Tue, Sep-07-10, 14:07
|
Senior Member
Posts: 606
|
|
Plan: PP
Stats: 210/170/170
BF:
Progress: 100%
Location: NJ
|
|
What makes this study meaningless:
1. Food frequency questionnaires. People filled out a form every few years. Very crude method of gauging what people ate.
2. This was a statistical analysis of tabular data. NOT a controlled study, not a clinical study, NOT an examination of actual people. It wasn't an experiment.
3. The risk ratios (hazard ratios) are MINISCULE. TINY. IF there is any effect at all, it is BARELY MEASURABLE, even if you have very good, very precise data. Which we don't. See #1.
Remember, data analysis like these are ONLY useful for generating some ideas worth following up on with more rigorous experiments. Most epidemiologists (legitimate ones, anyway) ordinarily don't consider an RR of less than 3.0 or worth looking into. And even then, it's still a tenuous, weak, relationhip at best. The RRs in this study were all like 1.25 or 0.95. In real world terms, basically nothing.
If numbers like these were presented to show, say, that eating saturated fat makes you live longer, it would be immediately skewered in the press, for the above reasons. But since the effect here is in vogue and politically correct, it gets a free pass. Nobody argues the flimsy, iffy numbers.
|