Terminology

In order to accommodate the growing variety of modern applications for analysis of means, the term group is used instead of treatment level throughout the documentation for the ANOM procedure. Likewise, the term group-variable is used to refer to the variable in the input data set that classifies the observations into treatment levels. In the ANOM procedure, a group-variable plays the same role as a CLASS variable in the GLM and ANOVA procedures, and it is syntactically the same as a subgroup-variable in the SHEWHART procedure.

The nomenclature for ANOM charts is the same as that for Shewhart charts: $\bar{X}$ charts for means, p charts for proportions, and u charts for rates. Consequently, the syntax for the ANOM procedure is patterned after the syntax for the SHEWHART procedure. However, there are some important differences between ANOM charts and Shewhart charts:

  • Analysis of means is formally a test of hypothesis, whereas a Shewhart chart is used to distinguish between special and common causes of variation.

  • In an ANOM chart, the horizontal axis corresponds to the group-variable, and it identifies the groups, which can be displayed in any order. In a Shewhart chart, the horizontal axis corresponds to the subgroup-variable, and it identifies the order in which the subgroup measurements were taken.

  • An ANOM chart displays response summary statistics for a set of groups (treatments) at a specific time. A Shewhart chart displays subgroup summary statistics for a specific process where the subgroups are made up of measurements taken over successive points in time.

  • In an ANOM chart, the decision limits are determined by a specified significance level ($\alpha $), which is the probability that under the null hypothesis of no treatment differences, at least one of the response summary statistics will exceed the decision limits. In a Shewhart chart, control limits are typically computed as $3\sigma $ limits.