Resampling Methods for Estimating Confidence Intervals in Gauge R&R Study

ABSTRACT Gauge Repeatability and Reproducibility (Gauge R&R) analysis is a theoretically well‐founded method to calculate the repeatability and reproducibility of a measurement system. The objective is to determine whether a measurement procedure is adequate for monitoring the performance of a process. In many situations, confidence intervals are necessary, because an interval estimate enables to analyze both how small and how large an effect may be. In this study, the Jackknife method is proposed for estimating the confidence intervals of gauge variability. This approach is compared with the bootstrap method, which, like the Jackknife, enables confidence interval estimation without assuming a specific data distribution, such as normality. Several Bootstrap methods can be applied, including the percentile method, the bias‐corrected and accelerated method, and the Student's t method. These confidence interval methods are compared using three Gauge R&R analyses: Average and Range method, Analysis of Variance method, and honest Gauge R&R study. To achieve this, a rigorous simulation study is conducted to evaluate the performance and to identify the most effective approach. In general, we recommend using the Anova method and the use of the Bootstrap percentile's method. The results of the study and their implications are discussed along with the limitations.