A critical survey of scientific methods in two psychiatry journals

In a critical review of 29 analytical papers from the 1985 Australian and New Zealand Journal of Psychiatry and the American Journal of Psychiatry, examples of 10 methodological errors were found. Twenty-three papers contained at least one error. The commonest error in design was the failure to assess outcome independently of knowledge of the subjects' groups. The most frequent error in analysis was the failure to control for the increased type I error rate when multiple comparisons were made. The high frequency of statistical errors in published articles indicates a need for research methodology to be included in postgraduate psychiatry training. The limitations of statistical designs highlight the importance of alternative investigative models in psychiatric research.