Sen, Jurečková and Picek (2003) extended the Shapiro-Wilk test to testing the normality in the presence of nuisance regression and scale and showed that the asymptotic null distribution (under the hypothesis of normality) of the extended test criterion is equivalent to that of the original Shapiro-Wilk test in the location-scale model. The simulation study in the present paper shows that both criteria are close even for moderate data sets.