Detection and Treatment of Outliers in Experimental Design: Real Data for Completely Randomized Design

Abstract

The presence of outliers values in the data leads to errors in statistical analysis due to the use of traditional methods of calculation, so it is necessary to switch to new methods that deal with these outlier values so as to ensure the accuracy of the calculations to the proper statistical analysis, and in this research resorted to the method adjusted boxplot to detect outlier values and then deleted and re-statistical analysis data have been used for a realistic agricultural experiment to completely randomized design in the College of Agriculture Wasit for 2017 has shown the result of statistical analysis that there is a difference in the results before and after deleting outlier values.