I often find that Sod's Law applies here.
For instance, when I'm installing additional smoke alarms in modern properties that have sheeted flooring and plasterboard walls throughout, neither of which the property owner wants damaged in any way, it is very rare to find that a loop-in system has been installed via ceiling lighting points. The obvious advantage with the ceiling loop-in method, is, of course, that the wiring for the alarm can be taken direct from the ceiling lighting point and fed through the ceiling void without any ceiling damage being incurred. But when the feeds are located at switch points, it is more often than not next to impossible to install wiring between the switch and the alarm without having to open up the adjacent wall and the ceiling, in which case the alarms would need to be installed on the wall rather than the ceiling.
Conversely, when the job is a light fitting changeover in older properties, I often find that a ceiling loop-in system is present. As other posters have pointed-out, it can be fiddly working at a height with a fancy light fitting and three pairs of 1.5 T&E. In this particular instance there is a distinct disadvantage and you end-up wishing that the feeds were at the switch rather than the light.
Both approaches or a combination thereof have advantages and disadvantages. In the end, it all comes down to what type of job you're doing and whether or not the law of Sod comes into play!