The Use of Statistical Analysis - In the Korean Journal of Otolaryngology-Head and Neck Surgery -.
- Author:
Myung hee SHIN
1
;
Young Ik SON
;
Yang Sun CHO
;
Hyo Yeol KIM
;
Byung Suk HA
;
Sun Il KIM
Author Information
1. Department of Otorhinolaryngology-Head and Neck Surgery, Samsung Medical Center, ungkyunkwan University School of Medicine, Seoul, Korea. yscho@smc.samsung.co.kr
- Publication Type:Original Article
- Keywords:
Statistics;
Otolaryngology
- MeSH:
Neck*;
Otolaryngology
- From:Korean Journal of Otolaryngology - Head and Neck Surgery
1999;42(11):1364-1369
- CountryRepublic of Korea
- Language:Korean
-
Abstract:
BACKGROUND AND OBJECTIVES: The statistical analysis is an essential procedure with the aim of ensuring that the results of researches are based on evidences rather than opinions. We evaluated the frequency, adequacy, and validity of the statistical methods used in the otolaryngology journals, identified types of errors in statistical analysis, and suggested the proper usage of the statistical methods. MATERIALS AND METHODS: 265 articles published in the Korean Journal of Otolaryngology-Head and Neck Surgery in 1997 were reviewed and evaluated. Each article was evaluated in terms of the adequacy, validity and frequency of statistical usage according to the preformed category. RESULTS: 104 out of 265 articles used analytical statistics, and the commonest statistical methods used in the articles were t-test. ANOVA and x(2)-test were the second and third commonest methods, respectively. In only 36.6% of articles, proper statistical methods were used for their data. Misuse of statistical methods (31.4%) was the most frequent errors found in the articles. 64.6% of the misuses was the adoption of a statistical test inconsistent with data characteristics, 18.8% of them was inappropriate use of parametric methods, and 10.4% of them was the use of x(2)-test when an exact test was needed (3.3%). CONCLUSION: Most of statistical errors found in the journals were misuses of statistical methods related to basic statistics. Closer communication between researchers and statistical analysts, and closer editorial attention to statistical methodology in the review process would improve the validity of articles.
