The Investigation of Statistical Descriptions in the Original Articles of Journal of the Korean Radiological Society

ABSTRACT

PURPOSE: The purpose of this study was to identify errors in statistical descriptions in articles published in the journal of Korean Radiological Society and to suggest guidelines for the reduction of such errors. MATERIALS AND METHODS: In order to investigate statistical errors, 139 of the 514 original articles published between July 1994 were analysed. To assess the validity of statistical descriptions, five items were evaluated. The five items included descriptions of basic data, statistical tests and their results, tables, articles which had presented two or more statistical methods or results, and the Chi-square test. RESULTS: The percentages of correct descriptions of basic data, statistical tests and their results, tables, articles which had presented two or more statistical methods or results, and the Chi-square test were 86.3%, 74.8%, 24.4%, 76.7% and 79.4%, respectively. In 50 of 139 articles (36.0%), statistical descriptions were correct. CONCLUSION: More than half of the articles assessed contained statistical errors; inaccuracies of this kind might raise doubts about the objective validity of an article and the authors should there fore be careful when performing a study which needs statistical descriptions.