Project Porky! - Body Fat scales

If you're really interested in getting an accurate absolute number, have a dexa body scan.

Expensive, but I think the only way to get an accurate view of your body composition.

Interesting stuff from Madcarew, thanks for that.

I have always assumed my Garmin Index scales had a smoke generator, which it used to blow it up my arse, as it reckons I'm 7% body fat. I'm skinny and tall, but I still find that figure very hard to believe. But as said above, they are consistent, so helpful to see trends rather than absolute empirical accuracy.

What madcarew says about algorithms: +1.

You could step from a Tanita to a Garmin to a Withings and expect them to nail mass consistently (they won't agree perfectly on even that). But even if they used consistent impedence technology between them, they throw it into the pot with population data and interpret it via algorithms - and they differ in that.

Some of them are getting smarter with different profiles (eg you can select 'athlete' so that it will assume more muscle/less fat) but the bottom line is that you should take composition readings with a pinch of salt. I'd maybe expect members of a cycling forum to give the algorithms more of a challenge with body fat than the general public.

The body fat readings are based on an algorithm which uses the factors that you put in (Sex, Height, Age etc) to approximate your fat %ge based on an impedance reading compared with sample studies. Studies show that it's within 3% for 60% of the population, within 8% for 85% of the population and bears little resemblance for the rest of the population. (If the scales say you are 24%BF, then 85% of the people who stand on it will have a BF of somewhere between 16% and 32%. That's quite a variation, so you couldn't really rely on it for accuracy. However, they are reasonably consistent longitudinally, so their reading remains relatively consistent (+/- 3-5%) for any given individual. So, as others have said, don't get too hung up on the actual number, watch the trend. Usual story is weigh youself at the same time each day, ideally when you first get up as your hydration levels etc are generally going to be similar going from day to day. And bear in mind that for an 80 kg person your weight can vary by as much as 2.4 kg each day without any change in your BF %ge. If you are glycogen depleted (about 400g glycogen in your body), as each gram of glycogen requires 2 grams of water to store, then that alone can produce a variance of up to 1.2 kg compared to fully glycogenised (made up word). And it is very easy for your hydration state to vary by 1 kg across the day, so the fact that we tend to be within a few hundred grams each time we measure ourselves is testament to our body's extraordinary ability to maintain homeostasis. Not an expert, but have a biology degree

My weight changes by over 1 kg from morning to evening - I can also shed about 1.5 kg just by going for a cycle when it's hot.

The only way I can get meaningful and consistent measurements is by weighing myself after I've got out of bed in the morning.

It also takes around 4 weeks to notice any actual changes in weight.

I've also heard that aerobic exercise and associated fitness levels can skew results. Something about building glycogen stores and glycogen being stored with water can lead to higher fat readings than with sedentary people. 

"bounce (blob) around a bit from morning to night"

The body fat measurement is closely linked to hydration levels, which vary throughout the day.

Measure yourself when you get out of bed to get consistent readings, and don't angst too much if they vary day to day. It's the weekly trend that matters ...

Have used Tanitas since the late 90s.

Do you mean Tanita? Your link brings up the Withings Body+ ones that I have.

I am not an expert, but...
from research I did when I was looking, it seems that various factors influence accuracy, and some claims that the time of day/amount of liquid you'd taken in made some difference (one review I read speculated [I think] that if you weighed yourself while you were fully hydrated, your skin was more likely to have sweat on it, which made the reading more accurate).

Those scales will be very accurate for weight, but best not to get too hung up on body fat% as you have the usual failings of impedence running from foot-to-foot (ie. even if it's bang accurate, it's not going to extrapolate it perfectly for the rest of your body). Some of the Tanita hand-and-foot sensor ones were supposed to be better for that reason - but when you look at their total product (scales and sensor hardware, their algorithms and software, and the apps to make sense of the data), they don't put their best stuff into that market segment - some of their products sell for £000s, presumably to doctors and physios.

From reviews I think they do some really accurate body composition ones at the next rung up, but you're into something like £400 for them. If you're not going to fork that out, as ReadingTim says, just make sure you're giving your current ones chance to be consistent with themselves (eg. weigh yourself at similar times of day), and view the overall trends.

A quick google brought up this, which tallies with other things I've read.  Suggest that, as with weight generally, you don't get too concerned by the accuracy of the actual value on the day, but monitor the trend over the last few weeks. 

That said, as weight etc does fluctuate during the day, measuring youself at roughtly the same time each time does try and fix at least one of the (many) variables. 

Apologies if that's too simplistic!