333
edits
Changes
Jump to navigation
Jump to search
Line 425:
Line 425: + +
Talk:Mystery:Young Hollywood Breast Cancer Victims (view source)
Revision as of 06:26, May 20, 2008
, 06:26, May 20, 2008→Deletion: clarify some epidemiological points
Humblpi, what evidence would you consider significant? If 10x the expected cancer rate is not alarming to you, what would you require? 100x? There isn't going to be a formal, double blind study, nor does there have to be if the expected incidence is exceeded by a high enough factor. Demanding a formal study is like the tobacco companies demanding more, and more, and more studies before admitting the obvious, as the tobacco companies did for decades as they continued to cause the deaths of millions.--[[User:Aschlafly|Aschlafly]] 22:50, 19 May 2008 (EDT)
Humblpi, what evidence would you consider significant? If 10x the expected cancer rate is not alarming to you, what would you require? 100x? There isn't going to be a formal, double blind study, nor does there have to be if the expected incidence is exceeded by a high enough factor. Demanding a formal study is like the tobacco companies demanding more, and more, and more studies before admitting the obvious, as the tobacco companies did for decades as they continued to cause the deaths of millions.--[[User:Aschlafly|Aschlafly]] 22:50, 19 May 2008 (EDT)
:I think you misunderstand my objections. Of course 10x the expected cancer rate would be alarming. But the quality of the evidence is judged not on the significance (or magnitude) of the results but on the reliability of the methods used to gather those results, and I'm afraid you have no clear methods at all. I am referring to proper epidemiological evidence, gathered by proper methods and properly analysed. And of course a randomised double-blind study would not be appropriate - that's not the right study design for this sort of question. What is required is a case-control study - exactly the sort of study carried out by Richard Doll, indeed, to demonstrate the tobacco/lung cancer link. [[User:Humblpi|Humblpi]] 02:26, 20 May 2008 (EDT)
:Andy, it wouldn't be possible to do a double-blind study. You can't randomly assign anyone to either be a rock star or to get a cancer diagnosis. I can't speak for anyone else I suppose, but I don't think anyone is asking for a formal study. The problem that keeps being raised is the estimate of the denominator, which appears (to me and apparently to others) to be a substantial underestimate; specifically, you haven't explained how you got the number. This is the point of Method sections in research papers - to explain how you did what it is that you did. Without that explanation, no matter how solid your case actually is, you leave yourself open to doubts. [[User:Murray|Murray]] 22:57, 19 May 2008 (EDT)
:Andy, it wouldn't be possible to do a double-blind study. You can't randomly assign anyone to either be a rock star or to get a cancer diagnosis. I can't speak for anyone else I suppose, but I don't think anyone is asking for a formal study. The problem that keeps being raised is the estimate of the denominator, which appears (to me and apparently to others) to be a substantial underestimate; specifically, you haven't explained how you got the number. This is the point of Method sections in research papers - to explain how you did what it is that you did. Without that explanation, no matter how solid your case actually is, you leave yourself open to doubts. [[User:Murray|Murray]] 22:57, 19 May 2008 (EDT)