Skip to content

Journal article

Why Carrying Shopping and Pushing a Stroller Confuse a Step Count

A step total depends on the signal and how it is interpreted.

Wrist step estimates can change during stroller pushing, carrying and household tasks. A 2026 validation study shows why the algorithm matters.

  • By Editorial Team
  • Published
  • 2 minutes read
Responsible editor
Illustration of a handheld screen showing a walking figure beside abstract footprints and a sensor-like circle.

Imagine finishing a walk with a stroller, carrying shopping home, or spending the morning moving between household jobs. Then you check your step total and find a number that does not match your impression of the day. It is an awkward moment: how should that number fit with the activity you remember?

A tracker estimates steps from movement signals rather than watching the feet touch the ground. The sensor's location and the rules used to interpret its signal both matter. A 2026 study applied five counting algorithms to the same wrist-device data and checked their estimates against steps identified in video recordings. This separated the movement recorded by the sensor from the way software counted it.1

The direct-observation analysis included twenty adults wearing wrist and thigh research devices. Their 35 recorded sessions covered about 97 hours of daily life. Researchers labeled the activities as well as counting steps, so they could connect discrepancies with what a person was doing. Conventional walking and running were generally easier to count accurately than modified walking and mixed activities.1

Modified walking included carrying a load and using stairs. Several algorithms undercounted these steps, while one overcounted them. Mixed movements such as household tasks could also generate excess counts. Closer inspection of unusual results identified difficulties during stroller pushing, hiking, typing and eating. The problem was therefore not simply too few counts: the direction of error depended on the task and algorithm.1

This also explains why an overall total can be misleading. Missed steps during one activity and excess counts during another can partly offset each other, while neither activity has been measured accurately. The observed pattern of errors matters alongside how closely a total matches the video count.1

Devices were worn for seven days, but video verification covered selected sessions. The separate comparison of daily device totals was not continuous observation of every step. These findings concern particular research sensors and processing methods in a small sample; they cannot rank current consumer watches or provide one correction for every walk with occupied hands.1

When you look back on a busy morning, the walk has details of its own: the route you took, the shopping you carried and the person you were with. Those are the things you might mention when someone asks what you have been doing. A step total belongs in that everyday conversation, alongside the rest of your account.

References

  1. Hyde ET et al.. Performance Evaluation of Device-Based Algorithms to Estimate Step Counts in Free-Living Adults Compared with Direct Observation. Medicine & Science in Sports & Exercise, 2026.

Continue reading

Conversation

Discussion

0 comments

No comments yet. Start the conversation.

Join the discussion

No account or email is required. Your display name and comment will be public.