Not sure what variance has to do with it? Thing is, you can always "patch" an X leads to Y theory when faced with opposing evidence by blaming confounding factors. But that is not very scientific. See https://en.wikipedia.org/wiki/Russell%27s_teapot
I'm just trying to say that very few things are actually that causal in big complex systems. I'm trying to say that such issues are very likely have many-many factors, and we see an aggregate effect.
And so in case of such aggregates arguing for-against them with very simple first-order reasoning is kind of useless. Because each piece of data (supporting or counter evidence) just explains a part of the big picture. So, without accounting for all data we can't really make general statements about what causes the aggregate effect that we see.
Yes, we can use "very simple first-order reasoning" even when we argue about very complex problems. If you say that "swans are white" it is enough for me to find a single black swan to prove your hypothesis wrong.
