Accuracy of buffers and self-drawn neighbourhoods in representing adolescent GPS measured activity spaces: An exploratory study.

Abstract

There continues to be a lack of understanding as to the geographical area at which the environment exerts influence on behaviour and health. This exploratory study compares different potential methods of both researcher- and participant-defined definitions of neighbourhood reflect an adolescent's activity space. Seven consecutive days of global positioning system (GPS) tracking data were collected at 15s intervals using a small exploratory adolescent sample of 14-18 year olds (n=69) in West Yorkshire, England. A total of 304,581 GPS tracking points were collected and compared 30 different definitions of researcher-defined neighbourhoods including radial, network and ellipse buffers at 400m, 800m, 1000m, 1600m and 3000m, as well as participant-defined self-drawn neighbourhoods. This exploratory study supports emerging evidence cautioning against the use of static neighbourhood definitions for defining exposure. Traditional buffers (network and radial) capture at most 67% of activity space (home radial), and at worst they captured only 3.5% (school network) and range from capturing between 3 and 88% of total time. Similarly, self-drawn neighbourhoods captured only 10% of actual daily movement. Interestingly, 40% of an adolescent's self-drawn neighbourhood was not used. We also demonstrate that buffers capture a range of space (22-95%) where adolescents do not go, thus misclassifying the exposure. Our exploratory findings demonstrate that neither researcher- nor participant-defined definition of neighbourhood adequately captures adolescent activity space. Further research with larger samples are needed to confirm the findings of this exploratory study.