02.Causation04b.Mediation and moderation analyses - sporedata/researchdesigneR GitHub Wiki
1. Use cases: in which situations should I use this method?
- Mediation analysis should be used when a variable is in the causal pathway between a risk/intervention and the outcome. For example, blood loss leads to transfusion, and the resulting outcomes could be postoperative infection and renal failure. In this case, transfusion is a mediating variable between blood loss and the outcomes.
- Interaction happens when two risk factors being simultaneously present has an effect that goes beyond just the sum of their individual effects. Using a fictitious example, diabetes might increase postoperative mortality by 1%, while chronic renal failure adds another 1%, but having both concomitantly could increase the risk by 5%.
An alternative are the so-called logic regression models (not to be confused with logistic regression), which investigates Boolean combinations of multiple variables. See some related publications.
2. Input: what kind of data does the method require?
- Datasets with an outcome variable and at least two risk factors
3. Algorithm: how does the method work?
Model mechanics
Reporting guidelines for Methods
Data science packages
Suggested companion methods
Learning materials
- Books
- Articles
- Evaluating mediation and moderation effects in school psychology: A presentation of methods and review of current practice
- Statistical Mediation Analysis for Models with a Binary Mediator and a Binary Outcome: the Differences Between Causal and Traditional Mediation Analysis
- A tutorial for conducting causal mediation analysis with the twangMediation package