Emil O W Kirkegaard

Do @AmirSariaslan style sibling control designs for your favorite predictions and let's see how they fare. As I recall, no effect of eg pop density on anything of interest in Amir studies.

If you have register data, you can also do a simple test using twins. Must be an association within MZs if causal. Probably best thing to start with.

Yep, so we can be sure there is genetic confounding in the cross sectional design, and also some C confounding.

In the study we controlled for confounding effects of socioeconomic factors, urbanization (pop. density), age and gender. We don't dismiss the effect of genetics and gene-environment interactions could be an interesting next step from this first exploratory study.

It was great that you controlled for those things. Given the approach, it is hard to make a better paper. I will list some of my thoughts about potential confounding issues:

SZ has heritabiliy = ~0.8, and small shared environment effect. And would a large part of that remaining variability be explained exactly by green things. Seems a priori not so likely.

The studies of @AmirSariaslan don't show evidence of environmental effects on mental disorders, in a neat sibling control design.