[mister_m]

Yeah I don't find their data convincing really. Ignoring that they have a biased sample to work from, and not knowing how the app works (do users elect when to tell their mood to the app? does the app ask? this could be a source of bias certainly), it is hard for me to look at these graphs and say wow. I'd imagine there are a number of confounding variables not being controlled for here. Life events that have nothing to do with sleep, etc? I'd be curious to see if that sort of thing was considered.

There is no statistical analysis happening here. This is just a presentation of what appear to me to be dubious looking graphs. I'm not convinced of any of this.

Additionally at the bottom of the page: "For the sleep analysis, we looked only at moods logged within 3 hours of the user waking up to reflect the impact of the user’s last sleep on the user’s mood."

That certainly introduces a lot of uncertainty.