According to another commenter here, the research that spawned the WSJ article said Instagram made 41% feel better, 31% worse, and no impact for the other 28%.

I haven't found the original material to verify, but if this is accurate, seems like the net outcome argument works in Instagram's favor.

It doesn't necessarily work in Instagram's favor. If the 41% and 31% had a baseline of both feeling "ok" then you took 72% who had a baseline that is acceptable and polarized them into groups who felt better or worse. Is it more favorable to have more people who have a stable baseline or to have one group have a good baseline and another similar sized group with a bad baseline?

I don't know, but at the very least the article title is grossly misleading.

Not an ethicist, but I don't think this is how ethical evaluations work in most frameworks.

It could be one input into a "utilitarian" model, which in my opinion describes all of the even remotely viable ethical frameworks.

Definitely, but still saying that "some people benefit from it" is difficult to agree/challenge. Numbers at least give some information.