Perhaps I'm not making my point clear enough, or it is clear but so simplistic that one skips past it: we can't test whether things would have been better or worse under Clinton because we can't run the experiment of Clinton being president for the past three years. And the same can be said for any other hypotheses that would require a working time machine to test.
I'd still dispute your point.
Yes, you "can't test whether things would have been better or worse under Clinton because we can't run the experiment of Clinton being president for the past three years".
Consider,
you can't measure an electron's momentum and position accurately simultaneously. But that still doesn't preclude you from building fantastic microprocessors relying on aggregate properties of those electrons.
you can't ever prove anything in physics with the degree of certainty like you can in Mathematics. We're still able to hurl rockets to space pretty accurately.
What you can't measure is not connected with reality, and not relevant for reality. You start with what you *can* measure and that forms the basis of your scientific reality.
This basic, simple truth is what entirety of science rests on.
When you focus on "irrelevant" - a.k.a. non-measureable - issues (e.g. whether we can build a time machine and test something in a very specific way) you are espousing the right-wing dogma of the tribal epistemology (
https://www.vox.com/policy-and-politics/2017/3/22/14762030/donald-trump-tribal-epistemology - can't find a non-left-wing source), either knowingly or unknowingly, that is so prevalent among the Trump supporters today. The same thought process leads to denying other scientific truths and realities in many different topics. If you can dispute the existence of *any* objective processes, truths etc - then it becomes easy to play footsie with truth and establish false equivalences.
The scientific method is to start with the best possible observation/test you have, form conclusion, and then update that conclusion when you have better tests and observations available.