@karelbrits @fionag11
IIRC the research looked like it was originally based in the US cities.

I get that that there may be places where the output of the model doesn't entirely match another place. But when they do a match they then give an explanation of the change. I don't get why they can't do an explanation of the change if there's no match. They must have the data, it's what they used for matching.

It doesn't really explain the matching method well either so, what variables are matched?