STATISTICAL ERRORS AND HOW TO AVOID THEM, OR NOTES ON CORRECT ANALYSIS OF QUANTITATIVE DATA IN BREEDING
2021
Sinjushin, A.A.
All stages of breeding process involve acquisition and processing of quantitative data. The correct statistical analysis provides a possibility to draw adequate conclusions and make results of different researches comparable. When analyzing papers published in ‘Zernobobovye i Krupyanye Kul’tury’ (‘Legumes and Groat Crops’), we found several frequent errors in statistical analysis. Among them, one may list use of parametric methods (such as Pearson’s correlation coefficient or Student’s t-test) without preliminary evaluation of distribution of variables, ineffective choice of ‘most typical’ value (mean instead of median), use of means without illustrating the amount of variation, demonstrating the results of statistical analysis without estimates of their statistical significance etc. Some of analytical procedures were exemplified by own data from phenotyping of germplasm collection of a garden pea. An algorithm was also proposed which may reduce the risk of incorrect use of statistical methods
显示更多 [+] 显示较少 [-]