The fat people I know don’t appear to give a damn about “the standards of our society”. My morbidly obese grandmother (BMI around 40) does think “why bother? It’s easier to just eat whatever I want and never exercise”, but she also thinks the same about lots of other things (e.g. she smokes over a pack of cigarettes a day, it’s hard to convince her to wear a cast when she breaks a bone, etc. -- she appears to be just waiting to die), and my parents and my sister (BMIs all around 30) say “I’m not that fat—look at grandma for instance” (and they also insist that I (BMI around 25) am absurdly skinny).
BMI is a ridiculous measure of obesity as far as I’m concerned. I know this doesn’t directly respond to your comment, but I’m listing my main arguments here for posterity.
For anecdotal context I’m about ten or fifteen kilograms above the weight I was when I was in secondary school (i.e., fully-grown), ate a (modulo American (hi John_Maxwell_IV!)) healthy diet, and led an active lifestyle. I consider this a reasonably good approximation to my “healthy weight”, but it corresponds to a BMI of 27.3. For philosophical context I’m not a “fat apologist” either; it is painfully obvious that obesity causes higher morbidity. It should go without saying that higher morbidity is (ceteris paribus) bad.
In no particular order:
The units are absolutely meaningless. If humans are to a zeroth approximation cylinders with comparable aspect ratios and average density, then weight is proportional to volume, which is proportional to height cubed, but BMI is the ratio of weight with height squared. NB: This is only a zeroth approximation; tall people tend to have different aspect ratios, and the power law approximation is probably better served by an exponent somewhere between two and three. See MacKay for more details.
BMI gives less information than intuition on extreme cases: army1987′s grandmother is indistinguishable from a 2000-era bodybuilder. The classical response to this objection is that BMI needs to be augmented with other statistics (typically given ad-hoc: waist-to-height ratio, waist circumference, body fat percentage, and etc.), but this is not how BMI is used “in the wild.” I’ve observed several online cases (yes, this is weak evidence here) of both 25-30′s becoming obsessed with getting down to 20-25 (even at the expense of muscle), and on the other end 17-20′s becoming obsessed with not going over 25 (even through gaining muscle).
Even if BMI alone doesn’t help us classify the ~35+ and the ~20- folks, it is possible to argue that it is useful for the people in between. At this point Goodhart’s law kicks in: the “easiest” (for some value of ease) way to lower your BMI is to get a liposuction (or more absurdly, an amputation), but that has little effect on overall health. Less absurdly, I claim that body composition is just too varied and complicated to be reasonably treated with a single statistic.
There is a good amount of fake accuracy in reporting BMI to three significant figures, as mass in kilograms is usually only accurate to 1.2-1.5 places. Even healthy people gain and/or lose two or three kilograms over the course of a year.
For obesity diagnosis (i.e., in a medical setting), BMI is easily replaced with, in decreasing order of accuracy and cost, 1) displacement measurement of body fat percentage, 2) electrical measurement of body fat percentage (despite being horrifically lossy), 3) qualitative visual assessment of body fat percentage.
For overall health (i.e., in a non-medical setting), BMI is less easily replaced with body fat percentage. Setting aside akrasia, there’s a great deal of psychological baggage that tends to mind-kill one’s ability to judge one’s own weight. To make matters worse, my own criteria (weight maintained after puberty in an active lifestyle) is depressingly ineffective in light of widespread childhood obesity. I admit I don’t yet have a good solution to this problem, but I claim that BMI obsession is overall more harmful than whatever lay purpose it serves.
I don’t know for certain, but it seems to me that a common failure mode for people struggling with weight is paying too much concern to various statistics, invented or not.
Disclaimer: I don’t believe the following is good general advice. I don’t have any theory backing it up.
At this point, in my own current work, the most useful thing I’ve implemented is applying the nameless virtue to this. If I want to look better, then I need to have goals that involve looking better; if I want more endurance, I need to have goals that involve running harder and longer, and etc. It seems trivial but I don’t see very much of it in the fitness community.
I am confused by this exchange. When we talk about cognitive bias, we accept data that contradicts self-reports all the time. So there is substantial evidence that self-reports about mental processes are unreliable. But when the topic shifts to social norms, self-reports are considered reliable again? I notice I am confused.
BMI assumes you are the normal semi-sedentary modern person. It’s not meant to be used on serious athletes or weightlifters. For 95% plus of the population, BMI is a pretty accurate metric.
BMI assumes you are the normal semi-sedentary modern person.
More importantly, BMI assumes you are of average height. Human weight doesn’t actually scale by the square root of height, so BMI has a systemic bias for tall people (too high) and for short people (too low).
As far as I recall, BMI was designed as a tool to compare whole populations (where the height bias averages out) and people who created it explicitly said that it’s not a good metric to evaluate individuals.
While it’s true that BMI is a rough metric and gets rougher when you’re dealing with unusual proportions or body compositions, those effects are often exaggerated. An athletic male of 6 feet 6 inches (99.8th percentile) and 210 pounds, which is about what you’d find in your average pro basketball player, would score as normal weight.
Too rough for my taste. Once your average pro basketball player adds 10 lbs of pure muscle and become 6′6″ at 220 lbs, BMI will declare him to be overweight.
Yeah, I weigh about the same as I did two years and a half ago, but I looked and felt much fatter back then. (OTOH none of the other people I mentioned in the grandparent exercise regularly, and anyway I only used the BMI because it was the quickest quantitative measure I could think of.)
Oh, I should clarify that my arguments were not meant to chastise you for using BMI. I don’t have any problem with it in the way you use it; it looks like it falls under argument two but it’s clear that you’re using it as a rough signal of degrees of morbid obesity.
