Statistical methodology in general psychiatric journals.

Abstract

Development of psychiatric research methods requires systematic review of their status. This study describes the frequency with which various statistical research designs and methods are reported in general psychiatric journals. All original research articles in four psychiatric journals in 1996 were reviewed: The American Journal of Psychiatry (AJP), The Archives of General Psychiatry (AGP), The British Journal of Psychiatry (BJP) and the Nordic Journal of Psychiatry (NJP). Evaluation included 448 regular articles, which reported original research findings based on systematic collection and statistical analysis of research data. Aspects measured were research design, statistical methodology, description of procedures and presentation of results. Percentage frequencies of reported statistical procedures were compared between the journals. The policy of determining statistical significance was still the most generally used method of conducting research. Compared to earlier reviews, readers of psychiatric reports in 1996 more frequently encountered computer-dependent multivariate techniques. There were differences in the utilization and reporting of statistical procedures among the journals. The authors of psychiatric journals might apply these results in designing their psychiatric research to present intelligible and compact analysis combined with a high quality presentation technique, thus being in line with the policy and presentation of the leading psychiatric journals.