The techniques of exploratory data analysis include a resistant rule, based on a linear combination of quartiles, for identification of outliers. This paper shows that the substitution of the quartiles with the median leads to a better performance in the non-Gaussian case. The improvement occurs...