3 ms·

It's a minor argument for something that tends to come up. It's come up a few times on HN some studies that show sperm quality is much lower in obese men, and t

It's a minor argument for something that tends to come up. It's come up a few times on HN some studies that show sperm quality is much lower in obese men, and that there are more obese men. There also tend to be small effect sizes that suggest within the healthy weight bin, sperm quality has dropped a small amount over time. But its naive to think that this means it still isn't a weight effect. The bin argument is basically saying that if the entire population gets fatter, you would expect the average weight within a bin to go up, even if some people move into the next. This depends somewhat on the shape of the curve, and the bins. But on a whole it's simply saying that the average weight of people between X lbs and Y lbs is higher than it was before. Really what it's saying is that the bins obfuscate what could be better analyzed as a continuous variable effect. The bins are arbitrary and misleading. People can fall within the same bin at t1 and t2 but be fatter at t2. > I think obesity is a huge factor, maybe 80%. But I don't know why you feel so strongly that every other explanation is completely false. The evidence and effect size is simply extremely strong, and the prevalence of obesity is very high. I estimate obesity to account for more than 90% of the sperm effect and diet to account for more than 90% of the obesity effect. I'm not opposed to the idea that plastics might have some small single digit effect size. Nor am I opposed to the idea that we might see an inflection point where more plastics start being significantly worse. But for now, it seems to be a tiny, unimportant part of the story.

titzer 5y ago
> The bin argument is basically saying that if the entire population gets fatter, you would expect the average weight within a bin to go up, even if some people move into the next. This depends somewhat on the shape of the curve, and the bins. But on a whole it's simply saying that the average weight of people between X lbs and Y lbs is higher than it was before. It does depend on shape of the distribution, but it is generally not the case that the average within a bin will go up. It is certainly not true for a bell curve, where if you push the curve out to the right (and flatten it), some bins will have higher averages (because they are weighted more to the right than before), but some bins will have lower averages (because they are weighted more to the left than before). If we wanted to go way deep on this, it really depends on the second derivative of the curve within a bin, not even the first derivative (a flat or even straight line through the bin will always give an average right in the middle).
- spywaregorilla 5y ago
  Apologies you're right. I oversimplified. Though I would assert that my premise for this particular distribution held true and that binned analyses were misleading. I tried to find the article. I could not. This particular example was had no lower bound on the left below overweight, and was skew right with the peak of the curve in this lowest category if memory serves. I remember thinking it was visually obvious that the bin would get on average fatter but I suppose it does not generalize.

titzer 5y ago

> The bin argument is basically saying that if the entire population gets fatter, you would expect the average weight within a bin to go up, even if some people move into the next. This depends somewhat on the shape of the curve, and the bins. But on a whole it's simply saying that the average weight of people between X lbs and Y lbs is higher than it was before. It does depend on shape of the distribution, but it is generally not the case that the average within a bin will go up. It is certainly not true for a bell curve, where if you push the curve out to the right (and flatten it), some bins will have higher averages (because they are weighted more to the right than before), but some bins will have lower averages (because they are weighted more to the left than before). If we wanted to go way deep on this, it really depends on the second derivative of the curve within a bin, not even the first derivative (a flat or even straight line through the bin will always give an average right in the middle).

spywaregorilla 5y ago
Apologies you're right. I oversimplified. Though I would assert that my premise for this particular distribution held true and that binned analyses were misleading. I tried to find the article. I could not. This particular example was had no lower bound on the left below overweight, and was skew right with the peak of the curve in this lowest category if memory serves. I remember thinking it was visually obvious that the bin would get on average fatter but I suppose it does not generalize.