Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

BMI is trivial to measure conclusively, using tools that are almost universally available and ready. And at population scale, it's proven to be a strong and pretty signal for analyzing in studies like these as the ways in which BMI can fail for individuals is too uncommon for it to matter to population studies.

It's a perfect tool to use for studies like these, as it can contribute understanding society-level concerns about the obesity epidemic and what impact we can expect it to have if it continues not to be tackled.

But like with almost all medical and nutritional research, drawing conclusions for yourself as an individual based on these population-scale studies is inevitably fraught. In the case of BMI studies, you can often get a fair sense of whether you should be concerned or not, but do need to take into account other things that you know about your lifestyle and health. Your doctor, presumably, does exactly that.

Meanwhile, body fat percentage tends to be very hard to measure accurately in the first place and has different techniques that tend to have different error characteristics, making it hard to correlate it between studies done through different clinics, let alone against someone's personal assessment at home or with their personal doctor/trainer/whatever. It probably has high relevance to some medical and nutritional outcomes, but isn't a very useful input variable for research because its inescapably noisy.



> BMI is trivial to measure conclusively

That does not make BMI a good measure.


It does for some applications, and not for others. In epidemiology you're looking at the number across broad populations, where a lot of the "three dimensions" idiosyncrasies cancel out. You're likely to run into trouble taking literal BMI constants and turning them into individualized patient care instructions, though.


Or basically: there are outliers, and those tend to complain. But the average billions cancel them out statistically.

I'm an outlier myself, and so is anyone who did some serious sports for a significant amount of time. There are other reasons to be an outlier too. But they're not so statistically significant.


ditto. Always a bit frustrating when the sadly overweight nurse looks at my height and weight, charts it, and mutters to themselves "borderline". My last visit, at least the doctor deigned to actually eyeball me and said "you look like you are quite active, we can ignore this".


Oh I'm inarguably overweight, but not to the point the BMI would indicate, considering that when i was younger and in top form i was between 29 and 30.


You are saying it is a not dreadful statistical measure.

That is doubtful, but possible

It is used for individuals. That is where it is harmful


> harmful

It's hard to conceive of anyone meaningfully "harmed" by being a BMI outlier.

The only real occassion it comes up is when some lean and uncommonly athletic person gets miscategorized as obese and get warned of increased risks to their wellness if they don't lose weight. A warning which they generally laugh at because they know their own situation.

That's hardly "harm" -- "mild social nuisance" is more apt.


The harm is that are more likely to get prescribed unnecessary medications. Most people are not confident enough or not knowledgeable enough to wave off what doctors say.


> It's hard to conceive of anyone meaningfully "harmed" by being a BMI outlier.

"You are fat. You must diet."...."Oh my Dog!! I am fat!!! I will fast"


Potentially for health insurance but I get the actuarial argument, especially since higher than typical weight regardless of fat composition puts higher than typical strain on your cardiovascular (and other) systems


I apologize if this is tangential, but I have noticed BMI research is often heavily criticized, but the same issues that plague BMI can easily be applied to psychometric measurements of intelligence (IQ), but somehow IQ is spared the same beating for some reason I must not understand.

Perhaps my low IQ is showing, but why is one metric heavily criticized while the other is not? (Rhetorical)


> somehow IQ is spared the same beating

I think this is the part you're getting wrong, which is why you're confused.

IQ was contentious from its introduction in the beginning of the 20th century, caught some traction in the mid-century, and has mostly been on a decline of esteem and respect again since. It's never been very broadly accepted, let alone spared from criticism, nor have most other psychometrics, although (as always) there are bubbles of academia, industry, and the general population that swear by them and work to legitimize them.

Kurt Danziger has published really great histories of all this.

I'll add that his books are especially interesting in the context of all the AI buzz as insight into the history of "intelligence" (the word, the socially constructed concept, etc) helps wrap one's head around what it means when people working in CS try to repurpose its terminology in their own field.


IQ is as far as I can tell quite well accepted as a diagnostic tool. The pushback is in its application across whole populations.


FWIW, anecdotal data is also considered a valid diagnostic tool too, and we know how unreliable such information can be.

Also, by populations, do you mean groups of particular types of people or just groups of people in general? If anything, I thought IQ really only had anything close to merit at large population levels. Much of what I have read clearly fails at an individual level.

However, if you are talking about IQ's application in regards to populations like one race vs. another, then yes, I think the pushback is well deserved.


I agree with you about your last statement! I think it's actually a pretty interesting comparable to BMI. BMI: useful epidemiologically, across large populations, but much less useful individually. IQ: much less useful across large populations, but useful as a diagnostic/clinical tool in individual settings.


[flagged]


No, it hasn't. What ever else you want to say about the research results you're describing, "widely accepted in the field" is not a thing you can hope to say about them. The objections scientists have to these kinds of statements is fractally deep.

