Process Capability Analysis--Normal and General Non-Normal Distribution (Raw Data) - Specs Tab
Select the Specs tab in the Process Capability Analysis--Normal and General Non-Normal Distribution (Raw Data) dialog box to access the options described here.
Type
The Type box contains four options: Nominal ± delta; Lower, Nominal, Upper; Lower, Nominal; and Nominal, Upper. Specify which method (see below) you want to use to enter the specification range. For details regarding these methods and the meaning of the specification range, refer to the Introductory Overview.
Nominal ± delta
If you select Nominal ± delta (the default setting), Statistica expects you to enter the nominal target specification and the deviation (Delta) defining the specification range.
Lower, Nominal, Upper
If you select Lower, Nominal, Upper, Statistica expects you to enter the lower specification limit (LSL), nominal target specification, and the upper specification limit (USL). With this setting, you can thus define a specification range that is asymmetrical about the design center (nominal target).
Nominal value
Select this check box in order to specify a nominal value; clear the check box to omit the nominal value.
Nominal value, ± Delta value, Lower specification limit, and Upper specification limit (Specifications)
Use these options to enter the nominal target specification, Delta value, and the lower and upper specification limits (specification range). The options that are available is dependent upon the setting of the Type box (see above).
User-defined Mean
By default, the mean is estimated from the data. However, if you select the User-defined Mean check box, you can enter a different mean value in the adjacent box (or adjust the value with the microscrolls).
Sigma limits
These limits will determine the process range that is shown in the summary histogram (Summary histogram button) and that is used for the computation of the process (machine) capability indices (Summary: Current variable button).
In short, this value will determine what is considered the practical range of values produced by the process. The default setting, 6 times Sigma (± 3 times Sigma), is the commonly recommended value.
In order to define the process range for the histogram, the Sigma limits specified here will always be applied to the total process variation. Thus, when analyzing multiple sample data (such as when you want to compute capability indices based on the within-sample inherent variation and performance indices based on the total variation), then the process range indicated in the histogram will still be determined from the process variability (Sigma for the total sample times the Sigma limits parameter.
Shewhart, 1939, warns against the use of the within-sample Sigma because it may greatly underestimate the true process variability for the entire process; this point is also emphasized by Hart and Hart, 1989, page 216).
User-defined Sigma
By default, the process standard deviation (Sigma) is estimated from the data. However, if you select the User-defined Sigma check box, you can enter a different Sigma value in the adjacent box (or adjust the value with the microscrolls).
Within-sample
By default, the within-sample Sigma is estimated from the data. However, if the User-defined Sigma check box is selected, you can enter a different within-sample Sigma value in the Within-sample box or adjust the value with the microscrolls.
Note: Computation of within-sample sigma
The within-sample inherent variation is used to compute the capability indices for data consisting of multiple samples (such as data collected for constructing variable control charts; see the Introductory Overview for details).
If Ranges is selected on the Process Capability Analysis Setup--Raw Data dialog box - Grouping tab, the within-sample Sigma is estimated as R-bar/d2 (see, for example, Montgomery, 1985, 1991; see also Quality Control); if Standard Deviations is selected, then Statistica will compute the within-sample Sigma as S-bar/c4 (see, for example, Montgomery, 1985, 1991); if Variances is selected, Statistica will compute the square root of the average within-sample variance.