Extract

1 Introduction

I am grateful to Auerbach et al. (2023) and Leung (2023) for their insightful comments. They have highlighted important perspectives that were overlooked in the article and delineated the scope of my arguments. In this rejoinder, I elaborate on some of the ideas presented in the article in light of the comments, and highlight where I agree and disagree with the discussants.

2 Interpretability of exposure effects

2.1 The necessity of correctly specified exposures

Auerbach et al. (2023) call attention to the fact that expected exposure effects can be difficult to interpret, and they provide a carefully constructed example to illustrate what can go wrong. The discussants argue that expected exposure effects are not automatically useful and that correctly specified exposures are necessary for making a study relevant to policy. They write: ‘we do not see how misspecified exposure effects can generally be useful for policy making’. The caveat indicated by ‘generally’ in the quote refers to two exceptions where the discussants find that misspecified exposures potentially could be useful, but, as they note, these exceptions are edge cases.

You do not currently have access to this article.