[nradov]

Your hypothesis is incorrect, or at least not justified by the data. The whole field of human nutrition actually is in a very bad place. Most of what we think we know comes from low quality observational studies which don't rate well on the evidence based medicine scale.

By all means let's continue research. But people generally shouldn't rely upon most of it when making dietary choices. A better approach is to conduct your own n=1 informal experiments and determine empirically what works best for you.

[rafaelero]

But you see, I don't agree that these conclusions are based on low quality data. These conclusions are not merely based on observational studies, as you pointed. This question is being asked for more than 70 years and a lot of different types of designs have been tried: putting people on strictly monitored diets and seeing what happens to their biomarkers, finding people with weird mutations that lower their LDL cholesterol, etc. By the way, the only type of data that disagrees with the expert's conclusion are, you guessed it, observational data. And that is simply because there are so much individual variation that these studies end up always underpowered. If you track the same individual and changes his/her diet, the conclusion is inescapable: saturated fat increases LDL cholesterol. And LDL cholesterol is causally linked to heart disease.

I don't mean to say that this narrative can't be pushed back, but these articles comes from the [wrong] premise that these findings are not solid. They are. It's the same as trying to advance the idea that trans fat is not dangerous to someone's health: you surely could try, but the prior should be low. That's is why I advanced the hypothesis that people are craving for these types of news that validate their lifestyles: they don't seem to be properly calculating their priors and are dismissing too quickly solid evidence.

[nradov]

Go back and read those studies again. No clear causal link has been found between those biomarkers and all-cause mortality.

[rafaelero]

Oh, dear...