Assessing Robustness of Inference in Symmetrical Nonlinear Regression Models