Beyond Meat Is Struggling, and the Plant-Based Meat Industry Worries

Re:Beyond Meat Is Struggling, and the Plant-Based Meat Indus... by sph on Tuesday November 29, 2022 @06:22AM (#33786171)

Epidemiologic nutritional studies are proven to be totally useless. They literally are a questionnaire asking people what do they eat, take some health markers and drawing conclusions. I spent the morning reading through the data of that "plant-based diets reduce colorectal cancer" on the front-page of the Guardian [1], and it's yet another pile of cherry picked inconclusive crap people will keep sharing on social media.

From just 2 minutes with your paper: "Higher heme iron intake appeared to be significantly associated with a 31 % (95 % CI 4–67 %) elevated risk of developing CHD" --- lmao, loving that 95% confidence interval between 4 and 67%.

Passing those out as proof of real science keeps the disinformation and terrible practice alive. If I had a dime every time a random commenter gave me a link to an epidemiologic nutritional study, I could fund a double blind randomized one myself.

1: https://www.theguardian.com/science/2022/nov/29/plant-based-... — have fun reading the actual numbers in Tables 2, 3 and 4 of the linked paper.

[ Parent | on HN ]

Re:Beyond Meat Is Struggling, and the Plant-Based Meat Indus... by wholinator2 on Tuesday November 29, 2022 @10:09AM (#33788477)

I'm curious how one gets a 95% confidence on such a wide margin. I'm having trouble visualizing data that would give such a result. Is this just a result of a very wide distribution of data? Or is this done with shady statistics. The 95% seems to imply that there's data on either side of their chosen bounds. To my understanding, the real "beef" with the article is the choice of representing the data as "31%" when there's such a wide distribution. A more accurate statement would be that, "nearly everyone experienced some heightened risk of CHD with the actual risk varying largely, but firmly positive". Thoughts?

[ Parent | on HN ]

Re:Beyond Meat Is Struggling, and the Plant-Based Meat Indus... by iopq on Wednesday November 30, 2022 @10:45PM (#33812070)

people who had less than 4% elevated risk composed 2.5% of the group, and people who had over 67% elevated risk composed 2.5% of the group

if normally distributed, two thirds would have elevated risk between 14% and 48%

that is to say, almost everyone has an elevated risk - so I have no idea why you are complaining about confidence intervals

[ Parent | on HN ]