It's concerning that such a low sample was published in Nature. One can imagine a similarly underpowered study showing that a direct injection of HIV does not cause seroconversion, because the two or three samples happened to have CCR5-D32.
Why is it concerning? They don't make any strong conclusions and say Although the extraordinary high serum concentrations reached should compensate for this, physiological characteristics of the individual may have led to THCA-A not being incorporated into hair to a measurable extent. in the section titled Limitations of the study.
Admittedly the sample is extremely small, but your example doesn't match what they discuss. The claim "X does not cause Y (under any circumstances)" takes a lot more samples to be convincing than "X is not a reliable indicator of Y because it has false-positives", where a few examples of such false-positives already support their existence very well.
They are not trying to do statistics, they are just exhibiting a counter-example to a commonly accepted "truth".
If everybody thinks swan are white, you just have to show one black swan to prove than the theory "all swans are white" is false.
I think it's still interesting because the common assumption is that it shouldn't be possible at all. If you drop one apple you expect it to fall. If it doesn't then it's time to up your sample size and do further testing.
7 comments
[ 3.7 ms ] story [ 28.1 ms ] threadIf it's 99.9%, accurate, it still has a use.