1. There are two principal types of ways of measuring quality: attributes (counting the number of occurrences in one or two categories) and variables (measuring a characteristic or the deviation from a standard).

2. There are two principal types of Quality Assurance: sampling inspection of incoming outgoing materials (acceptance sampling) and control charts for ongoing processes (process control).

3. Control charts differentiate between the process being in control (within an accept range of random variation) and out of control (outside the acceptable range). Process control involves making frequent small sample inspections to detect whether the process in control or out of control. Out of control means either above the upper control limit below the lower control limit. The limits for control charts are established first by obtaining 20 to 25 small samples of same size at different times from the process. This forms the basis for judging future samples.

4. The four types of control charts described in this chapter are: x-bar chart for means, R charts for ranges, p-charts for fractions of defective items, and c charts for the number of defects.

5. Key managerial decisions about the uses of control charts are where in the process to use them, what size sample to take, and what type (variables or attributes) to use.

6. Run tests such as the number of runs with respect to the median, and the number of runs up and down, can be useful in checking for nonrandomness. Run tests should be used in conjunction with control charts. Both approaches test for randomness. A process that is in control will exhibit randomness in a sequence of observations; the appearance of nonrandomness suggests that the process is not random (i.e., not in control).

7. A process capability ratio provides an indication of the ability of the process to meet specifications. It is computed as the ratio of a specified dimension to the process width on that dimension. A value of 1.00 means the two are equal; the greater the value of the ratio, the more likely it is that the process output will fall within the allowable variation (specification).