5.3.2. Cadence

The subjects' responses to the follow-up question provided valuable insight as to why they perceived certain cadence measurements as inaccurate. These reasons are sometimes explained by improper use of the system. For example, as a follow-up to choosing "cadence" as inaccurate in the survey, one subject responded: "Cadence too low; possibly because I held phone in hand". The cadence algorithm was designed to work around the hip area or on the shoulder, which explains why during this run the average cadence of the runner was as low as 72 steps/min.

In another example, a subject reported that "My measured cadence was also lower than expected—I was mostly running on beat to songs that had 150+ bpm". The measured cadence of the run is depicted in Figure 9b. The average cadence of the run was 111 steps/min, which includes the segments in the beginning of the run as well as some segments when the runner slowed down (or perhaps momentarily stopped running). After excluding these data points, the average increased to 124 steps/min, which is still lower than the subject's self-reported cadence of 150 steps/min. For reference, Figure 9a depicts the speed during the same run, which shows that the runner would slow down in many instances. One potential reason for those momentary slow-downs is the urban path that the runner selected, which is corroborated by the run's GPS data (the path passed through many street intersections).

**Figure 9.** (**a**) The estimated speed (m/s) during run 54 by subject b01k1o; and (**b**) the estimated cadence (steps/min) during run 54 by subject b01k1o.
