One hundred and thirty one health articles (as "original articles") reported in medical journals published in Korea during the period from Jan. to Dec. 1971. were selected as the study materials for the purpose of resumption and criticism.
Of a total of 131 articles, thirty-nine were not required the statistical application for the data processing and drawing their conclusions and ninty-two, 70. 2%, were needed to apply. The 92 articles were divided into three groups by the types of study: 54 for experimental study, 30 for clinical case analytic study and 8 for field study. Each report was resumed and criticised under the statistical aspects in terms of data processing and the validity of the conclusions.
The results of resumption of the 92 articles needed statistical application are summarized as follow;
1. Of the 54 articles of experimental studies resumed, thirty-one articles, 57.4%, were considered´ acceptable. Among the 38 articles classified as clnical case analytic and field studies, only 28. 9 per cent were considered acceptable. In total 45. 7 per cent or less than half were considered acceptable.
2. Types of errors encountered are as shown in Table 3.
3. The most frequently encountered error was that the conclusions were drawn without having made any statistical test. This type of error was found in 14 out of 35 errors presented in experimental studies, 11 of 35 in clinical ca´~e analytic studies and 5 of 9 in field studies and 30 out of 97, 33.2%, in total.
|