We thank the editor for the opportunity to offer our comments on this ambitious and interesting article investigating the causal relationships between "wealth" and "health". Because our views on what constitutes evidence for causal effects may differ somewhat from the authors' as expressed in this article (which we henceforth refer to as AHMcMR), we begin with a description of a simplified version of their central analyses eschewing causal language, and then summarize (also in a simplified version)the authors' basic causal conclusions. We then offer our view of causal inference and the meaning of direct and indirect causal effects. A simple counterexample is used to show the sense in which AHMcMR's causal conclusions about direct effects may be misleading. The basic issue is an excessive reliance on observed relationships among observed variables, without attendant discussion of justifying assumptions. We do not delve into other statistical issues, such as their use of hot-deck single imputationmethods to deal with missing data.
展开▼
