simply discarding such an observation is usually not recommended.serious efforts should be made to verify the validity of the data values or to determine if some unusual event did occur. However,for purposes of illustration here,we do reestimate the regression without that observation. The results of the analysis are given in table 8.17 ,where it becomes of the regression analysis.The residual variance has decreased from 366 to 106,the F statistic for testing the model has increased from 134to448,the estimated coefficients and their p values have changed drastically so that now x3 is the dominant independent variable,and the degree of multicollinearity between x2 and x3 has also increased.In other words,the conclusions about the factors affecting production have changed by eliminating one obserrvation