In this paper we present two new capability indexes for multivariate independent processes which are extensions of the univariate capability index. These new indexes are based on approach of Niverthi and Dey (2000) and Mingoti and Gloria (2008) and they are sensitive to differences between the specification and the process means. A comparison was performed taking into account several different scenarios which showed that the index based on the Mingoti and Glória approach performed better than the index based on the Niverthi and Dey approach, although both were adequate to evaluate the capability of multivariate non-centered processes. Confidence intervals for the true capability index was built using Bootstrap methodology. The performance of the capability indexes estimators was evaluated by Monte Carlo simulation.
Independent Multivariate Processes; Capability Indexes; Multivariate Normal; Confidence Intervals; Bootstrap
