A Robustness Study of X̄ Charts with Variable Sampling Intervals

Publisher Website
Google Scholar
Abstract
Recent theoretical studies have shown that, for all but very large process shifts, control charts using variable sampling interval (VSI) schemes are more efficient in their detection of shifting processes than the more conventional fixed sampling interval (FSI) schemes. This article, through simulation, considers the properties of the VSI X̄ chart in an environment where the process data are not normally distributed but are contaminated. In addition, it evaluates the behavior of VSI charts where a trimmed mean, a winsorized mean, or the median is used as the chart statistic. Comparisons between the FSI and VSI X̄ charts indicate that the VSI chart continues to be more efficient. Further, trimming the mean and/or including a rule for reducing the amount of switching between the two different sampling intervals improves the efficiency.

This publication has 7 references indexed in Scilit: