> Double-blind experiments on this are probably impossible, but it's not like large population studies are totally worthless.
Large population studies in itself are fine, it's just the quoted study is worthless. Socially there's just so many differences between the two groups that it becomes almost impossible to validly compare these two.
A better study design would be to treat each region separately. And compare 'time spent outside' with health outcomes in each region. That would give a much more reliable and useful insight in if time spent outside equals better/less health. Bonus points if you also ask them if they regularly use sunscreen.
That data would actually be useful. That way you could compare if spending time outside is healthy in itself, or if it has to do with the sun. For example if the sun would be a net negative, you would find relatively more skin cancer in the groups in sunnier climates that spend a lot of time outside versus those in greyer regions. If the sun would be a net-benefit, you would see less skin cancer instead. And quite likely you would see that people who spend more time outside have better health outcomes no matter which region you are in, although the size of those groups may differ greatly depending on the climate.