Research Article: Evaluating the Feasibility of Measuring Travel to School Using a Wearable Camera

Date Published: November 1, 2012

Publisher: Elsevier Science

Author(s): Paul Kelly, Aiden R. Doherty, Alex Hamilton, Anne Matthews, Alan M. Batterham, Michael Nelson, Charlie Foster, Gill Cowburn.


The school journey is often studied in relation to health outcomes in children and adolescents. Self-report is the most common measurement tool.

To investigate the error on self-reported journey duration in adolescents, using a wearable digital camera (Microsoft SenseCam).

During March–May 2011, participants (n=17; aged 13–15 years) from four schools wore wearable cameras to and from school for 1 week. The device automatically records time-stamped, first-person point-of-view images, without any action from the wearer. Participants also completed a researcher-administered self-report travel survey over the same period. Analysis took place in November 2011. Within- and between-subjects correlation coefficients and Bland-Altman 95% limits of agreement were derived, accounting for the multiple observations per individual.

Self-report data were collected for 150 journey stages and SenseCam data for 135 (90%) of these. The within-subjects correlation coefficient for journey duration was 0.89 (95% CI=0.84, 0.93). The between-subjects correlation coefficient was 0.92 (95% CI=0.79, 0.97). The mean difference (bias) between methods at the whole sample level was small (10 seconds per journey, 95% CI= −33, 53). The wide limits of agreement (±501 seconds, 95% CI= −491, 511) reveal large random error.

Compared to direct observation from images, self-reported journey duration is accurate at the mean group level but imprecise at the level of the individual participant.

Partial Text

Physical activity is associated with important health outcomes in children, including body composition, type 2 diabetes, and cardiovascular fitness.1–6 Active travel to school, including walking and cycling, can be an important contributor to physical activity levels.7,8 Conversely, travel in motor vehicles is a sedentary behavior representing a lost opportunity for physical activity.9

Participants were volunteers aged 13–15 years (n=17; 11 girls, 6 boys). Data collection took place between March and May 2011.

The current study demonstrates for the first time that a wearable camera is a feasible technique for use in a school travel setting, for multiple days of data collection. The obtained images give an objective assessment of travel mode and an accurate and reliable measure of duration. This study also demonstrated some issues associated with the device: there are particular settings with which participants are not comfortable wearing the camera (notably at a friend’s house visited during the journey home); journeys cannot always be determined when light levels are very low; images can be lost when the lens is obscured by clothing; the device can be forgotten for some journeys; and the 10-second epoch between image capture introduces a small error on calculation of duration. Protocols should be developed to address these issues and minimize data loss.