Can't say enough good things about this book. I recently co-wrote a primer on causal inference for anthropologists and this book addresses the exact tension we struggled with—how to grapple with exposures that are deemed not "manipulable" while incorporating the advances of modern causal inference.