Comparison of the diagnostic methods used for detecting influential observations in multiple regression

In regression, an influential observation and influential sets would cause noticeable differentiations on fitted values. Since these differentiations decrease explainable of model, detecting the influential observation or the influential sets in data is important for efficiency of regression analysis. In this study DFFITS, DFBETAS, COVRATIO, Cook Distance, S statistics and graphical technique used for detecting influential observation or influential sets are examined. These methods are compared with regard to ratios of detecting influential observation and influential set in data which includes an influential observation or influential sets.