I’m sure Harvard produces good engineers, doctors, nurses, etc. (Though maybe not lawyers, since I know plenty of bad politicians with a law degree from Harvard.)
It’s just they have long-running studies where they use food frequency questionnaires (FFQs). The original idea of the nurses study, for instance, was to get a relatively homogeneous groups of folks to study: mainly women of the same socioeconomic status. One problem, according to Gary Taubes, is they never verified the socioeconomic status. So, you could have a single mother with three kids living in a not-so-wealthy area and a woman whose husband is a doctor or CEO and living in a posh area. Who is most likely to be healthier? Yet they never test this.
And if you were a nurse sending in FFQs every once in a while, would you admit that you’re eating EVIL red (or processed) meat? The people who have the temerity to admit they eat red meat probably drink more, exercise less, smoke more, have more stress, don’t give a crap, etc. Or maybe you’re one of those people who have tons of money and are eating “healthily” with high carb but also meats like salmon, crab, the expensive stuff, playing tennis or whatever, going to the gym with a personal trainer, etc. The people who say they eat red meat are probably completely different from the people who don’t or say they don’t.
And they’ll try and “control” for that, but the chances that you put data like that into an algorithm and “correct” for something like obesity or being overweight and have it be anywhere near reality are near zero. (Not to mention that this is going to be done with bias – Harvard “knows” that obesity = bad, even though I’ve seen quite a few podcasts/youtube discussions where people who are obese are metabolically healthy.)
Anyway, the chances that anything is true based on a study that uses FFQs is near zero. And this includes the studies that go my way, meaning a study where red meat is “shown” not to be bad.