Journal article
A Tutorial on Aligning Questions and Models in Causal Mediation Analysis
The Journal of experimental education
02/2026
DOI: 10.1080/00220973.2026.2612929
Abstract
This tutorial offers practical guidance on causal mediation analysis for observational studies, emphasizing the critical alignment of research questions, causal estimands, and statistical models. Through educational research examples, the tutorial addresses analytical challenges, such as distinguishing between full and partial mediation, controlling for confounders, and avoiding collider bias. The tutorial demonstrates how to employ appropriate statistical methods (e.g., path analysis or regression modeling), contrasting correct analytical strategies with common misspecifications arising from misaligned estimands or statistical approaches. The tutorial illustrates the decomposition of a total effect into direct and indirect components to clarify causal mechanisms. Directed acyclic graphs (DAGs) are introduced as tools for visualizing causal assumptions, guiding variable selection, and ensuring alignment across the research question, estimand, and analysis.
Details
- Title: Subtitle
- A Tutorial on Aligning Questions and Models in Causal Mediation Analysis
- Creators
- Juyoung Jung - University of IowaAriel M. Aloe - University of Iowa
- Resource Type
- Journal article
- Publication Details
- The Journal of experimental education
- DOI
- 10.1080/00220973.2026.2612929
- ISSN
- 0022-0973
- eISSN
- 1940-0683
- Publisher
- Taylor & Francis
- Number of pages
- 21
- Language
- English
- Electronic publication date
- 02/2026
- Academic Unit
- Education Administration; Psychological and Quantitative Foundations
- Record Identifier
- 9985139297202771
Metrics
1 Record Views