Specifically for exercise, I think there's a viscous cycle where once you become unhealthy, it becomes much harder to exercise. So, it's unclear which directio

Specifically for exercise, I think there's a viscous cycle where once you become unhealthy, it becomes much harder to exercise. So, it's unclear which direction the causation runs. How can you remove this self-selection bias and prove the direction of the causal relationship without paying healthy people to not exercise?