In Response Surface Methodology, the relationship between the response variable and the independent variables in a restricted area of operation was used models which approximated by a second order polynomial function. While the model parameters are usually estimated by the Least Squares Method (OLS). However, this method is highly sensitive to outliers, because outliers are very likely to produce a substantial residual and often affects the resulting model. Thus, if this model is implemented

more »

... sing the resulting model estimate to be biased and resulted in errors in the determination of the actual optimal point. Therefore, we need a model of the response surface that is resistant to outliers. As an alternative, we use the M-Estimation, for estimating the parameters of the response surface model. In this paper, we demonstrated the use of M-estimation in the response surface model with a case study of tire tread compound problem. From the simulation results has been shown on the tire tread compound problems. M-estimation gives better 1804 Edy Widodo, Suryo Guritno and Sri Haryatmi results compared to OLS. So that, in case there is data that contain outliers, M-estimation can be used as an alternative estimator.