5 ms·

Yes, but the difference is subtle. The theory might still be true after all, but only for a particular subset. "We've tested the theory that 'swans are white',

Yes, but the difference is subtle. The theory might still be true after all, but only for a particular subset. "We've tested the theory that 'swans are white', and it turns out the swans in our test were all black". (The conclusion isn't that swans aren't white, it's that some swans are, while others are black.) (add.: the article mentions the 'WEIRD' category. Obviously the applicability/set range of a theory matters. If it only holds for "people who, between 2 and 5 pm on Sunday are walking around in Reading with a bowler hat.". Well, that's nice, but not very useful.)

ajuc 8y ago
This is common enough that it seems psychology as a whole is more art than science. It's like alchemy - they noticed some patterns and have some success exploiting them, but the abstractions and explanations they provide are mostly bullshit, so the predictive power is basically 0.
- anvandare 8y ago
  The problem is the Universe of which it speaks can not be reduced to a handful of beautiful and concise mathematical formulae. A group of humans (or even a single human) is more chaotic and complex in their behavior than a group of atoms ever could be. All Carbon-14 atoms are the same. No two humans, however, are.
  - ajuc 8y ago
    A group of humans is a group of atoms. Even a simple double pendulum is chaotic, yet we can describe its movement quantitatively. For me the failure of psychology are very similar to the failures of pre-enlightenment physics, chemistry, biology. Wrong abstractions, storytelling instead of formal models, unfalsifiable claims. My girlfriend was seeing a therapist, and the methods used were basically clairvoyance. Telling the patient "draw a tree", then interpreting the kind of tree as an indication of inherent characteristics of the patient. When she told me that - I've drawn a tree as well, and it turned out my tree was basically the same (the most common kind in our climate, in the most traditional drawing possible). I'm nothing like the description she got from the therapist and neither is she (but in a different way). Of course - she accepts the interpretation of that picture the same way people find matching details in their horoscopes. Psychology sounds to me a lot like bloodletting or voodoo programming, and scientific community should distance itself from people claiming its discoveries to be scientific, and its therapies to be scientifically-proven.
hn_throwaway_99 8y ago
But the issue is still major, because the results are nearly always announced with broad applicability. I.e. the headline was "The simple act of smiling can make you happier!" not "The simple act of smiling can make you happier if you are an 18-22 year old college student living in Virginia." Plus, a lot of these explanations seem really statistically dubious. The whole point of applying p-values is to determine the probability that your sample represents population at large. If you did a very poor job picking that sample and extrapolating to a larger population that wasn't warranted, that's a real problem in and of itself.
- anvandare 8y ago
  Absolutely. And it's sadly also correlated to the overall poor quality of science news reporting in our societies. One of my favorite PhD comics: http://phdcomics.com/comics/archive.php?comicid=1174 http://phdcomics.com/comics/archive.php?comicid=1174
- epistasis 8y ago
  That problem has nothing to do with science, though. If you are getting your "science" from newspaper articles, then it's not actually science. You're getting some over-worked, underpaid journalist's attempt at getting a catchy article.

ajuc 8y ago

This is common enough that it seems psychology as a whole is more art than science. It's like alchemy - they noticed some patterns and have some success exploiting them, but the abstractions and explanations they provide are mostly bullshit, so the predictive power is basically 0.

anvandare 8y ago
The problem is the Universe of which it speaks can not be reduced to a handful of beautiful and concise mathematical formulae. A group of humans (or even a single human) is more chaotic and complex in their behavior than a group of atoms ever could be. All Carbon-14 atoms are the same. No two humans, however, are.
- ajuc 8y ago
  A group of humans is a group of atoms. Even a simple double pendulum is chaotic, yet we can describe its movement quantitatively. For me the failure of psychology are very similar to the failures of pre-enlightenment physics, chemistry, biology. Wrong abstractions, storytelling instead of formal models, unfalsifiable claims. My girlfriend was seeing a therapist, and the methods used were basically clairvoyance. Telling the patient "draw a tree", then interpreting the kind of tree as an indication of inherent characteristics of the patient. When she told me that - I've drawn a tree as well, and it turned out my tree was basically the same (the most common kind in our climate, in the most traditional drawing possible). I'm nothing like the description she got from the therapist and neither is she (but in a different way). Of course - she accepts the interpretation of that picture the same way people find matching details in their horoscopes. Psychology sounds to me a lot like bloodletting or voodoo programming, and scientific community should distance itself from people claiming its discoveries to be scientific, and its therapies to be scientifically-proven.

hn_throwaway_99 8y ago

But the issue is still major, because the results are nearly always announced with broad applicability. I.e. the headline was "The simple act of smiling can make you happier!" not "The simple act of smiling can make you happier if you are an 18-22 year old college student living in Virginia." Plus, a lot of these explanations seem really statistically dubious. The whole point of applying p-values is to determine the probability that your sample represents population at large. If you did a very poor job picking that sample and extrapolating to a larger population that wasn't warranted, that's a real problem in and of itself.

anvandare 8y ago
Absolutely. And it's sadly also correlated to the overall poor quality of science news reporting in our societies. One of my favorite PhD comics: http://phdcomics.com/comics/archive.php?comicid=1174 http://phdcomics.com/comics/archive.php?comicid=1174
epistasis 8y ago
That problem has nothing to do with science, though. If you are getting your "science" from newspaper articles, then it's not actually science. You're getting some over-worked, underpaid journalist's attempt at getting a catchy article.