Claims that politics are preventing genetic and psychometric research also seem to drive practitioners in the field pretty nuts; there is clearly and obviously a ton of this research happening "in the west", it just doesn't say what race science activists want it to.

I owe another thread a deeper answer than this, so I'm keeping this brief.


[flagged]


No, they haven't. The opposite thing has happened (I owe another thread details): as we've shifted to genomics, the estimates have cratered. This isn't an "activist" thing.

You know you're in trouble when people start implying that actual genomic and quantitative psychological research in academic labs is suspect because of "taboos". In fact, since you made that point, and since I really shouldn't be digging into this before responding to the other person I said I would respond to: I'll leave it there.


[flagged]


In case anybody was wondering whether you were commenting in good faith.


> IQ is as far as I can tell quite well accepted as a diagnostic tool

Inside certain bubbles, sure, because there's pressing demand to have something quantifiable and sufficient supporters willing to assert its legitimacy.

But good luck taking your test results outside of those bubbles to earn you some opportunity or respect, even if you can say your test was administered to the highest standards by the most respected administrators.

Nobody here will take your results to mean anything. Nobody at your work will take your results to mean anything. The recruiter at some new potential job won't either. The prestigious university you want to attend won't care. Most of your partners and friends probably won't.

Many of them may already have a sense of whether you're "intelligent" or not, either from personal interactions or from other practical indications. But good luck finding somebody of material influence who would hear your IQ score results and change their mind. It just doesn't have that weight.

That's what it means for it not to be treated seriously outside of bubbles. There are supporters that back it, and there are institutions that rely on it, but outside of those very specific places, there's very very little trust in it.


In any bubble where IQ is considered legitimate, just mention that Richard Feynman apparently had a recorded IQ of 124 or so (or Lewis Terman's study of gifted children).

You'd think someone tried to convince those in the bubble that the Earth is flat. In my experiences, I have been met with all kinds of rebuttals like, "The test he took was probably over-weighted in verbal vs. spatial reasoning" or "He probably just blew the whole test off."

No one has a copy of the actual results to my knowledge, but he and his sister (who oddly enough scored higher) verified the story back in the day according to biographers. So, not telling what is the true story.

For the sake of argument, let's assume Feynman's score was indeed accurate. I think it makes people extremely uncomfortable for two reasons:

1. Some individuals with high IQs (IQ >= 130) feel threatened or ashamed by the fact that someone with an high but unremarkable score was capable of truly remarkable contributions to their field.

2. The scores are truly not always accurate or might potentially measure something that isn't intelligence.

If Feynman's score was indeed an inaccurate measurement of his intelligence, then that says more than I need to know about IQ testing.

I mention Feynman because I am in agreement with you:

> Many of them may already have a sense of whether you're "intelligent" or not, either from personal interactions or from other practical indications.

No one would consider Feynman to be unintelligent by any stretch of the imagination. However, at the end of the day, his score truly doesn't matter. Feynman nor anyone else is not remembered because of their intelligence, but rather, his contributions. Besides, exceptional people tend to be, well, exceptional.


The pushback is because it suggests some uncomfortable possibilities...

Also wait until people find out the correlation between IQ and obesity (runs for cover)...


> IQ is as far as I can tell quite well accepted as a diagnostic tool.

not at a scientific level, no


Are you sure about that? I'm shoplifting an argument from Twitter from a quantitative psychologist.


> there are bubbles of academia, industry, and the general population that swear by them and work to legitimize them.

Perhaps you a right, because this is my main exposure to psychometric. From much of what I have read from various journals, there are clearly those that are contrarians, but mainstream academic psychology presents IQ as if IQ is some infallible measurement of human intelligence.

Despite IQ's merits being contentious (like BMI), it is still widely used regardless. IQ testing is still used for clinical diagnostic purposes (why I had to take one), entrance in to some gifted academic programs, in judicial contexts, etc..

Based on my personal research and experiences, I align with the contrarian views vs. the mainstream, academic views.


IQ is definitely not spared. It doesn't take a Mensa member to realize that human intelligence cannot be reduced to a single number.


IQ routinely gets destroyed. As an example, I was an undergraduate in the 1990s. My statistics class skewered the use of IQ back then. According to some friends with undergraduates, they still use that as an example of statistics being used badly.


IQ is an interesting one. There are indeed plenty of issues in actually measuring it and deciding what we're even measuring to begin with, but at the same time, I've heard some people try to say it doesn't even exist at all.

Which seems odd, because I've yet to see anyone who doesn't believe that stupidity exists and it's the other side of the same coin.


> I've yet to see anyone who doesn't believe that stupidity exists and it's the other side of the same coin.

