(a) Grouping is usually by route when there is a small number of routes or by groups of routes that are similar.
(b) You are going to determine the similarity of service units on your expectations from your prior knowledge of your service.
(c) The most efficient grouping depends on other efficiency options you consider:
(1) If your sampling plan follows the base option, define groups so that within each group, they are similar with respect to PMT per service unit. For example, you may group heavily used long routes separately from lightly used shorter routes.
(2) If your sampling plan follows the APTL option, define your groups so that within each group they have similar APTL. For example, if your express route customers travel particularly long distances, you may separate your express and local services as two groups. As another example, if some of your local routes are particularly longer than the other local routes, you may further divide your local service into two groups.
(3) If your sampling plan follows the PPMT option, define your groups so that the routes within each group are similar with respect to the fraction of a route’s length that passengers typically ride. For example, if on some routes passengers tend to ride most of length of the route (as is the case for some express routes) while on other routes passengers tend to ride for only a small part of the route’s full length, separate those routes into different groups. Consider another example. If a system has a mix of radial routes (routes with one end in the downtown), diagonal routes (routes that pass through the downtown), and crosstown routes (routes that don’t pass through the downtown), the average fraction of route length that passengers ride tends to be smaller on the diagonal and crosstown routes and longer on the radial routes. In this case, radial routes could be grouped separately from diagonal and crosstown routes.