Normal Regression vs Robust Regression

Hello all,

I am doing regression analysis using Curve Fitting tool available in MATLAB. Aim is to find a polynomial function between 2 design variables and 1 response i.e. z=f(x,y). I used robust regression methods from curve fitting tool box and fitlm(,,'RobustOpts','on') function. Both of them gives same estimates for polynomial coefficients but different R-square and adjusted R-square values. So can somebody please tell me which one of them is correct? Is the R-square (for robust regression) from curve fitting tool is more accurate or one from fitlm() function is more accurate?

Thank you in advance,

Nik