Thirty-Five Years of Research on Neuro-Linguistic Programming. NLP Research Data Base. State of the Art or Pseudoscientific Decoration?

315 articles in the NLP Research Data Base (1974-2009); 63 (20%) published in ISI Master Journal List journals across 30 different journals; of those 63, 33 (52.4%) were empirical tests of NLP tenets, 14 (22.2%) had no empirical value, 16 (25.4%) were unrelated to NLP

All ages; NLP applied in education, therapy and business

Systematic classification of the entire NLP research base by whether findings support, do not support, or are uncertain about NLP's tenets, restricted to indexed journals. Sampling frame is deliberately adversarial-proof: the database is maintained by NLP proponents and was recommended to the author by the Polish Institute for NLP as the most comprehensive source, so it cannot be dismissed as a sceptic's selective corpus. Classification is by a single rater (the author, a founder of the Polish Sceptics Club) with no inter-rater reliability reported, and the author's rhetorical register is openly hostile - "pseudoscientific rubbish". NLP matters for education because the VAK modality scheme underpinning much learning-styles practice descends from it. CAUTION on the numbers: this paper's own ABSTRACT misreports its results, swapping the supportive and uncertain percentages. The body text (p.61) and the arithmetic are authoritative.

Of the 33 classified studies, 9 (27.3%) produced results supporting NLP tenets, 18 (54.5%) produced non-supportive results, and 6 (18.2%) were uncertain - a 2:1 ratio of non-supportive to supportive by count, which the author reports as 3:1 combining count with methodological weight. Weighting by journal impact (Polish Ministry scoring, 479 points total) shifts it further against NLP: 25.7% of the weight supportive, 58.6% non-supportive, 15.7% uncertain. The supportive studies were mostly uncontrolled pre-post designs of 21-day residential trainings; the non-supportive studies more often had control groups and multiple measures. No study unambiguously demonstrated the existence of distinct primary representational systems, and only one (Yapko 1981) supported a benefit from matching them. The author's verdict is that NLP functions as pseudoscientific decoration rather than an evidence-based body of technique.

Not applicable.

Witkowski explicitly frames his analysis as reaching "the same conclusions as the ones arrived at by Sharpley (1984, 1987), Heap (1988) and others". Sharpley 1987 classified 44 studies: 6 (13.6%) supportive, 27 (61.4%) non-supportive, 11 (25%) partial - the same 3:1 ratio against. Heap 1988 analysed 63 studies and found the representational-system and eye-movement claims unsupported; Dorn 1983 found no reliable method of assessing a primary representational system. Witkowski also reports that within his non-supportive category "most results of research from this category were replicated", whereas the supportive studies were largely single, uncontrolled pre-post designs. This is a replicated null, not an unreplicated one.