425
Process Capability: Engineering and Statistical Issues
same interval width. Conversely, estimates of process capability for non-normal processes are much morę variable and this further underscores the need for checking the necessary assumptions. It should be noted, however, that these distributions were selected only for illustration purposes.
Finally, although the entire October 1992 issue of the Journal of Quality Technology concemed the topie of process capability, it should be noted that the editor wams against the use of process capability indices. In his editorial, Nelson (1992), States that devoting an entire issue to this topie
... should not be construed as implying that JQT is promoting the use of capability indices. In fact, it is elear from a statistical perspective that the concept of attempting to characterize a process with a single number is fundamentally flawed.
As will be discussed below, considerable study is first needed before simply calculating a process capability index. Indeed, we suggest that an index should not be called a "process capability index" unless it is calculated from the data collected in a capability study and unless the analysis that we describe is performed and reviewed prior to the calculation of the index.
What Is a Process Capability Study?
The Western Electric (1956, page 34) Statistical Quality Control Handbook defines the terms ‘process capability study” and ‘process capability” as follows:
... the term “process capability study” means the systematic study of a process by means of statistical control charts in order to discover whether it is behaving naturally or unnaturally; plus investigation of any unnatural behavior to determine its cause; plus action to eliminate any of the unnatural behavior which it is desirable to eliminate for economic or ąuality reasons. The natural behavior of the process afler the unnatural disturbances are eliminated is called the “process capability. ”
Thus, the implication is that a process' capability should only be assessed after the process has been brought into statistical control and after special causes of variability have been eliminated. Of course, once that has been done, the issue of how best to assess the process' capability still remains. The Handbook goes on to say (pages 45-6):