> I'd guess that you can equate "screen time" to "smartphone use" in most of the population but here. Similar to what people are using the screen for.

I'm fairly confident this is not a safe assumption.

> To include so many variations of screen would complicate a lot the study

How so, though?

> without a prior about why it should matter, you are probably safer starting with something more homogeneous.

That's not very scientific, is it? There's no prior as to why it shouldn't matter, which is what the authors seem to think.