But there are also people who believe they are fine alone but are negatively affected by it, and people who have lots of friends and interaction but nonetheless lack connection. People aren't very good at judging their own emotions.

Not having a better way to measure doesn't mean this measure is sufficient.

"People who self identify as lonely" is a different class of people from "people who are negatively affected". It's worth researching both groups. This study happens to be about the former.

> Not having a better way to measure doesn't mean this measure is sufficient.

It necessarily does mean that. Empiricists (such as scientists) must work with the tools with which they are equipped. Sure you're not going to get deductively-true results out of it (true for any scientific field), and certainly psychological findings are on the emphatically less-certain side of the scientific fields, but that doesn't imply that results aren't meaningful.

Granted, scientific reporting is so terrible the hedging the (good) scientists engage in to reflect this uncertainty invariably goes out the window. But c'est la vie.