Recently I made a comment about activity trackers being one of the expected gifts for christmas. As a result, I got interested myself and – like always when I buy something – spent a few days researching the most popular models before finally purchasing Fitbit One.
Why the older Fitbit One then and not popular alternatives you wear on the wrist such as Flex and Force? Well, wrist versions have several advantages such as always being on your wrist regardless of what you do, but – at the end of the day, an activity tracker is all about accuracy and when I did my research I found that wrist trackers tend to have awful accuracy.
The accuracy of Fitbit Flex and Force
I first became aware about the problem when I visited the Fitbit Force product page on Amazon and noticed that more people had rated it 1/5 and 2/5 than 4/5 and 5/5. Sure, it wasn’t the only reason the buyers gave it such a low rating but it caught my attention.
Snipe found it to overestimate 2200+ steps daily and, when talking to her on Twitter, she made it clear that she didn’t see the point with the device until it improves. blackSPecials had a 30–80% overestimation depending on activity and mentioned that one hour of guitar playing would show 2000–2500 steps. Her recommendation? Carry it around your neck. The member luckydays27 on the MyFitnessPal forum also noticed how it counted steps even though she was sitting stil, talking with a friend – so she decided to return it and continue using One.
Her results with Fitbit One?
Weight lost based on Fitbit numbers: 24.59657143 lbs (11.16 kg)
Actual weight lost based on her scale: 25 lbs (11.34 kg)
Obviously there’s more than one source of errors and it seems to be a little bit to good to be true.
I’ve considered buying one just to check the difference with Fitbit One but figured it wasn’t worth the money. If anyone would like to lend me one, I’m happy to do it though.
The accuracy of Fitbit One
Why do I hail the accuracy of Fitbit One? Well, at least two studies have tested it:
Twenty-three subjects were fitted with Fitbit and Fitbit Ultra accelerometers, two industry-standard accelerometers and an indirect calorimetry device. Subjects participated in 6-min bouts of treadmill walking, jogging and stair stepping. Results indicate the Fitbit and Ultra are reliable and valid for step counts and determining energy expenditure while walking and jogging without an incline. – Source.
No significant differences were noted between Fitbit One step count outputs and observer counts, and concordance was substantial (0.97–1.00). Inter-device reliability of the step count was high for all walking speeds (ICC≥0.95). Percent relative error was less than 1.3%. – Source.
The latter study pointed out that there was a difference between the distance on the Fitbit device and the treadmill monitor. So I downloaded the fulltext and found that they used the default settings. In their discussion, they developed their thoughts about the problem.
To keep it short; Fitbit calculates stride length based on an average from your height, length, etc. You can change the stride length in the settings but this method to estimate length makes it extremely difficult to measure the device’s accuracy based on only distance.
By using Google Maps and Runkeeper (GPS), I’ve found my Fitbit One to differentiate less than 100 meters during a 6 km walk without changing the settings. I’ll compare it with my Garmin Forerunner 405 as well, but as I mentioned, distance isn’t a great way to test it.
Word of caution about the sleep tracking
Bland-Altman plots indicated that Fitbit overestimated sleep efficiency and total sleep time. Sensitivity for accurately identifying sleep was high within all sleep stages and during arousals. Specificity for accurately identifying wake was poor.
It’s the conclusion from this study and I’m not really surprised. Sure, you can change the sensitivity in the settings but the recurring problem with this kind of sleep trackers is that it measures movement and if you’re trying to fall asleep, you’re probably not moving…