McGuigan reviewed all papers published in the British Journal of Psychiatry in 1993 and presenting numerical results.
Of the published 248 articles 164 (66%) presented numerical results. Of these articles 65 (40%) contained statistical errors. Many errors were not serious, but some were serious enough to cast doubt on the conclusions.
The author concludes that the statistical error rate is unacceptably high.
McGuigan SM. The use of statistics in the British Journal of Psychiatry. Br J Psychiatry 1995;167:683-688.