Spatial_models_with_a_causal_lens
Published:
Understanding spatial models with causal inference: a non-mathematical primer =============================
Rohan Arambepola and I wrote this review on spatial methods and how the fit into a causal framework. I thought it was really useful at the time and still do.
However, we didn’t manage to get it published (a lot was going on and a lot is still going on) and I just don’t have the energy for it anymore. The paper is now probably a bit out of date, which makes it even harder to find time to update and submit. And preprint archives are a bit funny about review articles.
So I’m going to host it here. It can live here indefinitely. Maybe it will help some people think things through.
The core arguments are that there are a huge variety of spatial methods that do various things. One way to understand what they are doing is to view them through a causal lense. Then you see that most models are just “adjusting” for space in one way or another.
We use examples from malaria mapping and species distribution modelling. SDM in particular has a ton of fairly ad hoc methods and maybe this review will help people understand what those methods are trying to do.