Don’t ignore the standard for inspection by variables when designing sampling schemes

Don’t ignore the standard for inspection by variables when designing sampling schemes

There are essentially two types of data to be concerned about when devising sampling schemes:
– variables data, and
– attribute data

Variables data are related to characteristics that are measured, such as weight and length. Attribute data relate to characteristics that are assessed on a pass/fail basis, usually by visual inspection; e.g. presence or absence of visually identified defects, such as scratches.

There are two corresponding standards; a standard for inspection by variables, such as ANSI/ASQ Z 1.9, and a standard for inspection by attributes, such as ANSI/ASQ Z 1.4.

A key difference between inspecting by attributes versus inspecting by variables is the size of sample involved, with the sample size for inspection by attributes being much larger.

If you intend to inspect by variables (measurements) then you have the option of using either standard. It is very common, in my experience, to find that people devising sampling schemes in industry ignore the standard for sampling by variables, even though they intend to inspect by variables. Instead they use the corresponding standard for inspection by attributes. The consequence is that they frequently are inspecting unnecessarily large samples.

Suppose you have a lot of 10,000 parts to be inspected by variables. From Table 1, you could choose to inspect by attributes, draw a sample of size 500 and use a pass/fail approach to test the quality of the parts. Alternatively, you could use the standard for inspection by variables. If you have adequate historical data, typically obtained from previous inspections, which enables you to calculate the process standard deviation, you could use the Variability Known method, and the sample size could be as low as 44. Even if you don’t have the historical data necessary to establish the process standard, you can use the Variability Unknown method, in which case, the sample size will be reduced to 150, a considerable reduction from the 500 that applies in the inspection by attributes case.

Using the standard for inspection by variables involves calculating some simple statistics, easily performed with a hand-held calculator. It appears to me that that the need to get involved with statistics is sufficient to deter many people from using the standard for inspection by variables. Instead, they will suffer the unnecessarily large sample sizes that will result from use of the standard for inspection by attributes, and test the measured characteristics using a pass/fail approach.

Inspection personnel can be readily trained to use the appropriate standard for inspection by variables, and thereby greatly reduce the time involved in inspection, when the characteristics are measured.