Error root mean square is a method in which the general mean and error variance are obtained using the S/N ratio of a nominal-is-best characteristic, and their ratio to the variance of the unit space is used. In the RT method, a distance is obtained using variation and sensitivity, which is a mean effect. This method is conceptually similar to error variance. Liver disease data were used to compare error root mean square with the RT method. As different items of liver disease data uses different units, the data were standardized and the differences between the results of analysis with and without standardization were studied. A strong correlation between the two was found. A normalized scale is presented in which the error variance of the unit space is calibrated to obtain an error root mean square.