SHEWHART CONTROL CHARTS - AN IRREPLACEABLE TOOL OF EXPLANATORY DATA ANALYSIS WITH UNDERESTIMATED POTENTIAL

Abstract

In this paper some issues relative to the gap between the traditional theory of control charts and real problems practitioners encounter are discussed. We consider both the general reasons for this discrepancy and different examples of misunderstandings. The trend to develop statistics as mathematical branch of science in the area of statistical process control has led to (i) ignoring many real complexities; (ii) creating many new types of charts that rarely help practitioners to improve their processes. We offer some practical advices, such as the introduction of two types of the assignable causes of variations (internal and external); the refusal from a traditional assumption that repetitive measurements are always normally distributed; the simple and practically convenient technique to calculate control chart limits for highly non-normal data. As the main direction for future efforts, we offer to start discussion about the implementation of Shewhart control charts into the program of school education.

Keywords

• shewhart control chart
• assignable causes of variation
• nonhomogeneity
• nonrandomness
• process stability
• capability indices