Thank you for this. Everyone I've met who does research in social topics like this is endlessly frustrated by "well duh" reactions, especially on topics like social ills or discrimination.
Beyond the simple fact of empiricism, having a study show this is valuable on a lot of practical levels:
- Common knowledge is often wrong, so it's worth validating.
- Common knowledge says "worse", but this starts to quantify the difference. (Maybe 10 crummy friendships are worse than 2 good friendships, but are they worse than 1? How much worse? Where does zero friendships fall?)
- "Everybody knows" isn't and shouldn't be a basis for law or policy, so studies like this are citable evidence for the problem's existence and severity; "bad friendships are depressing" doesn't help us evaluate loneliness as a public health risk. (Similarly, replacing "those districts look stupid and have biased results" with a statistical measure of gerrymandering has been quite important in some recent court cases.)
- Simple studies provide standard methodology and reference data for analyses and future work. There's a long history of people testing complex hypotheses like "can we fight loneliness using this intervention to help people make more friends?" which fall apart because they skipped foundational work like "is number of friends even a predictor of loneliness?"
Honestly, I feel a bit relieved to see studies like this at the moment. With the replication crisis taking "well understood" results down like dominos, there's a lot to be said for going out and checking the really fundamental stuff.
