User Controls

What's Wrong with Social Science and How to Fix It: Reflections After Reading 2578 Papers

  1. #1
    netstat Tuskegee Airman
    https://fantasticanachronism.com/2020/09/11/whats-wrong-with-social-science-and-how-to-fix-it/

    I'm about halfway through and it's a pretty interesting read.
  2. #2
    Meikai Tuskegee Airman
    Bumping this because it'd be a shame to see it so quickly buried from latest threads by the barrage of inane threads Speedy Parker just shat the forums up with.

    I read the whole thing. I understood... a portion of what was said. It was written well enough that it made for a good read regardless of your technical understanding though, so that was nice. I know this just deals with social sciences, but reading about the causes of the "reproducability crisis" always makes me worry about science in general. It's not like your peers attempt to reproduce your results as a standard part of the review process in any field. How much bunk gets through? Is established canon now?
  3. #3
    Originally posted by Meikai Bumping this because it'd be a shame to see it so quickly buried from latest threads by the barrage of inane threads Speedy Parker just shat the forums up with.

    I read the whole thing. I understood… a portion of what was said. It was written well enough that it made for a good read regardless of your technical understanding though, so that was nice. I know this just deals with social sciences, but reading about the causes of the "reproducability crisis" always makes me worry about science in general. It's not like your peers attempt to reproduce your results as a standard part of the review process in any field. How much bunk gets through? Is established canon now?

    if you have spent more than 2.5minutes reading that then your doing it wrong because the author only spent that amount of time on those 2578 papers.
  4. #4
    Kev African Astronaut
    The problem with social science is that it is not a science at all despite the name, its up there with "religious science"

    only natural science is science as defined by law, social science should be called social studies. real testing is impossible to do when the experiment involves people, there are so many variables you can never take out of the equation, anecdotal evidence and the guessing game becomes the de facto standard.
  5. #5
    netstat Tuskegee Airman
    Originally posted by Meikai Bumping this because it'd be a shame to see it so quickly buried from latest threads by the barrage of inane threads Speedy Parker just shat the forums up with.

    I read the whole thing. I understood… a portion of what was said. It was written well enough that it made for a good read regardless of your technical understanding though, so that was nice. I know this just deals with social sciences, but reading about the causes of the "reproducability crisis" always makes me worry about science in general. It's not like your peers attempt to reproduce your results as a standard part of the review process in any field. How much bunk gets through? Is established canon now?

    Have to assume it's an issue throughout all science because scientists are often motivated by factors outside raw scientific curiosity, the peer review process is intriniscally fallible, etc., but physical sciences tend to be better because in general the controls, data collection and statistical standards are better. Physical sciences also have robust theoretical frameworks that social sciences lack. In physics you have a standard model that is well fleshed out theoretically and experimentally and this can be used to judge whether a new finding is likely to be true or not. Like when the higgs particle was discovered, we had a pretty good idea that was right because it had been predicted to exist 40 years ago. When neutrinos were observed ostensibly moving faster than light, we knew that was probably an abberation caused by some mistake in the experimental design or measurement, and it turned out to be exactly that. In social science you have a lot of one-off studies without such frameworks and it's almost certainly harder to judge whether a new or contradictory result is overturning previously entrenched nonsense, a valid result specific to a certain population and not necessarily in conflict with other results, or bunk that will consistently fail replication.

    So yeah, we have to have epistemic responsibility and not assume something must be correct because it was published in a physics journal and peer reviewed, but I assign higher confidence to a peer reviewed finding in the hard sciences than I do to almost anything in social science, even findings that replicate.
  6. #6
    Grylls Cum Looking Faggot [abrade this vocal tread-softly]
    Jimmy'll fix it
  7. #7
    Lanny motherfucker
    Originally posted by Meikai It's not like your peers attempt to reproduce your results as a standard part of the review process in any field. How much bunk gets through? Is established canon now?

    Checking proofs in mathematics is common during the review process. Not a science but at least it's something. Even in computer science where reproducing studies is usually relatively cheap, you'd be amazed how many results either can't be reproduced at all, or can't be generalized beyond the test data set. "AI research" in the last 10 years has made this a hundred times worse.

    Originally posted by Kev The problem with social science is that it is not a science at all despite the name, its up there with "religious science"

    only natural science is science as defined by law, social science should be called social studies. real testing is impossible to do when the experiment involves people, there are so many variables you can never take out of the equation, anecdotal evidence and the guessing game becomes the de facto standard.

    So what's the acceptable amount of "variables" that can be involved in an experiment? How many variables appear in studies in the social sciences verses natural sciences? How would you count these? Quine pretty much leveled the idea that you can test exactly one atomic hypothesis in any given experiment. I mean I won't say that experiment design isn't more difficult when you're studying human behavior than when you're studying, say, human biology, but there's nothing fundamentally different about the scientific process between "natural" and "social" sciences.
  8. #8
    Kev African Astronaut
    Originally posted by Lanny So what's the acceptable amount of "variables" that can be involved in an experiment? How many variables appear in studies in the social sciences verses natural sciences? How would you count these? Quine pretty much leveled the idea that you can test exactly one atomic hypothesis in any given experiment. I mean I won't say that experiment design isn't more difficult when you're studying human behavior than when you're studying, say, human biology, but there's nothing fundamentally different about the scientific process between "natural" and "social" sciences.

    Jesus i dont even know where to start with this.

    How do you do a controlled, double-blind social experiment on which policies create more crime/suicides without huge lawsuits pending your way?

    The reason you cant perform scientific experiments on human beings is because they have rights, you can do whatever you want to a piece of material until it bends, breaks or melts and record the resulting data. the social studies will always lack this fundamental rigor.

    it disturbs me when i see the abstract studies compared to natural sciences, it gives the false impression that it has the same authority and then corrupt fuckheads exploit that to push all sorts of crackpot garbage. cargo cult science comes to mind, as does conversion therapy.
Jump to Top