In this paper a comparison between the parametric and the non-parametric methods used to find the constant C Ralpha, which is part of the Hayter and Tsui (1994) methodology to control multivariate processes, is presented. The Hayter and Tsui method is an alternative to the Hotelling T²statistical test with the advantage that it identifies automatically which quality characteristic is responsible for the out-of-control situation. Some multivariate processes p=2 and p=4 dimensional were simulated and the results showed that the parametric method was better than the non-parametric one. However, for sample sizes higher ou equal to 5000 they were equivalent. The results go against the Hayter & Tsui's who suggested that the non-parametric method could be adopted for sample sizes higher or equal to 500.
Multivariate processes; multiple comparisons; quality control
