logoalt Hacker News

adrithmetiqatoday at 2:22 PM1 replyview on HN

Exactly. Just another “study” finding a correlation without causation.


Replies

citadel_melontoday at 3:59 PM

Teasing out causation with empiricism is near impossible without eventually needing to rely on occum's razor to some extent or another.

Reliance on occum’s razor would probably be less needed if this was a random control trial, but still the study would be correlative with alternative explanations still plausible.

Regarding health, focus on calorie control and getting enough fats/carbs/protein. Eat whole foods that are high enough on the satiety index because they make calorie maintenance more intuitive so you don’t have to count calories if you don’t want to. Those (and maybe a few other tips) are the only things that have a large enough effect for one to determine with almost (only almost, because everything empirical is a confidence interval/correlation) certainty that they’re effective.

Any study saying that blueberries are “superfoods” or any other hyper-specific food recommendation, I immediately don’t trust it. There just isn’t any organization that would fund a RTC of such a niche finding, especially considering you would need to pay and surveil thousands of people over the course of their whole life to change their diet and stick to it. I don’t think even the NIH is giving out millions of dollars to a research team to find out if blueberries are superfoods.