Hertzog Melody
1 University of Nebraska Medical Center, Lincoln, NE, USA.
West J Nurs Res. 2018 Jun;40(6):907-930. doi: 10.1177/0193945916687782. Epub 2017 Jan 12.
The purpose of this study was to describe common approaches used by nursing researchers to test mediation models and evaluate them within the context of current methodological advances. MEDLINE was used to locate studies testing a mediation model and published from 2004 to 2015 in nursing journals. Design (experimental/correlation, cross-sectional/longitudinal, model complexity) and analysis (method, inclusion of test of mediated effect, violations/discussion of assumptions, sample size/power) characteristics were coded for 456 studies. General trends were identified using descriptive statistics. Consistent with findings of reviews in other disciplines, evidence was found that nursing researchers may not be aware of the strong assumptions and serious limitations of their analyses. Suggestions for strengthening the rigor of such studies and an overview of current methods for testing more complex models, including longitudinal mediation processes, are presented.
本研究的目的是描述护理研究人员用于检验中介模型并在当前方法学进展的背景下对其进行评估的常用方法。利用MEDLINE检索2004年至2015年发表在护理期刊上检验中介模型的研究。对456项研究的设计（实验性/相关性、横断面/纵向、模型复杂性）和分析（方法、中介效应检验的纳入、假设的违背/讨论、样本量/效能）特征进行编码。使用描述性统计确定总体趋势。与其他学科的综述结果一致，有证据表明护理研究人员可能没有意识到其分析的强假设和严重局限性。本文提出了加强此类研究严谨性的建议以及当前用于检验更复杂模型（包括纵向中介过程）的方法概述。
