Here’s the deal: those markers are proxies for health. But there have been numerous cases (Alzheimers research is particularly full of them, but it’s widespread in many biological systems) where changing the proxy marker does nothing to change the underlying condition. Causality doesn’t work like that. You think it’s A causes B, but in fact it’s Z causes A by one causal chain, and Z causes B by (potentially) another. So there’s your guy’s first fallacy.
The second is to conduct multiple trials in parallel on the same subject. Then, even if a change in the proxy variable actually means a change in health, you have no reliable way to untangle which factor or combination of factors was responsible for the change.
Third, a sample size of 1 or 2 is fucking stupid. It makes it impossible to tell if any measurements collected are releveant, or even repeatable. It also makes it impossible to tell if any fluctuation was random or actually caused by something you are trying to measure. And if you’re trying to measure an actual effect, you need a control group to compare it against. He has none.
Source: I was educated as a statistician and my focus was on experimental design in bio-science and pharma.
So, even assuming good will on this guy’s part, he’s a hobbyist doing junk science. If he really cared about helping humankind, he should have asked someone who knows how to do experiments to advise him on how to set up his protocols.
Here’s the deal: those markers are proxies for health. But there have been numerous cases (Alzheimers research is particularly full of them, but it’s widespread in many biological systems) where changing the proxy marker does nothing to change the underlying condition. Causality doesn’t work like that. You think it’s A causes B, but in fact it’s Z causes A by one causal chain, and Z causes B by (potentially) another. So there’s your guy’s first fallacy.
The second is to conduct multiple trials in parallel on the same subject. Then, even if a change in the proxy variable actually means a change in health, you have no reliable way to untangle which factor or combination of factors was responsible for the change.
Third, a sample size of 1 or 2 is fucking stupid. It makes it impossible to tell if any measurements collected are releveant, or even repeatable. It also makes it impossible to tell if any fluctuation was random or actually caused by something you are trying to measure. And if you’re trying to measure an actual effect, you need a control group to compare it against. He has none.
Source: I was educated as a statistician and my focus was on experimental design in bio-science and pharma.
So, even assuming good will on this guy’s part, he’s a hobbyist doing junk science. If he really cared about helping humankind, he should have asked someone who knows how to do experiments to advise him on how to set up his protocols.