5. One important aspect of the Potential Outcomes Framework (POF), is that treatment effects are estimated using information observed from different units, that is, the units in the treatment group (T=1) are not the same as the units in the control group (T=0). (This is the fundamental problem of causal inference). What two important assumptions, in your opinion, are required to make valid causal inferences?