Humans are also very, very bad at knowing what makes them happy. If a healthy person tells me they have one friend they see once a month and no other social outlets and they wouldn't want anything else, 99% of the time I'm going to assume they don't know themselves well enough to know that more relationships than that would make them happy.
Do these studies actually show that there aren't people who, for lack of a better word, are non-responders? Imagine if 95% of people require relationships to be happy and 5% don't. I would expect most studies to show the correlation exists, but it doesn't mean that 5% doesn't exist. What kind of upper bound do these studies provide here?
These studies never make sense because there's no way to objectively measure happiness, especially not in a way where you can compare two people's self-reported levels of "happiness". The word happiness itself is already is an over-loaded term that everyone has their own definition of. (At best you can measure proxy outcomes like measuring health based on relationships, but there are so many confounding variables it's hard to narrow it down.)