An assessment of statistical errors in articles in the Korean journal of veterinary research

Park, S.I. . Department of Internal Medicine, College of Veterinary Medicine

An assessment of statistical errors in articles in the Korean journal of veterinary research

1999

Park, S.I. (Seoul National University, Seoul (Korea Republic). Department of Internal Medicine, College of Veterinary Medicine)

The purpose of this study is to assess the suitability of the statistical techniques employed in papers published in the Korean Journal of Veterinary Research from March 1997 to March 1999 and it is hoped that the critical assessment may be of help to other researchers preparing their works for publication. Of the 246 original papers 94 were inclluded in the analysis. of 62 papers with the measure of central location and dispersion of data 34(54.8%) used them correctly:9 (39.1%) of 23 for t-test; 1 (33.3%) of 3 for correlation analysis; 7 (43.8%) of 16 for analysis of variance (ANOVA); 5 (62.5%) of 8 for chi-square test; 44 (71%) of 62 for description of p-value. A number of papers employed ANOVA did not perform subsequent analysis of multiple comparison. Compared to the results of others, relatively higher proportion of papers in the present study was evaluated as appropriate analysis. The reason is that papers described insufficiently on the study design were not included, and evaluation items were restricted to the cases violated seriously ingerent assumptions for each statistical technique. Statistical misuse or abuse appeared in the study is due to lack of knowledge on statistics and short of its importance for improvement the quality of paper. Because an inappropriate analysis can lead the readers to misunderatand on findings, observed statistical analyses must be valid,and correctly undertaken. It is suggested that more intensive statistical refereeing are needed,and the communication should be allowed for the controversial points.

Показать больше [+]