Causal inference vs "the usual statistics"

Do you think that causal inference methodology should be used in all circumstances?

Is there any particular situation you have in mind, because on its own the question is rather broad and open to interpretation.

To start, what would you consider to be causal inference methodology? Are we talking formal methodology (i.e. DAGs, counterfactuals etc.) or would you consider any thinking about a well informed study/analysis plan to be causal methodology?

Second, what is the context where we are talking about whether to apply causal methods? Are we looking to do causal research or prediction? Cause even in prediction a well-thought out causal structure underlying the variables you are considering to use in your models could be an asset I think?

1 Like