MSA for bounded ordinal data: a new approach

In measurement system analysis (MSA) the assessment of the measurement system’s precision is a very relevant topic as it allows to evaluate the capability of the easurement system. The standard method generally proposed to estimate the precision in this context is the gauge repeatability and reproducibility method (GR&R). For continuous output this method has received substantial attention both in the statistical literature and in the international standards. On the contrary, few studies are available to investigate the capability of the easurement system when output data are categorical, despite their wide application e.g. in industry, applied metrology and quality assurance. In our presentation, we are going to present a new procedure based on a mixture model to evaluate measurement error for ordinal data.