And since it has been a while since this link has been posted in an HN comment, and since commenters here have mentioned that they haven't even read the fine submitted article, much less the underlying study publication, I'll repost here a link to a classic article by Peter Norvig (LISP hacker who is now Google's director of research) on how to interpret scientific studies and analyze study designs.
Suffice it to say that studies of human nutrition are HARD. A faultless study design would have to have dictatorial power to observe patient behavior over the long term, measuring exactly absolutely everything that a patient eats. The study design of the study reported in the submitted article is basically the best kind of study design for a longitudinal study that will pass institutional review for study of human subjects and not bankrupt the researchers with its expense. The researchers have made best current efforts to reduce flaws in the study design to the utmost extent possible, and acknowledge what they don't know (because they can't infer it from the study design) in their publications and comments to the press.
Re:More Evidence Against Vitamin Use
by scott_s on Wednesday October 12, 2011 @01:57PM
(#3103760)
I encourage everyone to read the conclusions. (And they're actual conclusions! Not just a summary! Yes, that makes me so excited as to warrant three exclamation marks!)
Re:More Evidence Against Vitamin Use
by kahawe on Thursday October 13, 2011 @05:06AM
(#3106653)
Excellent comment and conclusion and should be a mandatory read after the nyt article, really, because the newspaper story draws (or suggests) a conclusion that isn't really there.
http://www.sciencebasedmedicine.org/index.php/vitamins-and-m...
And since it has been a while since this link has been posted in an HN comment, and since commenters here have mentioned that they haven't even read the fine submitted article, much less the underlying study publication, I'll repost here a link to a classic article by Peter Norvig (LISP hacker who is now Google's director of research) on how to interpret scientific studies and analyze study designs.
http://norvig.com/experiment-design.html
Suffice it to say that studies of human nutrition are HARD. A faultless study design would have to have dictatorial power to observe patient behavior over the long term, measuring exactly absolutely everything that a patient eats. The study design of the study reported in the submitted article is basically the best kind of study design for a longitudinal study that will pass institutional review for study of human subjects and not bankrupt the researchers with its expense. The researchers have made best current efforts to reduce flaws in the study design to the utmost extent possible, and acknowledge what they don't know (because they can't infer it from the study design) in their publications and comments to the press.