It seems that the German railroad Deutsche Bahn excludes those trains that break down en-route (or that never even start) when calculating arrival statistics…and there are a lot of such trains. (via Cold Spring Shops)
So it would be pretty inappropriate to compare DB’s schedule performance, calculated in this way, with the schedule performance of a railroad that did include broken-down or never started trains in the late-arrival category.
There are a lot of cases in which statistics may not be actually comparable in the way that they are assumed to be…for example, it seems that the US calculates infant mortality in a different way from most other countries, owing to the different treatment of premature births.
Journalists, in general, fail completely in explaining what the numbers that they are citing actually mean…probably because in most cases, they don’t understand themselves.