In that we have completed a close look at DMAIC, we began with the last issue a review of some the Black Belt Tools. In this issue we shall look at a necessary, but frequently ignored "Road Map".
The Road Map:
- MSA (Measurement System Analysis)
- Stability (Use a passive control chart)
- Normality (symmetry, use the Anderson Darling Test)
- Co-equal Variances (use Bartlett's or Levene's test)
If one is interested in conducting higher level statistical analysis such as:
- Capability
- ANOVA
- DOE
- Parametric Tests
- Non-parametric Tests
It is necessary that the data pass through the gates identified in the "Road Map". If the data fails to pass any of the gates it will compromise the integrity of the pursuant calculations.
MSA identifies the contribution of measurement error to the calculations. When measurement error exceeds 10% of the parent distribution subsequent calculations are significantly compromised.
Stability, this checks for the presence of "Special Causes". Special Causes interrupt the distribution of normal data by infusing anomalous events. This non-normal behavior can really kick the standard deviation, which in turn compromises the integrity of summary statistics.
Normality deals with the symmetry of the distribution. When distributions truncate by one-sided specifications, this results in non-normal data, once again the summary statistics are compromised.
If comparative analysis regarding the averages is the objective, statistically equal variances are required. Hypothesis testing for averages include ANOVA, Confidence Intervals, and T-Test.
Six Sigma Black Belts are trained in statistical methods and will expose the data to each of these gates prior to high order analysis. Knowing how much failure in each of these gate categories to tolerate is a large part of Six Sigma Training.
