Data is awesome and people generate a lot of data. It’s not surprising that a number of companies, such as Fitbit and Jawbone have moved into the space of capturing data from people. it seems that fitness is the first major (adopted) wave of “self quantification” which is being adopted.
Of course, bad data is not entirely useful data. To that end, I’ve wondered about the accuracy of some of these devices. While I cannot address their total accuracy, I have started to poke around whether they compute the same data for the same movements. Specifically, if I wore a Jawbone and Fitbit device for one week, would they have the same data? Fitbit vs Jawbone – one week of data – would I get the same data?
Turns out, no, not really.
This test was super simple and I will be digging into details in the future. As a quick test, I wore a Fitbit Charge HR and a Jawbone Up Move for one week. Both devices were worn on my hands (different hands) and appropriately calibrated and configured. If I took one off (to shower, etc.) I made sure to remove the other. After one week of logging, I captured these results.
This chart shows the total number of steps each device recorded over the span of one week. Interestingly, some days the devices were closer while others they were pretty far apart.
This graph shows the total difference (absolute) between the two devices over the week.
Overall, it looks like the Jawbone is either being conservative, the Fitbit liberal, or both are a little bit skewed. Time to recalibrate and try again.