| Common in industrial applications is the need for estimates for measurement precision error. Measurement precision error is important because manufacturers make decisions about product acceptance or rejection based on product measurements. A frequent method of determining measurement precision error is the Gauge Repeatability and Reproducibility Study (GR&R Study). A typical GR&R Study determines estimates of repeatability error, reproducibility error as well as estimates of total measurement precision error and the part-topart component. This dissertation compares three methods of performing GR&R studies on 10,080 simulated GR&R study data sets. The 10,080 simulations were derivations of 224 actual Gauge R&R studies. The three methods of analysis are Donald Wheeler's "Honest Gauge R&R Study," the Automotive Industry Action Group's Average and Range Method and the ANOVA Method. The study results were analyzed by ANOVA, Kruskal-Wallis and Pearson correlation. The analysis showed the three methods are different in their estimates of total Gauge R&R and the components of repeatability, reproducibility, and part-to-part. The analysis also estimated the pair-wise comparisons of the three methods and showed they are different from one another for total GR&R, repeatability, reproducibility and part-to-part. The correlation analysis showed Donald Wheeler's method to be correlated with both the Average and range method and the ANOVA method and the Average and range method to be correlated to the ANOVA method. For critical products the ANOVA method is recommended for Gauge R&R analysis, while for less critical products the Average and range method and Wheeler's "Honest Gauge R&R Study" approach are recommended. |