I do not think that is the argument though. Clearly, there is some sort of distribution of all human attributes. I think the argument is in how (in)accurately something as qualitative and arbitrary as intelligence can be measured.

I like the analogy of athleticism. Some people are more athletic than others, but again, such differences are generally highly contextual. I am sure some sort of test composed of factors like the number of push-ups, sit-ups, and time trial of a 100m dash could be used to generate some sort of AQ (athletic quotient).

However, scoring high on the 100m dash might positively correlate with performance in sports, but I would not wager that a high AQ is truly indicative of athletic performance as a whole. At least not in any transitive sense when mapped to many sports. For example, Usain Bolt might have a very high AQ because of his running ability, but his AQ might not serve him well in a game of ice hockey or swimming.


Sure, but in the case of a hypothetical AQ, we could do factor analysis and pull out different things, like the strength of various muscles or muscle groups, like all pure running tests had strong correlation with things like leg muscles. In IQ, this has famously been done and found that a single factor dominates.

Or when they saw that knowledge of arcane sport rules interfered, they could look for a more pure test of ability rather than knowledge, which they did with IQ by going from tests of specific factual knowledge to Raven's progressive matrices.

And when you do that kind of work, you find that you can develop better measures that have a lot of predictive power for whole populations, even if they do not tell you, say, how many goals they'll score in their next hockey game.


People say that, but then people like me are administered the tests for clinical diagnostic purposes (right or wrong), and even my state still considers the scores to be a valid measurement of intelligence, and thus a factor in whether someone convicted of murder is eligible for the death penalty.

I think the concept is bullshit myself, but regardless the damage is done, and still being done for what it is worth.


My crude understanding is, we don't have a good way to calibrate above-average scores, but it is a good measure for below average (provided there aren't arbitrary barriers like the language it's in).


That’s too bad, but I replied to something that claimed there is no criticism of IQ. IQ is criticized and used. Those are not exclusive states.


Yup. Polygraphs are still used in the intelligence community and court too.

I'm seeing the tide turn (imo), but by no means am I claiming it's no longer used (regardless of merit). I acknowledge I'm starting a personal anecdote which is subject to my own biases such as cohort affects.


Huh? I haven’t seen “IQ” seriously used in decades. Its flaw are well known.


I was administered the WAIS-IV for the purpose of a clinical diagnosis about 10 years ago. My state still considers it one of many factors in whether one is a candidate for the death penalty. Some gifted programs still accept the scores as necessary evidence for admission.


Army too, back then. IQ test and MBTI (Myers Briggs), if I recall correctly, then OCS considerations if ranked well enough, including an oral interview.


Even in popsci/social media I've seen more and more awareness that it's "bunk" science.

It's in the same category as all that "alpha male" theory, Meyers Briggs, polygraphs, and "left vs right brained" people.

I'm starting to see more awareness around the concerns of "ballistic forensics" too, which is heartening.


> Even in popsci/social media I've seen more and more awareness that it's "bunk" science.

Devil's Advocate: Then why is it so heavily used still? (Then again, I guess back to my initial point -- the same could be said for BMI.)


Proof that's it's still heavily used in serious scientific literature?


How do you define 'serious scientific literature?' In other words, what type of evidence would you like me to present? Does any peer reviewed research that uses IQ as a factor count as serious? (Obviously, I won't present research that is about IQ measurement.)


I have never had IQ test taken, but countless of times BMI has been calculated. Mind you I have had other mental tests done due to ADHD and Autism concerns, and some of them were similar about abstract logic, but they were combined with other things like focus, memory, multitasking, literary understanding and other mental capabilities. The results of these tests were presented as percentiles in the population.


Interesting. I was given an IQ test for ADHD diagnostic purposes. There were other test given, as well. And in my opinion, the other tests were more indicative than the IQ test was (which I think is falling out of favor with ADHD testing). Then again, there were some issues with my test and scoring, so it makes me even more dubious of the (psuedo)science.

I went through the arduous process merely for the formal documentation, which in hindsight has served me well. Not a single doctor has ever questioned my diagnosis. However, as the psych told me, a 10 minute conversation with me should honestly have been evidence enough. ;)


Interestingly, the actual ADHD tests didn't find any issues, my main issue was my mind wandering in social situations, listening, understanding, keeping track of conversations and my own input. But whatever the tests were they were stimulating enough that I had no problem focusing. It's also repetitive, routineus actions that I have trouble focusing on, so it's almost like untestable issues. I did get the diagnosis though, I don't remember with what justification since the tests didn't allow for that.


Do you mean "why is it still used in popsci"? I'd wager because it makes people feel a similar way to other pseudo scientific categorizations, but that's mostly speculation on my part. Add in a bit of inertia and the simplicity of it ("let's reduce the cognitive ability of people across the world to a single number!") and, to me, I can see the appeal for others.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: