5 ms·

I am especially bothered by one sentence: "Prior to my introduction to the world of low-carbohydrate diet, I hadn't paid too much attention to nutritional scie

I am especially bothered by one sentence: "Prior to my introduction to the world of low-carbohydrate diet, I hadn't paid too much attention to nutritional science. I worked on biophysics, where I formed the opinion that medical science was mostly garbage." It's hard for me to take someone seriously who once believed that medical science is 'mostly garbage.'

Ardit20 17y ago
Why?
- pierrefar 17y ago
  Because he clearly is biased against "medicine" and then finds a book that fits his bias and is trumpeting it as the eternal truth. He's committing the same offense he speaks out against in the same article.
  - jerf 17y ago
    Is he "biased against medicine", or is he "biased in favor of five-sigma+ results"? Does medicine use 95% confidence intervals because it is actually sufficient to base a medical science off of, or do they use 95% confidence interval because it is all they can get? The problem is, the universe doesn't care if all you can get is a 95% result. And he is right that at least some people in the softer sciences don't understand their confidence intervals in the slightest. That crack he made about standard deviations being computed for non-Gaussian distributions won't resonate with you unless you have a lot of statistical training, but it a damning criticism; doing that sort of thing is basically proof your entire analysis isn't worth the paper it's written on, in proportion to the deviation from Gaussianess. I've been witness to the training procedure on experimental stats for undergrad biologists, which consists mostly of endless hammering in of the p-value as mantra, with understanding neither asked for nor given (unsurprising, as given the math prereqs up to that point the true meaning of the p-value can't really be taught). At the time I wouldn't have realized how wrong it is to apply p-value analysis to a bi-valued distribution, but I bet they were asked to. I know there's a grad-level course on the topic for them, but I fear it may be the blind leading the blind there. I'm less willing to write off medicine as a whole. It has had some undeniable success in some areas (but it should be observed that weight management with much less than stomach stapling hasn't really been one of them). I'm playing a bit of devil's advocate here, because I think you may not have given the article a fair reading after you flipped the idiot flag on the author.
    - qwerx 17y ago
      Having worked in mechanical engineering and biomechanics, I also developed a certain skepticism with results in the "softer" (or "less controllable"?) sciences. In biomechanics I often saw complete scattershot X-Y correlations, which after enough statistical dancing, were revealed to be "statistically significant" enough to be publishable, though they obviously had very little predictive power. I personally know how hard it is to isolate all factors and get a decent correlation, and that's just the way it is in medical science, but that's what gives me skepticism. In the "hard, controlled" sciences, generally if you can't see an obvious correlation by eye you already throw it out. Statistics are used to quantify just how powerful the correlation is, not to prove there is a correlation at all.
      - larryfreeman 17y ago
        I've never understood this viewpoint. Would a mathematician be ok for dismissing physics since a physics experiment is not as rigorous as a mathematical proof? Calling "medicine" garbage because it is not as precise as physics is so silly that I don't really know how to respond. I am imagining a physicist who just had a massive hard attack saying to his doctor: "I'm sorry, please let me die, I don't believe your technique is based on sufficiently precise data."
jnorthrop 17y ago
Your quote is a bit out of context. He goes on to explain, "As a physicist, if I get an correlation coefficient, R2 < 0.9997 in an experiment, I would consider that a poor result. A nutritional researcher working with human patients cannot even dream of achieving the degree of control or characterization I can, and their data are overloaded with spurious noise." He is giving his opinion on why the science is "garbage" and explicitly states that "This isn't largely the fault of the scientists involved"
- 3pt14159 17y ago
  This part bothered me a lot. R2 of 0.9997! Thats ridiculously accurate. Unless he is measuring meters with an electron microscope I don't think he could ever get that close to accuracy.

Ardit20 17y ago

Why?

pierrefar 17y ago
Because he clearly is biased against "medicine" and then finds a book that fits his bias and is trumpeting it as the eternal truth. He's committing the same offense he speaks out against in the same article.
- jerf 17y ago
  Is he "biased against medicine", or is he "biased in favor of five-sigma+ results"? Does medicine use 95% confidence intervals because it is actually sufficient to base a medical science off of, or do they use 95% confidence interval because it is all they can get? The problem is, the universe doesn't care if all you can get is a 95% result. And he is right that at least some people in the softer sciences don't understand their confidence intervals in the slightest. That crack he made about standard deviations being computed for non-Gaussian distributions won't resonate with you unless you have a lot of statistical training, but it a damning criticism; doing that sort of thing is basically proof your entire analysis isn't worth the paper it's written on, in proportion to the deviation from Gaussianess. I've been witness to the training procedure on experimental stats for undergrad biologists, which consists mostly of endless hammering in of the p-value as mantra, with understanding neither asked for nor given (unsurprising, as given the math prereqs up to that point the true meaning of the p-value can't really be taught). At the time I wouldn't have realized how wrong it is to apply p-value analysis to a bi-valued distribution, but I bet they were asked to. I know there's a grad-level course on the topic for them, but I fear it may be the blind leading the blind there. I'm less willing to write off medicine as a whole. It has had some undeniable success in some areas (but it should be observed that weight management with much less than stomach stapling hasn't really been one of them). I'm playing a bit of devil's advocate here, because I think you may not have given the article a fair reading after you flipped the idiot flag on the author.
  - qwerx 17y ago
    Having worked in mechanical engineering and biomechanics, I also developed a certain skepticism with results in the "softer" (or "less controllable"?) sciences. In biomechanics I often saw complete scattershot X-Y correlations, which after enough statistical dancing, were revealed to be "statistically significant" enough to be publishable, though they obviously had very little predictive power. I personally know how hard it is to isolate all factors and get a decent correlation, and that's just the way it is in medical science, but that's what gives me skepticism. In the "hard, controlled" sciences, generally if you can't see an obvious correlation by eye you already throw it out. Statistics are used to quantify just how powerful the correlation is, not to prove there is a correlation at all.
    - larryfreeman 17y ago
      I've never understood this viewpoint. Would a mathematician be ok for dismissing physics since a physics experiment is not as rigorous as a mathematical proof? Calling "medicine" garbage because it is not as precise as physics is so silly that I don't really know how to respond. I am imagining a physicist who just had a massive hard attack saying to his doctor: "I'm sorry, please let me die, I don't believe your technique is based on sufficiently precise data."

jnorthrop 17y ago

Your quote is a bit out of context. He goes on to explain, "As a physicist, if I get an correlation coefficient, R2 < 0.9997 in an experiment, I would consider that a poor result. A nutritional researcher working with human patients cannot even dream of achieving the degree of control or characterization I can, and their data are overloaded with spurious noise." He is giving his opinion on why the science is "garbage" and explicitly states that "This isn't largely the fault of the scientists involved"

3pt14159 17y ago
This part bothered me a lot. R2 of 0.9997! Thats ridiculously accurate. Unless he is measuring meters with an electron microscope I don't think he could ever get that close to accuracy.