Statistical methods used in articles of the Korean Journal of Dermatology

Background: The Korean Journal of Dermatology, a well-known dermatology journal in Korea, has been an essential medium for dermatologists and a symbol of the Korean Dermatological Association. The members of the association have an obligation to improve the quality of the journal. The statistical analysis is associated directly with the study design and completeness of an article. However, there has been no report on the statistical analyses used in the articles of the Korean Journal of Dermatology. Objective: To investigate the usage of statistical methods in the journal and evaluate their accuracy and correctness. Methods: A total of 264 original articles from 30 recent volumes of the Korean Journal of Dermatology (2003, vol.1 to 2005, vol.6) were collected and analyzed for statistical methods. The descriptive statistics were excluded from the study. The methods were considered as 'correct', 'not-correct', or 'impossible to determine' in order to evaluate the correctness of the analyses. Results: Of the 264 original articles, 126 (47.7%) articles included statistical analysis. The authors most frequently the used t-test and contingency table method. The incorrect analyses appeared in 30 (23.8%) articles, and their most frequently cited reason was their unwillingness to use the nonparametric analysis. Conclusion: Recently, the statistical methods in about one fourth of inferentially analyzed original articles of the Korean Journal of Dermatology have been found to be incorrect, and the reasons for the incorrectness appear not to be so difficult for authors to rectify.