The principle difference between slotted and unslotted networks is regularity. Frames in a slotted network are always the same size (or have a bounded maximum size); frames in an unslotted network can vary in size in unpredictable (to the network) ways. Once regularity is in place, simplicity (or greater simplicity) follows in timing constraints, hardware design and protocol requirements, mostly by eliminating the need to find and handle corner cases.
Regularity and simplicity are pleasant features to have in a network being designed, so in the absence of external influences to the contrary, they are to be preferred over unpredictability and complexity.
A lot of the answers didn't notice that the question was phrased to emphasize the network designer's point of view, not the network user's point of view. It is true that unslotted networks don't require that PDUs be fragmented into slots, and that makes the network easier to use, but it also makes the network harder to design.
The definition of isochronous traffic makes this a difficult question. However, the essence of the answer is the idea of time or delay sensitivity: isochronous traffic is delay or time sensitive, while statistical traffic isn't. (It may be possible to handle delay sensitive statistical traffic at the end-points, where lots of buffering can be used cheaply, but it unrealistic - and, as it turns out, futile - to assume that much buffering in the network.)
This page last modified on 6 December 2004.