Body Fat% is a way off PART2

Got this scale yesterday, and can confirm that mine is WAY off as well. Garmin please fix the firmware, there's obviously something wrong. I'm very frustrated to see a generic reply 8 days ago from Garmin in this topic, basically not taking any responsibilities, very PR friendly for Garmin but user unfriendly reply. You basically say "You bought an overpriced scale, that doesn't work well and I'm going to act as if nothing is wrong, you are just f*cked".

if I see generic copy/paste response in this topic again, I'll lose my faith in Garmin and send the scale and fenix 6 pro back right away.

Don't act like there's nothing wrong, that only frustrates your customers. Take responsibility, say there's an issue and you're going to fix it, otherwise you will lose users. This is how you ruin a brand.

I am about to stupidly put out an opinion.  Is the product defective, probably no.  Is the result what you want/expect, the answer is no. So, let me defend my resounding maybe.

When you set up the scale you add height (used along with impedance to calculate a path length for the current flow), weight, age and are you normal/athletic.  THe height is part of the impedance normalization so the same algorithm can be used for all heights.  Weight... a fixed number to base calculation on.  Age is used to select the range in which you can expect numbers.  Body type is probably used the same as age.  So, we get on the scale, we get weight (a true collected data point) and we get normalized impedance.  These numbers are then used to calculate (and again I’m guessing that Garmin is using tables based on an independent  test menthol like DOXA.  And such tables are freely available for NIH, and such).  So, now we have all the Dara, and here’s where it falls apart.

I’m 6ft, 175 lbs, 67 years old.  No secrets here.

• On my old weight Guru scale, I get 175.4 lbs with 16.4% BF
• Using my Garmin the same day, I get 175.4 lbs (excellent, same base number) with 27.9% BF

I should be upset, and I was; but not so much now.

I then used another calculation mode, that asked for all the data, as before.  But instead on impedance, it asked for body measurements (like a caliper, which I think is the Gold Standard?).  When I use the measurement method, I get

• Using the US Navy calculations, I have 13.6% BF
• And using the BMI method for calculation, I get 27.9%

So, what we need to know to decide the quality of the data, is the method used to calculate the results.  It appears, from a small investigation, that Garmin uses the BMI method.  It would be nice to know

So, I’ve decided that the %BF should be used as a trending number, not a hard number.  And, so I’ve made some peace with the number.

I’ve ignored any variations on collection of impedance values because, near as I can see, the differences are very small.

I don’t work for Garmin, nor am I an apologist for them.  But, as a retiree that used to design scientific optimal equipment, I got curious on the results variation and this is the result of a couple hours of reading and talking with old friends.

Sorry, its not the answer most want.

Len

I think that your assessment is probably reasonable but whatever Garmin uses for calculating BF is not in accordance with DEXA, the gold standard.  I wish that it were.  Accordingly, since I like the integration of the scale with my tracking systems, I will continue to use it - I will, as you suggest, just use it as a relative metric.  Then, I will return to UC Davis every 6 months to get a good measurement.  After a few DEXA measurements, I can factor the historical delta between Garmin and DEXA into my collected scale metrics, then perform a calculation in my Excel spreadsheets to reliably tell me a BF measurement - I export my fit files from Garmin Connect to Excel every day so that I can see the progress of the measurements that are important to me.  Thanks for your thoughts.

Glad I could be useful.  I would be curious how the your DEXA data compares.

thanks for sharing your thoughts

It's been a Covid while since my last measurement.  I need to get a scan soon - UC Davis is operating.  When I get some good numbers I will definitely share the delta data.  The last measurement I had was between 8 and 9%.  I've not been training as hard since then but I still think the scale reads high at 20% (Covid).  I'll know better when I get a scan.  Cheers.

oh dear, after waiting about 2 months to get my S2 scales, i get my first reading, note that it is way off and find that there is a 2 month old thread (among many others) on such issues. and garmin are denying any problem.

to add my voice: i have had 2 previous scales - salter and fitbit. both gave my BF% reliably as 7 (fitbit) or 8 (salter) in athlete mode (around 14 in normal mode). my new S2 gives 18.6. my activity class is set to 8 so that should be athlete mode according to https://support.garmin.com/en-NZ/?faq=4yEmnMQhmb70eWsxxm8vr5&productID=530464&tab=topics 

i can understand some variation between different algorithms but more than double what others calculate? i haven't had any other sort of test done in years but to look at me and grab a skinfold, it is very clear that i am far below 18% BF.

i've owned so many garmin devices over the years and always seen complaints about problems that garmin deny but never had it myself so imagined it wasn't as bad as people make out. now though i am caught in the same situation and realise i've just been lucky previously and may need to rethink my loyalty. its ok to have a problem occaisionally but to deny rather than apologise and resolve is unforgiveable but seems to be a regular pattern with garmin.

the claim that "The body composition metrics provided by Index S2 compares to known gold standard methods (Dexa scan, Bod Pod, etc.) favorably on average." is laughable as averages are meaningless in this context, it is all about individuals!

update to note that i updated my activity level to the maximum 10 and my body fat is magically down to 15%. so it seems to be a bit more detailed than athlete mode on/off. if i could increase my activity level to about 15 i would probably get a proper BF reading! i don't think activity level 10 is actually appropriate for me - i am a serious athlete who trains hard but if i'm 15/10 then what is a pro who trains 2-3 times as much as me?

agreed, it seems like this is what garmin consider to be working. they seem to have no interest in addressing the fact that for significant numbers of users their new scales give significantly wrong body composition results as compared to any other measurement including their own previous scales. 

unfortunately garmin have quite a history of ignoring significant issues with their products

options:

1. have you tried setting your activity level to 10?
2. accept that you will get false numbers and just look at trends
3. send them back to garmin for a refund and buy a competitors product (only downside being that you can't then integrate into the garmin ecosphere if you have other garmin products... up to you if you see this as a deal breaker and ditch garmin altogether)

It would be nice if garmin just allowed for a calibration offset, similar to how the treadmill footpod calibration offset works...