In a standard mean chart, such as Figure 1, the extent to which the sample means are likely to vary under "normal" circumstances is calculated by measuring the standard deviation of the individual items within each sample, and then using this to predict the variability of the differences between samples, using a formula such as:

This is particularly important for range charts: this means that the range chart which accompanied the mean chart in Figure 1 was flawed because the variable plotted here is very unlikely to be even approximately normally distributed (it had a pronounced skew).

26, whereas the estimate calculated from the variation within the samples using the standard mean chart procedure is 1.

These values, and the corresponding ones for method A and the conventional mean chart method, are summarized in Table I.

A large number of different control charts are described in the literature - for example, mean charts, range charts, P charts, C charts, U charts, charts for individuals, moving averages charts, and so on.

Problem 3: The number of different models for calculating action and warning levels - mean charts, range charts, P charts, C charts, etc.

Furthermore, they can be used in place of any of the standard control charts, not just mean charts.

Furthermore, the same procedure works for mean charts, P charts, or range charts, or any other chart which is based on a random sample of measurements or observations.