Earlier quoted context omitted.
Actually, they didn't touch the lights at all, they only measured the lights that people already were exposed to. They write that "an interventional method, such as a randomized controlled trial using a LAN exposure, is not possible because of ethical concerns".
> Actually, they didn't touch the lights at all, they only measured the lights that people already were exposed to. But isn't it a flaw of the study? It only shows correlation without causality. What if being exposed to light is correlated to some other factor that is responsible for increased depression? In this case, decreasing the amount of light won't decrease depression.
But yes, it’s possible that some other factor is responsible, such as living next to a busy road with lots of street lights.