallander, 1992). This growing appreciation for process and person-oriented approaches resulted in increasingly complex research questions, and subsequently, the need for complex statistical techniques. In 1997, Holmbeck published an overview of moderators and mediators (Holmbeck, 1997), and expanded on this work in a 2002 Journal of Pediatric Psychology (JPP) publication which detailed additional statistical steps for examining moderation and mediation; namely, how to conduct post hoc probes of significant mediation and moderation effects (Holmbeck, 2002). The practical methods strategies in Holmbeck (2002) offered rationale for why post hoc probing should be done, and provided instructions on how to conduct these analyses using relevant pediatric psychology examples. This approachable method, combined with the zeitgeist of increased emphasis on theory-driven research, helps to explain the impact of these publications, both within and beyond the field of pediatric psychology (Aylward, Roberts, Colombo, & Steele, 2007). Nearly two decades later, the core message in Holmbeck (2002) regarding the importance of exploring complex questions persists, even as advances in technology have reduced computational efforts and facilitated the adoption of updated techniques (Hayes, 2013). The increasingly complex theories produced by researchers have motivated a great deal of methodological advances in the past decade (Preacher, 2015). In line with Holmbeck’s own recommendation to monitor and continually improve the quality of research and statistical procedures within pediatric psychology (Holmbeck, 2008), the field appears to be embracing the methodological advances in mediation (e. g., Psihogios et al., 2017) by using bootstrapping, Monte Carlo methods, or Bayes confidence intervals to test for mediation significance (Preacher, 2015). Although the Baron and Kenny (1986)“criteria to establish mediation” logic can still be found in modern research (O’Rourke & MacKinnon, 2018), it is waning in popularity and no longer recommended by mediation methodologists (Hayes & Rockwood, 2017). This pivot away from the requirement of a significant X→ Y relation (Baron & Kenny’s step 1) is particularly relevant to pediatric intervention research, where investigating mediation, even in the absence of an overall significant intervention effect, may provide valuable information (Hinshaw, 2007; O’Rourke & MacKinnon, 2018).Examples of updated moderation probing techniques can also be found within pediatric psychology. Okado, Long, and Phipps (2014) examined how stressful life events moderate the relation between parent and child distress. Authors used the Johnson-Neyman technique (Lorah & Wong, 2018) to identify the limits of significance for this moderation effect and found that parent and child distress were only