계명의대 논문집 10년간(1982-1991)의 통계적 방법론에 대한 고찰 = Review of the validity of ststistical methods in the Keimyung University Medical journal, 1982-1991

다국어 초록 (Multilingual Abstract)

Authors reviewed the 204 original articles published in the Keimyung University Medical Journal during 1982－1991 to give a status report of validity of statistical methods for the purpose of improving the validity of statistical applications. Checklists were assessed by the independent reviewers
from the department of statistics of a graduate school. Number of cross-sectional study was 81(39.7%) and that of animal study was 61(30.0%). Cohort study was merely 4(1.9%) which appeared first during 1988-1989.
Eighty-seven articles used no statistical method or descriptive statistics only, frequency of which decreased yearly and 68 articles(28.2%) used t-test Nonparametric test, survival analysis and multiple linear regression were used once respectively. Orphan p where no statistical method had been specified
and only the p value given was presented in 25 article(10.4%) which showed little fluctuations yearly. Representative sampling method was employed only in one article and no description of sampling method was noted in 28 articles(19.6%), but the yearly frequency was decreased steadily. Descriptions of statistical methods used in materials and methods was noted in 73 articles(59.8%), number of which appeared to increase yearly. Basic assumptions of the statistical methods used were discussed in 3 articles only(2.5%)Twelve articles (9.8%) described the significance level declared
statistically significant, showing yearly decreasing frequency. Applications of the statistical methods appeared to be appropriate in 30 articles(24.6%), frequency of which showing increasing yearly trends. These results suggest that the validity of statistical methods used appears to be seriously compromised in this period and has much to be done to improve the current situations.