It's nice that the article controls for many reasonable variables. Of course, there could be other confounding factors (like the employer's trust, perhaps loc

It's nice that the article controls for many reasonable variables. Of course, there could be other confounding factors (like the employer's trust, perhaps location, etc.), but those may be hard to quantify. I think, it's interesting to think about the reverse. "Developers that earn more are more likely to work from home". I'm curious whether this still holds for the very best payed and very worst payed? There could be other effects kicking in the tails of the distribution?