Respectfully, I disagree. An example I'd like to set forth as particularly relevant is work which falls in fields such as the medical humanities, STS (science, technology, and society), and similar fields which utilize a methodology called actor-network theory. This sort of work falls heavily in the realm of social science or humanities as it draws on philosophy, sociology, but also ultimately and intimately STEM fields and subjects. This sort of work often analyzes the role of technology, and more importantly the uptake of technology, /scientific advancement/, the network of actors which influence how, when, etc this happens. In a general sense these fields tend to yield work which elucidates the context of technology and science, how this context and actors in it influence or have influenced technology.
For example, Bruno Latour wrote a book more or less on the temporally specific factors which lead to the acceptance of the vaccination in scientific communities and in society. Essentially, what I'm saying is something similar or parallel to Thomas Kuhn. Scientific advancement and technological advancement in society and the acceptance of a research achievement as /scientific advancement/ is less about the literal research achievement/advancement and the science, but about social context and temporal specificity. I.e. Kuhn's "paradigm shift".
Annemarie Mol's work, who is both an MD and non-STEM PhD, often highlights the very unscientificness of medical science and technology in practice. Though it's not really her goal, I believe, her examinations illuminate how the guidelines of medical science and the abstract understandings of specific diseases are incongruous with the reality of disease pathology in real patients, even when looked at on a large scale.
I don't see how humanities, social science, etc are somehow less worthy of PhD study. I don't experience STEM as somehow much more difficult to self-study, nor doctoral programs as strictly about knowledge, understanding, nor the validation of general information within a field. Doctoral programs are often about reaching toward mastery of a specific subject to the point of researching it, and there's a lot of networking, career-oriented stuff, and teaching involved. There's a somewhat apprenticeship-like quality to it. Then a committee validates that this research contributes something to the field in a general sense, and also meets bare minimum ethical standards in research, and other similar standards. They ask questions to validate that the candidate has knowledge and can answer questions about whatever highly-specific subject they're working with. It's not an end-all degree; it's one which certifies in whatever specific field that the recipient can master required comps or quals at a minimum standard and then plan, execute, and draw conclusions about research in written form done at a bare minimum standard required for peer-reviewed publishing. In pure math that can mean proving something and then explaining it, but actually most pure math phds I know completed their degrees in less time than say... I think 10-12 years is average or expected fast degree completion for doctoral candidates in Anthropology at my institution. In pure Math it's something like 6-7. Though all that varies by individual program.
I think it's also relevant to note that I think you're a bit jaded about social science research. It's like anything, there's useless, thoughtless, and downright unethical research, and there's also research which the researchers are attempting to engage in best practices and the most accurate understandings of their research. Like, if you look at something like quantitative Sociological research, it's often done via online self-report surveys. No less, sociology phds now are taught to methodologically consider the potential or downright likelihood of the unreliable self-reporter, in addition to the ways that the framing of a survey question influences the respondent's answer. The social sciences methodologically acknowledge these issues because it is a big part of these fields in itself that this would be a major research concern. So I would expect an ethical researcher to frame the results of a quantitative research study utilizing self reports in the language of self reporting. I.e. "X% of some sub group of respondents reported they experienced feeling cold in the winter time". The methodology of social science fields is also to lay that stuff out on the table, often recognizing the culturally-specific aspects of the research, and/or bias in the research. In the social sciences, people are taught not to create hypothesis and expectation. I'm sure that's something people do, but research questions can be rather open-ended.
Conversely, when you consider human research in the life sciences, there's an issue with the quality of research results. There's a well-documented risk of confirmation bias, observer-expectancy effect, and other experimenter effects, the most important for me being how researchers can unknowingly give cues to human and animal subjects which often results in the subjects confirmation their experimental expectations. The same issue with the framing of research is a major concern in areas like medical research where it can majorly impact results.
I guess I'm saying... I'm not trying to negg STEM research or hold up other fields. More, I'm trying to point out that STEM research and its uptake in scientific communities is not objective, nor immune to the conditions which influence social sciences. STEM research is influenced heavily by socio-political factors, and there is plenty of less than ethically performed and analyzed STEM research; there's STEM research which is only about confirming the researcher's view. STEM fields have all these ills too.