Home Page of Inspection and Quality Control

sigma

engineering partnership

Inspection and measurement

"... an activity will in general have two valuable consequences: the physical outputs themselves and the change in information about other activities."

Kenneth J Arrow
Essays in the Theory of Risk-Bearing

Data that represents some physical characteristic and enables us to assess its deviation from target is called variables data. E.g. length, time, temperature.

Data that simply captures whether an item was conforming or non-conforming is called attributes data. Variables data can be turned into attributes data by comparing it with the specification limits, recording whether or not the item conforms, then discarding the original variables data.

Sometimes, the perceptions of a visual scrutiny can be turned into pseudo-variables data by using a Likert scale to characterise various degrees of damage on an arbitrary series. An example of this is the Beaufort scale for wind speeds.

Inspection uses measurements simply to categorise items as conforming or non-conforming. However, variables data also allows the use of statistical process-control (SPC) to:

Stabilise unstable processes
Centre processes on target
Reduce variation to tolerable limits
Maintain stability

This strategy is used by Shainin, "six sigma" and other practitioners. An effective SPC regime will:

Eliminate the risk of defective parts and the need for inspection
Minimise economic losses
Capture learning that can be deployed in future activities.

A stable process that is centred on target and has small enough variation will offer little chance of non-conformance. Such processes are known as capable.

Home page of Inspection and Quality Control

sigma HOME

engineering partnership

This page last updated 13^th July 2001