The only social sciences research that isn't a sham is the kind that can be used to consistently predict non-obvious things or consistently modify otherwise non-modifiable things.
They should all be called social studies and not “social sciences”. As you’ve pointed out, they’re not real sciences because the requirement for being a science is having a repeatable theoretical model.
To be fair, they’re not completely useless either. They just need to graduate from being an astrology into an astronomy, but I’m not sure how that’s possible without a near perfect world simulation.
I don't know of any such research.