Combining Scott-Knott and GR&R methods to identify special causes of variation

Abstract

Measurement error is an unavoidable source of variation in any decision-making process based on experimental research. Components of variation due to measurement system and manufacturing process must be estimated and special causes of variation should be reduced whenever possible. GR&R (gage repeatability and reproducibility) studies quantify these sources of variation by using analysis of variance. The main contribution of this paper is to conjoin GR&R and the multiple comparisons method of Scott-Knott in order to help practitioners identifying special causes of variation in empirical studies. Stainless steel cladding process has been evaluated to validate the proposed procedure. The experimental findings have shown that the well-structured method based on Scott-Knott test was effective in indicating the source of error due to reproducibility.