Almost all reports of sleep-related behavior are anecdotal --
-- because few of us have an accredited sleep disorders specialist sitting in a corner of our bedroom, taking notes every night. Likewise, few of us go to bed wired up to an EEG and various other monitoring devices, as is done during clinical sleep studies. If a sleep-related behavioral syndrome is sufficiently widespread, troublesome, or diagnostically useful, monitoring techniques may be devised that can record the event as it occurs; but there's no guarantee that that will happen.
To take just one example, cataplexy, one of the basic symptoms of narcolepsy, can be elusive in a clinical context. For many narcoleptics, episodes of cataplexy are triggered by specific emotional stimuli, which are different in every patient. Someone whose cataplexy is set off by emotions that don't normally occur during visits to a doctor's office is going to have to rely on anecdotal reporting. The only alternative is to try to arrange for circumstances that will prompt the desired response while the doctor is watching. If that sounds like the setup for a sitcom episode, let me assure you that the results can resemble one.
Anyone who thinks that "supported only by anecdotal evidence" is sufficient grounds to dismiss reports of some odd somnolent behavior doesn't understand sleep disorders.