Models, patterns, psychology and science

The big argument that seems to exist in psychology seems to stem from the perceived unscientific nature of the subject. The works of Freud, Jung and many of the others is based on observation of a reasonable number of snippets from anecdotal evidence from their patients. The psychologist then appears to have constructed their model of how people think, from the individual to larger generalisations.
This method is all very human in nature – we are built to be pattern-matching machines and therefore we need a model with which we do our matching. This is not a bad thing, to my mind, because this is the basis of scientific research. We construct a model, then we test it with logical ‘proof’ or with controlled experimentation. Experimentation is never taken as proof of a theoretical model, however an experiment, if repeatable, can be taken as proof that a model is wrong. Even if a model is wrong, though, it might still be useful. Newton’s theories were proven wrong (inadequate is the PC term) but they’re still useful.
With the human mind, these theoretical models are practically impossible to prove. Only experimentation and observation can be used to give them a stronger base. It is probably also true that our models are inadequate – the human mind is very complex, perhaps to the level of chaos theory. However it doesn’t stop them being useful models which have helped psychologists in turn help a large number of people.
Thinking about this, chaos theory might be a good starting point for setting the ground rules for logical thought about psychology.