Fitbit steps for running, walking and cycling

I recently got a FitBit activity tracker and have been having great fun competing with family members who live abroad to see which of us can take the most steps. It is a fun way for me  feel connected to them by looking at their activity logs even if time zone differences make it impractical to actually speak to them.

As part of a discussion about one of our step challenges I was asked if the FitBit was giving me credit for taking steps while I cycled. (I think I might have been subtly accused of cheating in the challenge).

The honest answer was that I didn't know how/if the FitBit was giving me step credit for cycling. So, in order to test this out, I checked what steps I was given credit for when involved in various different activities.

This is my results:

• Running - I checked my FitBit before and after a few runs and found that it gives me credit for almost exactly 1,000 steps per km. This seems consistent with the fact that I probably have a stride length of roughly 1 metre while jogging.
  
• Walking - when I go for a walk with the dog, I get credit for 8-10% more steps than if I ran the same distance. This probably means that my stride is shorter while walking than it is when running. The fact that it is slightly more variable is probably because I wander around a little while walking the dog, but rigidly put one foot in front of the other when running.
  
• Cycling - Twice daily I cycle the route between home and work so I have lots of chances to measure how many steps I get credit for when doing this journey. My FitBit seems to give me credit for somewhere between 700 and 1,800 steps for this 8 km journey. This is a strange reading from a few points of view:
  
• I do tend to take a few real steps at the start and end of the journey as I wheel the bike in and out of the shed, but this would probably be 10s of steps not 100s.
  
• On the other hand, it could be argued that each time I push the pedals up and down this is effectively the same motion as a step. But, if was what the FitBit counts as steps, then it is seriously underestimating the number of pedal turns that I would do.
  
• The other thing that is strange is that there is so much variation in step count - some journeys were more than twice as many steps as others despite the fact that I was going on exactly the same route.
  
  I think (but can't prove) that the key factor is how often I wobble. If I was cycling nice and steady on a flat road, my wrist would be staying still on the handlebars and the FitBit would count that as resting. However if I was wobbling from side to side, the up and down motion of my left wrist would probably move up and down in a similar way to if I was walking. The amount of wobbling I do would probably vary from day to day depending upon traffic and other factors.
  
• Working - I work as a software engineer so my work involves virtually no physical movement (which is why it is a very unhealthy occupation). The FitBit should not give me credit for any steps while I work. I tested this one day by looking at my step count when I came into work and then checking it 1 hour later.
  
  The FitBit said I took 7 steps in that hour - initially I thought it was giving me false credit, but then I remembered that during the hour I had left my seat once to look at a colleague's screen to help solve a problem he was having (his desk is beside mine so 7 steps sounds right for there and back). Therefore I think that it correctly gives me no credit for activity while working.
  
• Sleeping - I know that I tend to toss and turn a lot during my sleep. In fact my FitBit sleep data tends to show frequent periods of red in the middle of the deep blue sleep time (the red periods are times when the FitBit thinks I am awake/restless - I am not sure how it decides that). I thought that the FitBit might give me credit for steps when I was really just tossing about restlessly during the night. However, it seems that it doesn't.
  
  The FitBit resets its step counter at midnight and I would normally be asleep by then (or at least lying in bed trying to sleep). So I decided to check my FitBit on a few mornings straight away after I got up out of bed. It would normally give me credit for a small number of steps, but never more than about 10-15 steps per night.

Overall I think that the FitBit gives a reasonably accurate measurement of activity level even if it is off significantly when measuring cycling.

I have read on-line about a legal case over how inaccurate the FitBit heart rate monitoring function can be. I don't doubt that the readings are often wrong, but I think the real problem is why were people expecting that a simple wrist-watch with two LEDs built into it could really provide accurate readings? (Medical professionals use complex and expensive devices to take the same reading.)