Why I measure my Fat Loss with a Regular, Boring Scale
"It's not perfect but it's the best we have!"
tl;dr: All other methods are either less precise, more prone to user error, or too impractical/expensive to do daily. It doesn’t matter which scale you buy. Just get any $20 model from Amazon. No smart/body fat sensing required. All scales are imprecise. Still good enough.
Yea, yea, the scale isn’t perfect. Most scales aren’t that accurate. Weight can fluctuate quite a bit and most people are much heavier during the day or in the evening than they are in the morning.
But everything else I’ve seen is worse. And I’ve tried a lot.
Waist size is way less Accurate than Weight
Some people like measuring their waist circumference and claim that this is more reliably or accurate for them than scale weight. I measured both for years and this is absolutely not the case for me.
I went with pounds and centimeters because those are the smaller units (compared with kilograms and inches) and therefore make changes sound more dramatic, ha. Both weight and waist measurements were typically taken in the morning after using the bathroom but before consuming anything including fluids. Note I didn’t always use both scale and waist size hence there are some gaps.
You’ll notice that both curves are roughly in line and the trends are the same including peaks and troughs.
But the waist circumference measurement is way noisier. Within a few days I get inaccuracies of 5 or more centimeters at any waist size ever recorded, e.g. from my thinnest to my fattest. So it’s not just that it’s inaccurate when you’re very obese. It’s never very accurate. 5cm noise out of 110cm is 4.5%. There are nearly 10cm of noise in early 2017 near my all-time low making for 10 out of 95cm noise, or over 10%!
The weight, on the other hand, is way more accurate. Even crazy fasts barely reach the 10lbs of noise level, which is 5% at my lightest and 3% at my heaviest. “Normal” noise is around 3-5lbs (1% - 2.5%).
So if the trend line is the same for waist and weight but waist fluctuates by 3-5x the amount why would you use waist size? It’s just a shoddy signal. Maybe it gets better once you have six-pack abs, I don’t know. But I’ve used both from 200-300lbs and weight is just superior for me. So unless you tried both over a period of time and the waist signal is much more precise for you I’d just go with the scale.
Using a more precise signal allows us to get better feedback from our diet faster. If we have 5-10% noise in the signal we might have to wait months or years to find a signal.
Look at the period from August 2017 to July 2018:
You can see that my weight during this period is going up. But it’s only going up by 10lbs in a whole year, starting over 6 months in! Daily fluctuations are in the range of 5lbs and so we could be doing something right or wrong for several months before we’d ever have a chance to find out.
You think that’s bad? Wait for the same period using waist size:
The average seems to shift by 2-3cm over the course of 1 year. But you wouldn’t even notice anything happening for the first 7 months!
Get a scale. It’s much better for experimenting.
The one argument for waist circumference I can see is travel. It’s obviously much easier to pack a tape measure than a weight scale. Even if you have a scale at the hotel or gym they’re not all calibrated the same way and there are often huge differences.
Weigh yourself Daily but don’t Overreact
This may be a good time to explain why I advocate weighing yourself daily.
You can see above that there can be quite the fluctuation in your day-to-day weight (and even more in your waist circumference.) If you only took one measurement per week or even per month you could end up getting a very wrong signal. You might catch a crazy up/down trend and get the wrong idea about the impact you’re having with your current diet.
So I recommend you weigh yourself daily. I do it first thing in the morning after using the bathroom. My scale is actually in the bathroom so I don’t forget and it’s easy to do right away. I try not to drink any water before I weigh myself, but honestly if you had a sip it probably wouldn’t matter much.
Should you shower before or after you weigh yourself? It doesn’t matter. I often do both and the weight on the scale stays the same no matter how wet my hair gets.
Now when weighing yourself every day it’s important not to overreact. There are sometimes crazy days. Maybe you helped a friend move a couch in the blazing summer sun. You got super dehydrated. You had a giant social cheat meal.
There are a lot of things that can cause the next day to be a massive outlier. Try to look at the trend over a few days. Sometimes you can clearly attribute a specific action to your weight being way up or down the next day and sometimes you can’t.
Learn to look at the average trend of 3-14 days as you weigh yourself. That’s why I recommend putting your weight in a spreadsheet or something similar. That makes it easy to look at the trend. Where do you think I get all these graphs from?
High Signal vs. Noise ratio helps you Experiment
On ex150 accuracy is a mind-bogglingly .5% or so. Yes it is so consistent in weight that I often don’t go up a single pound over the course of weeks. Let’s zoom in on ex150.
Note how during the entire first 30-day ex150 experiment there was only a 2lbs fluctuation over the course of 11 days. Everything else was straight down.
During ex150-2 I went up by 1lb once early on and then again when I had a “social cheat meal” and ate a salad. Another 2lbs jump later was actually caused by pretty strenuous exertion the previous day which I am not used to and which probably caused some water retention. Then a 5lbs jump from another social meal, which consisted of tons of BBQ and coleslaw. Fiber or PUFAs? Not sure but it caused a weight jump at least 3x the weight of the food I ate.
You get the idea. On ex150 my weight is ridiculously stable and barely fluctuates by a pound. At most 2lbs unless I eat a cheat meal. Compare that to the high fluctuations I experienced before, often in the 3-5lbs/day range.
This is not just extremely motivating (although it is also extremely motivating (see what I did there?) because you almost never see your weight go back up) because you almost never see your weight go back up.
It’s also just a clearer signal. I can run an experiment in 14 days instead of 3 months. 30 days instead of 6 months.
