Electronic Thesis and Dissertation Repository


Master of Science




Jason Gilliland


The present study examined current methodological approaches to characterize the local food environment around children in London, Ontario, assessing variations in BMI and dietary preferences in relation to the choice of food environment measure. Taking advantage of a unique dataset that collected GPS trajectories of children’s schools and homes for a large sample of children between 11 and 14 years of age, two commonly-used approaches (i.e., network buffers and Euclidean buffers), and two novel measures of activity spaces (i.e., standard deviational ellipses and α-hulls) are used as ‘geographic containers’ (i.e., areal units) to derive food outlet measures. Results showed slight to low agreement in the percent of shared area between the various containers and the α-hulls. Kappa statistics further confirmed the slight to low agreement between the food outlet measures derived from activity space containers and Network and Euclidean buffer containers. There is considerable variation in the maximum number of outlets between the various group comparisons across gender, weight status and reported food outlet visit. In addition, results from logistic models point to consistent evidence of gender differences in dietary and weight outcomes across containers, but did not support an overall clear effect of food environment measures across choice of geographic container.

When assessing the role of local food environment on children’s outcomes, studies should select the appropriate geographic container definition depending on whether the focus is on opportunities (accessibility) or affordances (exposure).