Sensitivity Analysis for Identifying Parameter Changes in Attribute Data Utilizing Modified EWMA with Improved Square Root Transformation

The original Shewhart p- and c-charts monitor binomial and Poisson attribute data using normal-approximation control limits, which may be unreliable for small or moderate samples, low defect levels, or sparse nonconformities. This study proposes Improved Square Root Transformation-based modified exponentially weighted moving average charts, namely ISRT p-MEWMA and ISRT c-MEWMA, to improve detection of small and moderate shifts in attribute processes. The charts integrate the variance-stabilizing capability of ISRT with the memory mechanism of MEWMA. Monte Carlo simulations were performed under various baseline parameters, subgroup sizes, and shift magnitudes, with all charts calibrated to a common in-control average run length. Results indicate that ISRT p-MEWMA performs well for low-defect binomial data, whereas ISRT c-MEWMA is effective for low-count Poisson data. Real-data applications confirm the practical usefulness of the proposed charts.