Using Measurements & Metrics in Decision-Making

Statistical Quality Control Primer

Statistical Quality Control (SQC)- The application of statistical techniques for measuring and improving the quality of processes. SQC includes SPC, diagnostic tools, sampling plans, and other statistical techniques.
Statistical Process Control (SPC)- The application of statistical techniques for measuring and analyzing the variation in processes.
For any particular quality characteristic, variation can be quantified by sampling the output of a process and estimating the parameters of its statistical distribution. Changes in the distribution can be revealed by plotting these parameters over time.
The variation described by the statistical distribution of a quality characteristic has many separate causes. The Pareto principle illustrates that a few causes will have a major effect upon the total variation. A few more will have a somewhat lesser effect. Most will have a very small effect.
Shewhart at Bell Labs identified two major components to variation: a steady component that appeared to be inherent in the process, and an intermittent component. Shewhart attributed inherent variation, currently called random variation, to chance and undiscoverable causes, and intermittent variation to assignable causes.
Shewhart concludes that assignable causes can be economically discovered and removed, but generally random causes can not be removed without making basic changes to the process.
Each cause can itself vary, and be describable by its own statistical distribution. Each may be causing the observations to move in different directions relative to the others at any point in time. The standard deviation measures the dispersion of each of these distributions.
Individual standard deviations for the different causes add by the square root of the sum of the squares rule, meaning that the largest causes will have an exponential impact on the process observations.
Normal distributions are the result of many approximately equal causes of variation. Most measured distributions are not normal, as the Pareto principle predicts that the magnitudes of the causes will not be the same.
As random causes are eradicated in the improvement process, the causes tend to equalize and the process falls into statistical control.
Statistical control is not the absence of variation factors, but simply the continuous average counterbalancing of variation factors.

Return to... DOQS Home Page, DOQS Education Page