Re:Any claim without a URI should be treated as suspicious
by TeMPOraL on Monday July 20, 2020 @12:49PM
(#23899590)
> 9/10 times the disagreement isn't data/citations but conclusions or criteria.
Or assumptions behind the cited data, and effect size.
What I've seen very often (particularly around sketchy nutrition discussions about "natural" things and "evil chemicals") is that quoted research seems to support a given conclusion, but under assumptions that do not apply to the argument at all, or significance is so low as to make the result useless.
Some study showing that a bunch of mice got cancer after getting injected with more artificial sweetener than you'd drink in a year does not show that artificial sweetener causes cancer in men. A study that says you get 0.003% greater chance of cancer from eating red meat every day has an effect size so small it's probably bullshit, but even if, 0.003% increase is not something worth caring about at all.
Or assumptions behind the cited data, and effect size.
What I've seen very often (particularly around sketchy nutrition discussions about "natural" things and "evil chemicals") is that quoted research seems to support a given conclusion, but under assumptions that do not apply to the argument at all, or significance is so low as to make the result useless.
Some study showing that a bunch of mice got cancer after getting injected with more artificial sweetener than you'd drink in a year does not show that artificial sweetener causes cancer in men. A study that says you get 0.003% greater chance of cancer from eating red meat every day has an effect size so small it's probably bullshit, but even if, 0.003% increase is not something worth caring about at all.