The Models
DIN 55319 (the German industry norm, widely adopted by German auto manufacturers) and ISO 21747 (the international norm) are almost identical, with a few differences in notation and labeling. In Statistica , you can choose either the DIN 55319 or ISO 21747 notation and recommended approaches; in practice, the results will be near-identical in all but extreme cases.
The following table summarizes the distribution models and recommended approaches according to DIN 55319 and ISO 21747:
DIN 55319
Distribution Model:
| Characteristic | A1 | A2 | B | C1 | C2 | C3 | C4 | D |
| Location | c | c | c | r | r | s | sr | sr |
| Variability | c | c | r | c | c | c | c | sr |
| Skewness | c | c | c | c | c | c | c | sr |
| Kurtosis | c | c | c | c | c | c | c | sr |
| Instantaneous Distribution | nd | nnd | nd | nd | nd | nd | nd | as |
| Outcoming Distribution | nd | nnd | 1m | nd | 1m | as | as | as |
ISO 21747
Distribution Model:
| Characteristic | A1 | A2 | B | C1 | C2 | C3 | C4 | D |
| Location | c | c | c | r | r | s | sr | sr |
| Dispersion | c | c | sr | c | c | c | c | sr |
| Instantaneous distribution | nd | 1m | nd | nd | nd | as | as | as |
| Outcoming distribution | nd | 1m | 1m | nd | 1m | as | as | as |
Characteristics
The Characteristics column in this table denotes the different characteristics of the respective distribution, such as, location (for example, mean), variability (for example, standard deviation), skewness, and kurtosis.
- Instantaneous Distribution is the distribution of observations within each sample.
- Outcoming (Resulting) Distribution is the marginal distribution of all observations, across samples
Indices for characterizing the distribution type
- nd - normally distributed
- nnd - not normally distributed
- 1m - mixed distributed with one mode only
- as - any shape (also sometimes abbreviated asw for any shape whatsoever)
These labels can be applied to both the individual samples taken at a particular point in time (Instantaneous Distribution), as well as the resulting distribution of multiple samples taken over time (Outcoming Distribution).
Time dependency of distribution parameters
- c -the distribution parameter is constant (constant)
- r - the distribution parameter changes randomly only (randomly)
- s - the distribution parameter changes systematically only (systematically)
- sr - the distribution parameter changes systematically as well as randomly
The label s indicates that, even when used in sr, the time-dependent changes are deterministic and predictable, for example, as a trend.
For additional details, refer to DIN (Deutsche Industrie Norm) 55319 (see Deutsches Institut fuer Normung e.V., 2002)
Summary: Models
The time-distribution models considered in DIN 55319 and ISO 21747 are essentially identical. However, significant differences exist in some of the computational approaches for estimating process capability, given the different distribution models; some of these differences are primarily attributing to different (and somewhat confusing) labeling.