Vol. 18 No. 1
Detection and impact of outliers on bell pepper (Capsicum annuum L. var. grossum Sendt.) in field condition
Author(s): M. Roy, A. Saha, J. S. Basak, M. Saha, Md M. H. Middya, S. Roy and A. Majumder
Abstract: The presence of abnormal observations (even a single abnormal observation) in any agricultural field experiment may cause a serious damage to the whole experiment. Actually, the presence of such observations (or a single observation) will make a deviation from the normality assumption of ANOVA model. Such observations are considered as outliers in Statistics. Cook (1977); Andrews and Pregibon (1978) developed methodologies to detect the presence of outliers in regression model. Later Bhar and Gupta (2001) modified those statistics for detection of outliers in experimental design data set. Those modified statistics have been applied in field experiments on bell pepper (Capsicum annuum L. var. grossum Sendt.) for detection of presence of outliers in the data set. In all, more than 10 experiments in RBD layout for different parameters of the crop were judged for presence of outliers and only 6 of them were found with presence of significant influential outliers. After detection of outliers, the outliers were removed and the data set of each experiment was undergone through the missing plot analysis. It has been observed that in all of the cases after removal of detected outliers, the analysis gave more efficient and effective result. The results of outliers, the mean values of treatments (bell pepper) and error mean square (before and after removal of significant outlier) for each experiment have been tabulated.