I must have touched a nerve. I did like I said I was going too. I went to the library and did a computer search of psylit from 1974-current. A total of 24 articles were retrieved. Of those.
Buckner et. al. (1987). Investigated rater reliability of NLPers to judge presence of eye movement and match it to subject's self-report of the presence of senory modality thought. Visual and auditory channels were consistently identified, however, kinesthetic componets were not (p>.85). Given the p-value, it would be interesting to explore if something in NLP actively discourages this identification of this modality.
Salas et. al. (1989) found no empirical support emerged between type of modality eliciting questions or perceptual predicate usage and direction of gaze.
Einspruch et. al. (1985) examined 39 empirical studies pre-dating 1984, which contained critical and fatal design and methodology errors. Actually a nice article for anyone wanting to design research to test NLP.
Einspruch et. al. (1988) examined the NLP treatment of phobias. Eight weeks of treatment. Marked improvement was found, however, no complete cure was demonstrated.
Krugman et. al. (1985) tested the NLP single treatment for anxiety. Krugman's research design was good and seemed to address Einspruch's design concerns. No empirical support was found.
Helm (1990) demonstrates linguistic difference sex x race.
Helm (1991) examined grade acheivement and sex x sensory modality. He tracked the same students for two years. Kinesthetic attained the highest grades followed by visual and auditory. No demonstration of NLP's ability to intervene.
The rest of the citations concerned training issues and techniques.
Given the amount of empirical support demonstrated by behavioural techniques for phobia treatment, and the lack of suport for NLP why would someone prefer NLP?
[ I think that any therapist who takes the time to learn and actual try out some of the simpler NLP techniques will have no problem in answering the question at the end of Dow's article. However this should not make us forget that more careful, published research in NLP would definitely be desirable and useful not only in helping to convince mainstream psychologists of NLP's value, but also useful to NLP itself. It would help NLPers be more careful in separating those parts of NLP which we know for sure work, because they are easily demonstrated, from those parts which we've always taken for granted but whose effectiveness is more difficult to test. -- LL ]