It also helps to have a huge effect size. Wegovy makes you lose 15-18% of your body weight over 68 weeks? Good job, buddy! Gave it the old college try, did ya? Slow & steady wins the race.
Ain’t nobody on ex150 got time for that.
With an accuracy of 1-2lbs and an effect size of ~10lbs/mo we can clearly tell which way things are going after 14 days.
Let’s compare a low-noise vs. a high-noise diet experiment.
The above graph simulates a base diet that is weight-stable but fluctuates 2lbs/day. Say a weight stable ex150 once you’ve reached your target weight.
You can see that the red trend line of -10lbs/mo can easily be distinguished after 2 weeks, maybe even before that.
The yellow line of -5lbs/mo doesn’t leave the range of the random noise until after 9 days and we really need at least 14 days to tell it apart.
The green line of -2lbs/mo still almost touches the noise at 20 days out and we really need the full month to make any clear distinction.
Compare to this high-noise base diet with 5lbs of random fluctuations per day.
The red line still touches the noise at 7 days and we can’t tell anything happened. But after 14 days it is quite a ways from the lowest random point - in fact, it’s lower than the green line will be at day 30! A 14 day experiment is still very reasonable if our effect size is -10lbs/mo.
The yellow line ends up needing almost the whole month to distinguish itself this time. It’s almost touching the noise at 15 days! That means we could probably make a pretty educated guess by 21-25 days but it would make sense to wait out the full 30 days to be sure.
The green line? It never even leaves the noise area. How could it, with an effect size less than half of random, daily fluctuations?
You would have to run the green experiment for 1.5-2 months to even tell if anything is happening at all.
That’s why the combination of low weight fluctuation and large (expected) effect size is so powerful for experimenting and finding a diet that works. There are an infinite number of possible diets. If it takes us 6 months to evaluate one vs. 14 days then it’ll take us 12x as long to get it right.
Now imagine this scenario, where we use a noise level of 10%. Like the tape measure and waist circumference method above, remember?
You can’t detect even a respectable -5cm/mo trend with this after a whole month of experimenting!
The high effect size of the red trend line still takes over 15 days to cut through the noise. But at least we get a very definitive signal after that and have left the noisy area in the dust after 30 days.
Which scale to buy? Doesn’t matter.
I just went to The Wirecutter on bathroom scales and bought their recommendation. It’s less than $20 on Amazon. I’ve used other models in the past and they all work the same.
Don’t bother with a “smart scale” or electric-impedance body fat measurement. It’s not better and “smart” just means it’ll make you sign up for an app.
If you’re serious about experimenting and fat loss I recommend you create a spreadsheet or something similar to track your weight. You’ll want to compare things and generate your own graphs eventually. Their app won’t be good enough, I promise it.
What about DEXA?
Knowing your body fat percentage is cool but not required to quit obesity. It doesn’t matter if you went from 40% to 39.9%. The scale gives you the same information.
DEXA can be great to see if you lost a lot of lean body mass. It’s normal to lose some lean body mass when losing fat but obviously it’s better to lose as little as possible.
I don’t worry about this while still in the obese range. If I lose a few pounds of muscle in exchange for losing 100lbs that’s a good deal to me.
It is a good idea to do a DEXA once in a while. If your lean body mass is still at a reasonable level you can safely continue what you’re doing. Should it drop precipitously you might want to reconsider your approach.
My last DEXA showed I still had above-average (for my weight/height/age) lean body mass and that’s good enough for me.
Of course you also probably don’t have a DEXA machine in your bathroom and it costs money to do the scan. While they can often be had for under $100 and I recommend you do it once in a while, that’s prohibitive for everybody but millionaires. Plus DEXA scans expose you to a little bit of radiation. It’s safe if you do it once in a while but probably adds up if you were to perform one daily.
Scale is good enough to carry you over between DEXA scans.
Don’t bother. Just get DEXA’d every once in a while.
Some people claim they can get very reliable results using calipers to measure the fat in skin folds. The thing is that it’s extremely liable to user error. You have to get trained and use a very repeatable process or your results will be all over the place.
Why bother? Just use a scale and get DEXA’d every couple months if you really care about the body fat percentage.
The scale lies about its Accuracy
All scales are a bit inaccurate. It’s ok. They’re $20 and they’re good enough for our purposes.
My digital scale gives the weight in pounds with one decimal point, implying that it’s accurate to a tenth of a pound. But it’s clearly not that accurate. As I saw my weight dwindle over the last few months I noticed that it always had one or two favorite spots per pound. E.g. it’d always be 250.2 but never 250.3 or 250.5.
Just for fun I decided to test the scale’s accuracy. I weighed myself repeatedly, drinking exactly 100g of water between each weigh-in. I repeated this 10 times. Of course you’d expect my weight to go up linearly in 10 equal increments. But since the scale isn’t actually that accurate what you see instead is this:
Foolish scale! Why even show 1/10th increments if you can barely detect a pound? Oh well, no matter.
Conclusion: Just buy a $20 bathroom scale, weigh yourself every morning, track your weight on a spreadsheet, don’t overreact
Our goal here is to quit being obese. We probably need to lose 30lbs, 50lbs, maybe 100lbs or more. The decimal point doesn’t bother us at all.
As you’ve seen above with my ex150 weight graph the 1lb resolution is easily good enough to spot even a 1-week or 2-week trend.
So just buy any old scale and start experimenting.
I'll say that with the waist measurement, a big cofounder for me is bloating as I'm super prone to it, so unless I eat strict carnivore I can expect a huge variation in my waist measurement.
Have you tried measuring another body part, like upper arm or thigh or high bust, full bust, chest?