While you are right that it is naive, or maybe more exactly, intelectually dishonest to use abstracts (and not full papers or better yet the totality of evidence) to arrive at a strong conclusion you should forget first-hand accounts. There is no such thing like a reliable first hand account. Everyone is subject to cognitive and statistical bias. Everyone is subject to the placebo effect. Everyone is human. And we're all slaves to statistics (do you think the bull curve, including the SD, applies to everyone but you and your measurements?).It's just extemely naive on your part to be so trusting of a scientific abstract, while so untrusting of all first-hand accounts, esp. from people who have proven themselves to be trustworthy.
Thus even well-controlled case-reports, i.e. a level of evidence much higher than anecdotes, are inherently unreliable. Sometimes we're forced to use such weak evidence, but preferably, no.
Instead of saying long winded stuff like this why don't you write a dissertation on the reasons these conflicts exist and then proceed to prove point for point why personal accounts and biomarkers of several case reports are inferior to 'statistical data'. Personally I never trusted statistics fully. There is far too much randomness of phenomenon, variation in individual responses and genetic determinants to trust an abstraction called statistics. This is why prozac made many people worse than they were before using it. They took a statistical abstraction (I.E that such and such a cross section of the populace responded chemically well to the drug) and applied it to the whole human race. That's why they created many other drugs/doses to account for chemical variations. Not everyone is the same. But that is why I prefer natural supplements and a natural lifestyle. Nature, for the most part, caters to our needs, regardless of our genetic variability.