The fat people I know don’t appear to give a damn about “the standards of our society”. My morbidly obese grandmother (BMI around 40) does think “why bother? It’s easier to just eat whatever I want and never exercise”, but she also thinks the same about lots of other things (e.g. she smokes over a pack of cigarettes a day, it’s hard to convince her to wear a cast when she breaks a bone, etc. -- she appears to be just waiting to die), and my parents and my sister (BMIs all around 30) say “I’m not that fat—look at grandma for instance” (and they also insist that I (BMI around 25) am absurdly skinny).
BMI is a ridiculous measure of obesity as far as I’m concerned. I know this doesn’t directly respond to your comment, but I’m listing my main arguments here for posterity.
For anecdotal context I’m about ten or fifteen kilograms above the weight I was when I was in secondary school (i.e., fully-grown), ate a (modulo American (hi John_Maxwell_IV!)) healthy diet, and led an active lifestyle. I consider this a reasonably good approximation to my “healthy weight”, but it corresponds to a BMI of 27.3. For philosophical context I’m not a “fat apologist” either; it is painfully obvious that obesity causes higher morbidity. It should go without saying that higher morbidity is (ceteris paribus) bad.
In no particular order:
The units are absolutely meaningless. If humans are to a zeroth approximation cylinders with comparable aspect ratios and average density, then weight is proportional to volume, which is proportional to height cubed, but BMI is the ratio of weight with height squared. NB: This is only a zeroth approximation; tall people tend to have different aspect ratios, and the power law approximation is probably better served by an exponent somewhere between two and three. See MacKay for more details.
BMI gives less information than intuition on extreme cases: army1987′s grandmother is indistinguishable from a 2000-era bodybuilder. The classical response to this objection is that BMI needs to be augmented with other statistics (typically given ad-hoc: waist-to-height ratio, waist circumference, body fat percentage, and etc.), but this is not how BMI is used “in the wild.” I’ve observed several online cases (yes, this is weak evidence here) of both 25-30′s becoming obsessed with getting down to 20-25 (even at the expense of muscle), and on the other end 17-20′s becoming obsessed with not going over 25 (even through gaining muscle).
Even if BMI alone doesn’t help us classify the ~35+ and the ~20- folks, it is possible to argue that it is useful for the people in between. At this point Goodhart’s law kicks in: the “easiest” (for some value of ease) way to lower your BMI is to get a liposuction (or more absurdly, an amputation), but that has little effect on overall health. Less absurdly, I claim that body composition is just too varied and complicated to be reasonably treated with a single statistic.
There is a good amount of fake accuracy in reporting BMI to three significant figures, as mass in kilograms is usually only accurate to 1.2-1.5 places. Even healthy people gain and/or lose two or three kilograms over the course of a year.
For obesity diagnosis (i.e., in a medical setting), BMI is easily replaced with, in decreasing order of accuracy and cost, 1) displacement measurement of body fat percentage, 2) electrical measurement of body fat percentage (despite being horrifically lossy), 3) qualitative visual assessment of body fat percentage.
For overall health (i.e., in a non-medical setting), BMI is less easily replaced with body fat percentage. Setting aside akrasia, there’s a great deal of psychological baggage that tends to mind-kill one’s ability to judge one’s own weight. To make matters worse, my own criteria (weight maintained after puberty in an active lifestyle) is depressingly ineffective in light of widespread childhood obesity. I admit I don’t yet have a good solution to this problem, but I claim that BMI obsession is overall more harmful than whatever lay purpose it serves.
Do people have any incentive to be interested in the number on the scales itself, regardless of its effects on health and looks?
I don’t know for certain, but it seems to me that a common failure mode for people struggling with weight is paying too much concern to various statistics, invented or not.
Disclaimer: I don’t believe the following is good general advice. I don’t have any theory backing it up.
At this point, in my own current work, the most useful thing I’ve implemented is applying the nameless virtue to this. If I want to look better, then I need to have goals that involve looking better; if I want more endurance, I need to have goals that involve running harder and longer, and etc. It seems trivial but I don’t see very much of it in the fitness community.
I am confused by this exchange. When we talk about cognitive bias, we accept data that contradicts self-reports all the time. So there is substantial evidence that self-reports about mental processes are unreliable. But when the topic shifts to social norms, self-reports are considered reliable again? I notice I am confused.
BMI. Those things are fun. If I spend enough time in the gym I can get myself up to “Obese”.
BMI assumes you are the normal semi-sedentary modern person. It’s not meant to be used on serious athletes or weightlifters. For 95% plus of the population, BMI is a pretty accurate metric.
More importantly, BMI assumes you are of average height. Human weight doesn’t actually scale by the square root of height, so BMI has a systemic bias for tall people (too high) and for short people (too low).
As far as I recall, BMI was designed as a tool to compare whole populations (where the height bias averages out) and people who created it explicitly said that it’s not a good metric to evaluate individuals.
While it’s true that BMI is a rough metric and gets rougher when you’re dealing with unusual proportions or body compositions, those effects are often exaggerated. An athletic male of 6 feet 6 inches (99.8th percentile) and 210 pounds, which is about what you’d find in your average pro basketball player, would score as normal weight.
Too rough for my taste. Once your average pro basketball player adds 10 lbs of pure muscle and become 6′6″ at 220 lbs, BMI will declare him to be overweight.
Yeah, I weigh about the same as I did two years and a half ago, but I looked and felt much fatter back then. (OTOH none of the other people I mentioned in the grandparent exercise regularly, and anyway I only used the BMI because it was the quickest quantitative measure I could think of.)
Oh, I should clarify that my arguments were not meant to chastise you for using BMI. I don’t have any problem with it in the way you use it; it looks like it falls under argument two but it’s clear that you’re using it as a rough signal of degrees of morbid obesity.