So, the problem with these types of arguments is the evidence is often obscured by shell games with labels, or ignoring the elephant in the room.
My research sort of straddles statistics, epidemiology, neurogenetics, and social sciences, and my impression is there's a great deal of stuff that doesn't become interesting to people until it involves social sciences. I say this from the perspective of someone who often wants the math and whatnot to be more appealing to people when it's not.
You can take some model, or method, or advance, and if it sits somewhere in a wetlab or physics building, no one cares. But if you make it about relationships, or money, or happiness, or children's social development, it becomes interesting.
The value is so built into the fabric of these things we don't even recognize it half the time. It becomes another physics grant, or engineering grant, or medicine grant. They're social science in nature, but god forbid you call it that because it's seen as soft or irrelevant even though that's the thing people are actually most interested in. People don't give a damn about memory models in bare-metal programming, or electrical engineering, but they do care about being able to talk to their relatives on the other side of the globe.
The other thing that always strikes me as odd about these discussions and decisions is we as people are, well social beings. So do you want to leave the study of our social behavior and experiences to ... what exactly? Religious theory? Astrology? What else is there than scientific study of psychosocial phenomena? Are you saying that social phenomena shouldn't be studied at all? That it should be studied nonscientifically?
Within the social sciences there's a well-known phenomenon of people coming in from elsewhere and thinking they can solve all the problems because they're physicists, or chemists, or whatever, and that you can just slap whatever paradigm works in that to study psychological or socioeconomic phenomena. It's seen as naive, and it inevitably fails because it doesn't scale up to the population level, or doesn't address the difficulties of the real-world phenomenon as it is, or whatever.
There's also the problem that some of the biggest contributions of social science aren't what people think they are. Modern meta-analysis and preregistered replication research both came from psychological research for example (meta-analysis from clinical and educational psychology), and I doubt people realize that. But it's widespread in biomedical research, and should probably be applied more widely in many other fields. In certain ways it's more scientifically rigorous than what happens in a lot of biophysical science. But it goes against the stereotype of what "psychosocial research" is.