Improving Your Work BMI: The Importance of Context and Relative Measures

All models are wrong, some are useful – British statisicain George Box

In the 1970s, American physiologist and dietician Ancel Keys and a group of colleagues were looking for a way to quantify a healthy man.

They stumbled upon an experiment by a Belgian Mathematician in the 1830s. Lambert Adolphe Jacques Quetelet set out to find the “l’homme moyen” or the “average man”.

Quetelet thought that if he took enough measurements and compared them all he could find the ideal weight. When looking at the numbers he found that in general, weight went up with the square of the height of the person.

And so BMI was born.

This was not a very broad study.

Keys replicated this by only measuring 7426 healthy men from 12 different groups.

This created a measure that is inherently twisted towards the group it was modelled on. Because there wasn’t a wide variety of people studied the results mean that lots of people fall out of the normal range, me included.

It wasn’t till later

The Population vs The Person

With its small sample size, it is not a great measure for a person. However, it is pretty good for a population group.

Putting everyone’s information into the model can give a good indication of the health of the group, but putting just one person’s information in can give them an incorrect understanding of their health.

Take me for instance, when I was playing basketball I was 110kgs and 200cms and this gave me a BMI of 27.5.

A normal BMI is between 18.5 and 25.

I was as close to obese as I was to normal. I was also 5% body fat and had a resting heart rate of 40 beats a minute. By all accounts, I was extremely healthy but looking at my BMI alone told a completely different story.

I could have been 75kgs and still been in the normal range. I would have been skin and bone. The numbers lie, or at least don’t tell the whole story.

Specific vs Relative

As a specific measure, meaning the number is accurately telling us something, the BMI is a piece of shit.

As a relative measure, meaning that going up or down tells us something, then it is more valuable.

Now that I am injured and no longer playing basketball and exercising as much as I once was, my BMI is now 28.8, so it has increased from 27.5.

You can read this as that I am generally less healthy than when I was playing. Which is mostly accurate.

The number is interesting, but the scale of good or bad is wrong. My height definitely makes the scale less accurate. Context and nuance also matter.

BMI, KPI, and You

We love to measure things. Your performance is often measured against something. Even though we know people game the system and short-term incentives are often terrible,

Despite this, some of the measures might be interesting.

You want to assess if more or less is better or worse and look at the relative movement rather than the specific number.

You want to look at the trend of movement to what you think is an improvement or problem.

If your workplace is solely focused on KPIs and does not understand nuance then by all means game the hell out of the number, but make some internal metrics that aim for relative measures so you can track your performance.