Alf,
I agree- its very difficult to "prove" that NTS are critical to a safe outcome, but there is plenty of evidence to support the proposition. Every accident or incident that is put down to pilot error adds weight to this idea.
LOSA data is of course subjective, but looking at the papers produced by Helmreich et al (University of Texas), I think anyone can see that there is a lot of useful information available provided the observers are sufficiently well-trained and capable. If nothing else, it can help to identify trends if there is a history of, for example, unstabilised approaches using a particular procedure at a particular airport. There may be a plethora of different reasons why pilots have got the aircraft into this situation, but unless someone has flagged up the fact that there is a recurrent issue not related to aircraft serviceability, then it will not get addressed until it is too late.
As Mesh says, the CAA require technical consequences before NTS causes a failure on a line check. This does not, of course, prevent a comment or exploration of CRM issues in a debrief even if the line check was passed.
There is always a problem with debriefing an event, and having your episodic memory massage the facts. But, at least a well-facilitated debrief (particularly if video debriefing stations are available), will allow you to re-visit the events and allow you to formulate better ways of coping with particular situations.
There is always something to learn - there is always someone who has had an experience that you have no come across yet.