When I started writing this post, I had just finished my academic semester, and was feeling very relaxed. Relaxed enough that I wanted to take a slice of my then-ample free time to write about something that has bugged the hell out of me for years. Unfortunately, inertia, changing blogs and going back to uni conspired to delay actual posting until now. Without further ado, however, I present: MY RANT AGAINST BMI!
The BMI, or Body Mass Index, is otherwise known (at least to me) as the arithmetically-derived pseudoscientific twaddle used to diagnose Australia as the 'Fattest Nation on Earth', having recently pipped the US for that crown. You heard it here first, kids. You know, unless you watch the news or read the paper, that is.
A little history (I'm drawing heavily on relevant sections from Fat Politics by J. Eric Oliver for my facts and figures). The BMI was developed in the 1830s by a Belgian astronomer and physiognomist named Adolphe Quetelet. He was interested, not in measuring human health, but in exploring the applicability of mathematical probability to humans. Quetelet collected data from French and Scottish army conscripts, plotting their heights and weights. He found that, for those who weighed roughly the statistical average for their height, weight was proportional to height squared. This is the source of the BMI formula (BMI = weight (kg)/height (m)²). As Oliver notes, however, Quetelet went a step further. He fallaciously reasoned that, since the weight of the 'average' conscript in his sample was proportional to their height squared, that weight and height should always exhibit this relationship. In case you're wondering, this relationship is the basis for 'ideal weight' calculations. It is also noteworthy that his sample was presumably comprised of only adult, white males, presumably with some form of standardised military physical training.
To provide a clearer example of Quetele's reasoning is 'bayad, m'kay', let's look at another example. Aristotle famously assumed that women were essentially passive, as Athenian women had no formal political role and little influence. In doing so, he treated something that was contingently true of Athenian women (their lack of political power) as a necessary fact about all women.
Anywho, let's look at a little testing theory. In order for any test to be useful, it needs to consistently gauge the target variable (e.g. weight). After all, there is no use for a set of bathroom scales that gives wildly varying results when the same person steps onto them several times in a row. Tests that deliver consistent results in this way are called reliable. Perhaps more importantly, it's crucial that a test measure what it purports to measure. Testing mavens call this property 'validity'.
Mechanically speaking, the BMI is a simple ratio of weight (in kg) divided by height (metres) squared: w/h². For a start, no such simple measure can be sensitive to different builds, proclivities for water retention, body fat percentages, bone density or any number of other variables that might reasonably affect this value. Examining Quetelet's methodology reveals that there is nothing magical or even special about this ratio -- it was simply an observed relationship in a subset of his very limited sample. Even if it weren't for Quetelet's poor reasoning in jumping from observation to injunction, his restricted sampling makes it impossible to make any valid inferences about the height-weight relationship for anyone not represented in his sample.
Let me clarify why this is a problem. Imagine that I went into the Olympic Village and took the body fat percentages of every female sprinter. I then went home and analysed the data and found that the average body fat rating was 14%. From that, I concluded that every man, woman and child of any age, fitness level and build should have 14% body fat. Hopefully, this reasoning seems as silly as it is, and this is precisely the logical process underlying BMI.
So this is the measure used on evaluating the health of nations; hence, it will be used to direct funding, justify lap band surgery and drive the diet industry. I'm not arguing that people are not getting heavier, eating poorly or exercising less than, say, 20 years ago. What I am saying is that BMI is an inappropriate measure of that trend. Empirically, the strongest correlations with coronary heart disease are not with BMI, but with levels of tummy fat. BMI is widely used not because it has any particular intrinsic merit, but because it is easy to measure. The problem, of course, is that it doesn't measure what it should, and that is a dangerous failing indeed.
The BMI, or Body Mass Index, is otherwise known (at least to me) as the arithmetically-derived pseudoscientific twaddle used to diagnose Australia as the 'Fattest Nation on Earth', having recently pipped the US for that crown. You heard it here first, kids. You know, unless you watch the news or read the paper, that is.
A little history (I'm drawing heavily on relevant sections from Fat Politics by J. Eric Oliver for my facts and figures). The BMI was developed in the 1830s by a Belgian astronomer and physiognomist named Adolphe Quetelet. He was interested, not in measuring human health, but in exploring the applicability of mathematical probability to humans. Quetelet collected data from French and Scottish army conscripts, plotting their heights and weights. He found that, for those who weighed roughly the statistical average for their height, weight was proportional to height squared. This is the source of the BMI formula (BMI = weight (kg)/height (m)²). As Oliver notes, however, Quetelet went a step further. He fallaciously reasoned that, since the weight of the 'average' conscript in his sample was proportional to their height squared, that weight and height should always exhibit this relationship. In case you're wondering, this relationship is the basis for 'ideal weight' calculations. It is also noteworthy that his sample was presumably comprised of only adult, white males, presumably with some form of standardised military physical training.
To provide a clearer example of Quetele's reasoning is 'bayad, m'kay', let's look at another example. Aristotle famously assumed that women were essentially passive, as Athenian women had no formal political role and little influence. In doing so, he treated something that was contingently true of Athenian women (their lack of political power) as a necessary fact about all women.
Anywho, let's look at a little testing theory. In order for any test to be useful, it needs to consistently gauge the target variable (e.g. weight). After all, there is no use for a set of bathroom scales that gives wildly varying results when the same person steps onto them several times in a row. Tests that deliver consistent results in this way are called reliable. Perhaps more importantly, it's crucial that a test measure what it purports to measure. Testing mavens call this property 'validity'.
Mechanically speaking, the BMI is a simple ratio of weight (in kg) divided by height (metres) squared: w/h². For a start, no such simple measure can be sensitive to different builds, proclivities for water retention, body fat percentages, bone density or any number of other variables that might reasonably affect this value. Examining Quetelet's methodology reveals that there is nothing magical or even special about this ratio -- it was simply an observed relationship in a subset of his very limited sample. Even if it weren't for Quetelet's poor reasoning in jumping from observation to injunction, his restricted sampling makes it impossible to make any valid inferences about the height-weight relationship for anyone not represented in his sample.
Let me clarify why this is a problem. Imagine that I went into the Olympic Village and took the body fat percentages of every female sprinter. I then went home and analysed the data and found that the average body fat rating was 14%. From that, I concluded that every man, woman and child of any age, fitness level and build should have 14% body fat. Hopefully, this reasoning seems as silly as it is, and this is precisely the logical process underlying BMI.
So this is the measure used on evaluating the health of nations; hence, it will be used to direct funding, justify lap band surgery and drive the diet industry. I'm not arguing that people are not getting heavier, eating poorly or exercising less than, say, 20 years ago. What I am saying is that BMI is an inappropriate measure of that trend. Empirically, the strongest correlations with coronary heart disease are not with BMI, but with levels of tummy fat. BMI is widely used not because it has any particular intrinsic merit, but because it is easy to measure. The problem, of course, is that it doesn't measure what it should, and that is a dangerous failing indeed.
Posts
2 comments:
So you won't get mad at me for eating the last of the coconut jelly while you were out? I needed it for my bum.
PS: amusing if ill-informed link -- http://www.theage.com.au/national/nine-million-australians-are-a-ticking-fat-bomb-20080619-2tjv.html
Well, according to BMI, I should be somewhere between 75.5kg and 83kg. At the moment I'm around 85, which means I'm overweight.
When I was in the middle of that weight band (79-80kg), my mum thought I looked like I was dying.
So much for BMI - I'm with you!
Post a Comment