In my research work I have IV, DV and Mediator. According to Barron and Kenny's condition of mediation following steps should be met.
Step-1. I.V should significantly predict Mediator.
Step-2. Mediator should significantly predict DV
Step-3 I.V should significantly predict DV.
Step-4. When Mediator is added, I.V should predict DV less than that as in case of step 3.
In my research work a condition has arised where at Step-1, IV has not predicted mediator at all.
So can I proceed further for mediation analysis?
Dear Muhammad Imran Khan,
most researchers would agree, that Step 2 and 3 are necessary parametric assumptions for applying mediation analyses.
Step 2 tests the a-path, or put in other words, does the exposure affects the mediator?
If you have strong theory, especially for your a-path, that suggests a mediation process to take place, then you should re-consider non-paramatric assumptions on mediation, that could impact the significance of your a-path. These are:
1 - Assumption on confounding, e.g., see "sequential ignorability assumptions" Imai et al. (2010) or "unconfoundedness" assumptions Pearl (2014). In the case of the a-path, it could be the presence of one or more X-M-confounders that might suppress the X-M association.
2 - The presence of XM-interactions e.g., see Valeri & Vanderweele (2013). Traditional mediation analysis (e.g., Baron & Kenny approach), assumes the share of mediation to be equal across all levels of the mediator/exposure. However, in reality, XM-interactions are the rule rather than the exception. Maybe your a-path only holds for certain combinations of levels in X and M, thus yielding insignificance.
In past two decades there have been a lot of advances in the field of mediation analysis, especially from the causal inference or potential outcome framework. Generally, I would suggest concerning the non-parametric assumptions of mediation analysis as a first step.
For an application in Stata with longitudinal mediation and XM-interaction you can also see my publication Demirer et al. (2022)
Demirer I., Kühhirt M., Karbach U., Pfaff H. (2022) Does positive affect mediate the association of multimorbidity on depressive symptoms? Aging & Mental Health, 26(1): 65-76. https://doi.org/10.1080/13607863.2020.1870209
Imai, K., Keele, L., & Tingley, D. (2010). A general approach to causal mediation analysis. Psychological Methods, 15(4), 309–334. https://doi.org/10.1037/a0020761
Valeri, L., & VanderWeele, T. J. (2013). Mediation analysis allowing for exposure–mediator interactions and causal interpretation: Theoretical assumptions and implementation with SAS and SPSS macros.Psychological Methods, 18(2), 137–150. https://doi.org/10.1037/a0031034
Pearl, J. (2014). Interpretation and identification of causal mediation. Psychological Methods, 19(4), 459–481. https://doi.org/10.1037/a0036434
I will suggest you use the VAF approach which does not require such standards for the mediation.
Recommended readings
Article The Relationship of Emotional Exhaustion to Job Performance ...
Article Emotional exhaustion and its consequences: a comparative stu...
- Badges
- Science topic
- Similar topics
- Correction