[Knowledge Sharing] Mediation analysis with Common Error 1: Establishing Mediation Hypotheses without theory-Based
当代中介分析:常见问题一： 建立中介假设，却未提供理论依据 by Shiau et al.(2026)
Source from: Shiau, W.-L., Shih, C.-H., Qureshi, I., Yuan, Y., & Teng, C.-I. (2026). Do matched and mismatched null hypothesis significance testing of mediators matter? An example of fintech continuance and guidelines for the future. Decision Support Systems, 211, 114764.
Guideline 1: Establishing Theory-Based Mediation Hypotheses
Researchers should establish each mediation hypothesis before conducting confirmatory analysis. A hypothesis should identify the proposed X → M → Y mechanism, and its theoretical development should explain what M represents and how it transmits X’s influence to Y. Researchers should also specify whether the hypothesis concerns a single mediation mechanism or the entire model. Merely stating that M is important or that “M mediates” does not provide sufficient mechanism-based justification. Mediation identified after examining the results should be reported transparently as exploratory or post hoc analysis.
指引一：建立有理论依据的中介假设研究者应在分析之前，提出中介假设，并说明X如何透过M影响Y。假设发展段落需要交代M代表的理论构念，以及M为何能影响X并影响Y。同时，研究者应界定假设是针对单一中介机制，还是在整体模型提出。仅说明M很重要，或写出M具有中介作用，仍不足以建立完整的机制论证。
研究者应该/不应该做的事
应做事项 | 不应做事项 |
1. Clearly state the mediation hypothesis in the hypothesis section. 1. 在假设段落中陈述中介假设。 | 1. Only the entire mediation model or post hoc was tested, ignoring the original mediation hypothesis. 1. 仅检定整体中介模型或进行事后检定，而忽略原先的中介假设。 |
2. (1) Write the mechanism proposition in one sentence: express it in mechanism language: “X influences Y through M”. 2. (1) 以一句话写出机制命题，并使用机制语言表达：「X 透过 M 影响 Y」。 (2) Explain what theoretical construct M represents? How does it carry X’s influence and transmit it to Y? (2) 说明 M 代表何种理论构念，以及它如何 影响X 的影响并影响 Y。 | 2. (1) Write M is important, or M relates to Y without a clear X→M→Y mechanism proposition. 2. (1) 仅写出 M 很重要或 M 与 Y 相关，却未明确提出 X→M→Y 的机制命题。 (2) Establish the variable in the model without providing a mechanism-based justification. (2) 将变数纳入模型，却未提供论证。 |
3. Specify the mediation model level (Micro vs. Macro) 3. 设定中介模型的层级 (微观与宏观)。 | 3. Only write M mediates. 3. 仅写出 M 具有中介作用。 |
Reference
MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. Annual Review of Psychology, 58, 593–614. https://doi.org/10.1146/annurev.psych.58.110405.085542
Rungtusanatham, M., Miller, J. W., & Boyer, K. K. (2014). Theorizing, testing, and concluding for mediation in SCM research: Tutorial and procedural recommendations. Journal of Operations Management, 32(3), 99–113. https://doi.org/10.1016/j.jom.2014.01.002
Shiau, W.-L., Shih, C.-H., Qureshi, I., Yuan, Y., & Teng, C.-I. (2026). Do matched and mismatched null hypothesis significance testing of mediators matter? An example of fintech continuance and guidelines for the future. Decision Support Systems, 211, 114764.
转载本文请联系原作者获取授权，同时请注明本文来自萧文龙科学网博客。
链接地址：https://wap.sciencenet.cn/blog-3444471-1557914.html?mobile=1
收藏
