ABSTRACT
We analyze the performance of traditional R charts and introduce modifications for monitoring both increases and decreases in the process dispersion. We show that the use of equal tail probability limits and the use of some runs rules does not represent a significant improvement for monitoring increases and decreases in the variability of the process. We propose to use the R chart with a central line equal to the median of the distribution of the range. We also suggest supplementing this chart with a runs rule that signals when nine consecutive points lie on the same side of the median line. We find that such modifications lead to R charts with improved performance for monitoring the process dispersion.