Tracking steps
What I have learned from tracking my steps for three years and seven devices
Counting daily steps seems to be the second most popular form of self-tracking. I have tracked my steps for 3 years, and lately with multiple devices. In this post I’ll try to analyze some of the data I have collected.
What you measure is what you get
I started tracking my steps about three years ago, in the end of November 2013. I installed the Moves app and began to glance at the data on a daily basis.
The first thing I noticed was that tracking changed my behavior very quickly.
Prior to installing Moves, I was optimizing my daily commute quite efficiently by avoiding walking. After I started tracking my steps, they became a friend instead of a foe. If I had to choose between a bus going 500 meters closer to home and another leaving 10 minutes earlier, I would more often choose the latter and walk some extra steps. I started to take short walks just for walking’s sake — but still very infrequently.
Development of steps
I had no goals (e.g. losing weight) and no health issues I was worried about. One reason of installing Moves was to just keep track of where I had been. I was mostly just curious.
Unfortunately I have no data about my steps before Moves. I have a feeling that I started walking more after installing Moves, but no hard data to back up that assumption. At least my attitude towards walking changed.
In the data I have collected, I can see some slight changes of behavior.
As you can see from the graph above, I don’t reach the well known 10 000 steps per day recommendation. (My actual daily steps would produce a very noisy and spiky graph, thus the graph above shows 28 day rolling averages. I occasionally get 10 000 or even 12 000 daily steps, but averaging smooths them out.)
During the end of 2013 and most of 2014 my average daily steps were below 3 000 more often than above. If you look at the data from 2014 (light green line in the graph), you can see that the number of days with more than 3000 steps seems to slightly increase during the first nine months of the year, although there are also some lazy days and weeks. I attribute this slight increase to my changed behavior, but it could just as well be caused by the weather.
In September 2014 I started to reach 5 000 daily steps more often — and kept doing so until April 2015. From April to August 2015 the daily average remained between 4 000 and 5 000 steps, and then jumped back to more than 5 000. The reason for increased walking was another app: Ingress. (You may have heard of an Augmented Reality game that was a source of all the Poke Stops in Pokémon Go. That’s Ingress.) In August 2015 my son started school and I took a 5 month leave from work. 5 days a week I walked about 3 km in the morning and another 3 km in the afternoon to escort him to school and back.
During 2016 my step count declined back to the level of 2013. Playing Ingress requires less walking than when I started (there are much more portals), and I haven’t had enough incentives to walk. I know I should walk more, and I like the way it refreshes my mind, but just like going to the gym — it’s just too easy to postpone. I should find a way to increase my walking again for 2017.
Measurements from different devices
In December 2015 my Pebble smartwatch started counting my steps. During 2016, I have bought a bunch of other step-counting devices as well. The graph below shows my daily steps during two weeks in December 2016 counted by 7 different devices:
- Lumo Lift
- Moves app
- Nexus 5X phone (using Google Fit)
- ŌURA ring
- Pebble smart watch (I have various models)
- Spire mindfulness and activity tracker
- Wii U Fit Meter
As you can see, they don’t always agree on my total daily steps.
One thing is quite clear: Moves gives significantly lower step counts than others. It only counts “active” steps — when I’m continuously walking or running.
The most significant reason for differences in step counts between devices is the fact that I temporarily lose them, forget or choose not to wear them. That results to zero steps counted for one or more days.
Some devices don’t sync the steps in the cloud unless I use their apps and make sure that the connection between the device and the phone isn’t interrupted, which also leads to data loss.
My Spire and Lumo Lift are pretty recent purchases, and I cannot say anything definitive about the reliability of their step counts. ŌURA shows low step counts since I mostly use it for tracking my sleep and rarely wear it during the day.
I regard the Wii U Fit Meter as kind of a reference point I can compare others to. I have the meter in my pocket more often than not (although I frequently lose it for a few days) and it seems pretty accurate. (What I dislike is that Nintendo offers no API to my data. I manually type my daily steps into a spreadsheet.)
More analysis
All the devices are positioned differently on my body. Lumo Lift is attached to my shirt just under the collarbone, Spire to my waist, Wii U Fit Meter in my pocket. ŌURA is (obviously) worn on a finger and Pebble on a wrist.
I don’t know the details of the algorithms the different gadgets use to count my steps. I would assume that the devices on wrist or on finger will have to take extra measures to separate signal (steps) from noise (hand-waving).
Both at home and in the office my phone is usually laid on a table, and cannot count all my steps. This affects both Google Fit and Moves. When I look at the hourly step counts during a day, the differences show.
The graph below shows by step counts during December 14th 2016 measured by Lumo Lift, Moves, my phone and my watch. I chose that day because all the four devices had collected data the whole day, and it’s a pretty regular day for me. I only have daily step counts from other devices, and can’t analyze them on a hourly basis.
Moves counts more than 800 steps between 5 AM and 6 AM, where the others are quite unanimous in about 330 steps. Pebble and Nexus 5X measure almost 1 200 steps from 6 to 7 AM, Lumo Lift more than 1300 and Moves just above 200. Until 3 PM, Moves doesn’t count any steps and Nexus only a few. Pebble and Lumo Lift (which I wear all the time), count some steps also during office hours — for some reason Pebble shows some steps an hour earlier than Lumo. Between 3 and 5 PM I have walked with my phone, probably consulting colleagues about something. The 6 PM peak is my commute back home.
Conclusions
If you are interested in counting your steps, take into account following:
- Different devices measure different things. Don’t compare apples (e.g. steps counted by your watch) to oranges (steps counted by Moves). Try to decide whether you want to count all your steps or only the “active” ones, and choose your devices accordingly.
- Try to find a device that’s easy to wear. A phone or a watch blends into your life, is hard to lose and does much more than just counts steps. More specialized solutions (devices like Lumo Lift or Spire) can provide insights that a watch can’t. Wellness bracelets generally beat smart watches in battery life and they can be worn around the clock (although you probably aren’t going to collect many steps during the nights). The distinction between a watch and a wrist-worn activity tracker is blurring all the time.
- If you want to increase your level of activity, tracking can help (what you measure is what you get). Additional motivational aspects (e.g. setting explicit targets, competing against friends, getting a dog, playing Ingress or Pokémon Go) can help even more. Find out what works for you.
- Perhaps you want to know how the changes in your activity affect your health or other aspects in your life. In that case I suggest you click on the Follow button below.
This is my first post in a series I’m planning to write about self tracking and data correlation. It’s also my first post in Medium. If you’re interested in seeing more graphs and thoughts about my data, go ahead and follow this publication (Tracking Self). Also, I’d very much like to hear comments and your own experiences.
