If you did read the paper, you'd have noticed that people who slept more regularly in this research were also wealthier, and more physically active. They did say they adjusted for these variables, but who knows. Maybe they didn't adjust them enough, maybe there are more covariates they didn't consider.

I don't know if they can have controls for "specific activity that changes their sleep schedule".

In the past few years I went from having relaxed 11AM daily meetings to rigid 8AM meetings, and my sleep has suffered immensely. But nothing else in my life has changed, so it wouldn't show up in my socioeconomic data.