4 ms·

You know those studies saying that a cup of wine a day is good for you? Huge numbers, compelling evidence. Turns out they were all wrong due to a pretty simple

You know those studies saying that a cup of wine a day is good for you? Huge numbers, compelling evidence. Turns out they were all wrong due to a pretty simple and 'obvious' bias. Plenty of people who stop drinking did so due to health conditions, both diagnosed and undiagnosed. And so when you just compared drinkers to non-drinkers, you were biasing towards healthy individuals in the drinking side, and the effect was significant enough to create a statistically significant 'discovery.' All large scale observational studies are subject so such biases, and the one you linked to also seems a likely suspect there. The hidden confounders are pretty much infinite. The users of many of the worst of sorts of drugs, such as heroine and amphetamines, tend to be average to lower weight. They will die disproportionately young, which will drag their weight cohort substantially down. Another could be increases in things like body building, which show increases in BMI without the corresponding increase in body fat that it's supposed to represent. I'm not saying any of these are the silver bullet, but rather that when you see a big observational study like this, that sort of skepticism should be the first thing to enter into your mind, as it's extremely often the case when you read about a [shocking discovery] based on big observational studies. Correlation is not causation, even with really big population sizes.

kennywinker 3y ago
Seems like you have a belief, and this data that challenges that is rife with flaws while the data that supports that belief is good clean science. > The users of many of the worst of sorts of drugs, such as heroine and amphetamines, tend to be average to lower weight. They will die disproportionately young, which will drag their weight cohort substantially down. Right so this study was skewed by meth use, something that there’s a good chance they controlled for at least where they could, but other studies couldn’t possibly be skewed by drugs that cause weight gain and shorten lifespan. My takeaway from this and similar studies is that body fat (especially as categorized by BMI) is mostly orthogonal to health. It’s possible to be fat + healthy, fat + unhealthy, slim + healthy, and slim and unhealthy. And if that’s the case, burdening fat people with my own assumptions about their health is misguided, and dumb.
- somenameforme 3y ago
  They did not control for these or countless other variables that could be confounding. You can read the exact study here [1]. This is just a general weakness in large scale observational studies. The reason the wine comparison is particularly apt is because it fits the exact same scheme. Studies specifically on alcohol demonstrate a million and one negative side effects with little to show for it, yet [shocking discovery] shows that a glass a wine day is actually beneficial for longevity? There needs to be an explanation for why. And most often that answer is eventually found to be that [shocking discovery] was just reporting on a false correlation, driven by a confounding variable. It's the exact same here. There are countless extremely negative factors that being overweight weighs strongly into, yet somehow being overweight is good for life expectancy? There needs to be an explanation for why and, absent some revolutionary new discovery, by far the most likely explanation is a confounding variable. [1] - https://jamanetwork.com/journals/jama/fullarticle/2520627 https://jamanetwork.com/journals/jama/fullarticle/2520627
  - kennywinker 3y ago
    I see they did not. I also see that in denmark in 2023 (not in the 1970s and early 2000s when the data was gathered) about 1.6% of 16-34yo’s had used meth that year. The prevalence of full blown meth addiction is going to be dramatically lower. 0.5% seems high but let’s go with that - 50 people in this study. Seems a bit of a reach to say that’s what’s skewing the data. But the overarching point is: why is your default that fat is unhealthy and you must have bulletproof data to believe otherwise? My default is i don’t know one way or the other and I think health is probably way more complex than what % body fat you have and I don’t understand why so many people are determined to use it as a proxy for health.

kennywinker 3y ago

Seems like you have a belief, and this data that challenges that is rife with flaws while the data that supports that belief is good clean science. > The users of many of the worst of sorts of drugs, such as heroine and amphetamines, tend to be average to lower weight. They will die disproportionately young, which will drag their weight cohort substantially down. Right so this study was skewed by meth use, something that there’s a good chance they controlled for at least where they could, but other studies couldn’t possibly be skewed by drugs that cause weight gain and shorten lifespan. My takeaway from this and similar studies is that body fat (especially as categorized by BMI) is mostly orthogonal to health. It’s possible to be fat + healthy, fat + unhealthy, slim + healthy, and slim and unhealthy. And if that’s the case, burdening fat people with my own assumptions about their health is misguided, and dumb.

somenameforme 3y ago
They did not control for these or countless other variables that could be confounding. You can read the exact study here [1]. This is just a general weakness in large scale observational studies. The reason the wine comparison is particularly apt is because it fits the exact same scheme. Studies specifically on alcohol demonstrate a million and one negative side effects with little to show for it, yet [shocking discovery] shows that a glass a wine day is actually beneficial for longevity? There needs to be an explanation for why. And most often that answer is eventually found to be that [shocking discovery] was just reporting on a false correlation, driven by a confounding variable. It's the exact same here. There are countless extremely negative factors that being overweight weighs strongly into, yet somehow being overweight is good for life expectancy? There needs to be an explanation for why and, absent some revolutionary new discovery, by far the most likely explanation is a confounding variable. [1] - https://jamanetwork.com/journals/jama/fullarticle/2520627 https://jamanetwork.com/journals/jama/fullarticle/2520627
- kennywinker 3y ago
  I see they did not. I also see that in denmark in 2023 (not in the 1970s and early 2000s when the data was gathered) about 1.6% of 16-34yo’s had used meth that year. The prevalence of full blown meth addiction is going to be dramatically lower. 0.5% seems high but let’s go with that - 50 people in this study. Seems a bit of a reach to say that’s what’s skewing the data. But the overarching point is: why is your default that fat is unhealthy and you must have bulletproof data to believe otherwise? My default is i don’t know one way or the other and I think health is probably way more complex than what % body fat you have and I don’t understand why so many people are determined to use it as a proxy for health.