ABSTRACT
This chapter describes the estimation of process capability based on the simultaneous analysis of multiple variables. When a process is characterized by more than one variable and there are significant correlations among those variables, estimating capability by analyzing each variable separately can be misleading. Simultaneous modeling of multiple variables using a multivariate normal distribution allows overall process capability to be estimated properly. Multivariate capability indices including MC pk are presented. Multivariate tests for normality are described, as is the analysis of nonnormal multivariate data through the use of power transformations. Finally, the construction of multivariate statistical tolerance limits is discussed.
