In this paper, attempt to study effects of extreme observations on two estimators of finite population total theoretically and by simulation is made. We compare the ratio estimate with the local linear polynomial estimate of finite population total given different finite populations. Both classical and the non parametric estimator based on the local linear polynomial produce good results when the auxiliary and the study variables are highly correlated. It is however noted that in the presence of outlying observations the local linear polynomial performs better with respect to design mean square error (MSE) in all the artificial populations generated