Abstract
We thank Mokhayeri and Mansournia for their considered comments relating to our recent article, and we are happy to clarify some points raised.
As with all approaches, there are advantages and disadvantages of the methods we used here, and our paper acknowledges some key limitations related to our approach. Although we do not explicitly define each of the assumptions underpinning our methods, we reference the seminal paper that delineates these assumptions. Our paper acknowledges the assumption of no unmeasured confounding of the mediator–outcome relationship, and indeed we did not identify any variables likely to confound this relationship.
As with all approaches, there are advantages and disadvantages of the methods we used here, and our paper acknowledges some key limitations related to our approach. Although we do not explicitly define each of the assumptions underpinning our methods, we reference the seminal paper that delineates these assumptions. Our paper acknowledges the assumption of no unmeasured confounding of the mediator–outcome relationship, and indeed we did not identify any variables likely to confound this relationship.
| Original language | English |
|---|---|
| Pages (from-to) | 1731-1732 |
| Number of pages | 2 |
| Journal | International journal of epidemiology |
| Volume | 48 |
| Issue number | 5 |
| Early online date | 9 Jun 2019 |
| DOIs | |
| Publication status | Published - 1 Oct 2019 |
| Externally published | Yes |
Keywords
- mediation analysis
- methodological approaches
- parametric models
- Causal mediation
- measurement error
- bias