Effect of cross-section models on the validity of sterile neutrino mixing limits