This paper investigates the response model approach for the dynamic robust design problem. We derive relationships between the effect estimates of the loss model approach and those of the response model approach. We show that the bias problem for the static case solely exists in the estimation of process variance and does not exist in the estimation of the slope and intercept parameters. The two analysis approaches are compared by use of a real example.