start with descriptive distance-based statistics using something like the Near tool (Euclidean distance as a first run, Network distance if you have the network analyst.
Have you read any of the literature to support the hypothesis? There is some (fast food and proximity to schools)
