Hacker News Reader

Comment by Mistletoe | Hacker News Reader

This is why nutrition research is extremely confusing and contradictory often. I worked in a nutrition department and the amount of post menopausal obese women in our knee studies that said they just ate a half cup of green beans that day is astounding.

Now try to use data that flawed to make assumptions over a lifespan about human health. About the only studies I truly believe are the ones where people are at a facility and all food is provided to them and tracked.

View on HN

My favourite example of this is that the number of condoms used in the USA, according to surveys, is dramatically higher than the number of condoms sold in the USA.

Ew.

Perhaps they are using them more than once.

Or they are confusing imagination of how much they would like to have sex with reality. Or they feel the social pressure, to boast themself. Sexual activity is kind of a primitive success metric. Reporting low numbers means reporting low success ..

In surveys the amount of liquor consumed is often about half the amount of liquor that is sold.

Yeah, that's because it's assumed that the other half consumed is non-registered moonshine. I don't get how that would apply to condoms, though.

Grey import through Chinese websites like AliExpress? I doubt those individual sales are registered on the total sales numbers of the USA.

I lived in one of those food study centers after college. Most people there were pretty diligent about sticking to the program, but there was a big scandal when one guy was kicked out after discovering extensive cheating and several papers had to be retracted.

We were allowed out - I had a full time job, but couldn't eat or drink any food not provided by the center (we took a radio-isotope tracer with food, and had to collect our poop). It was quite interesting :_)

what was the consequence if you did eat outside food?

If the researchers found out then you would be eliminated from the study and no longer paid (and given free food) as a research subject.

Well, some studies validate their FFQs. Also, FFQs don't have to be perfect. They just have to create analytical clusters or continuity. As FFQs become less accurate, confidence intervals get wider but it just depends if the study is powered to handle it.

I find that most of the dismissal around FFQs is pretty vague and seems to come from a group of people who find the consensus in nutrition research inconvenient for them.

Reply on news.ycombinator.com