How Should We Think about Replicating Observational Studies? A Reply to Fowler and Montagnes