100% Inspection Is Not Enough

The discussion on the previous page implicitly assumed that a system of inspection and rework, while very costly, is at least notionally viable. Is this really the case though? Can 100% inspection ensure that no defective items are shipped? The answer to this is no. If defective items are produced, then it is inevitable that some will be shipped. While it is certainly true that for certain safety critical areas, 100% inspection is desirable as a safeguard, it should not be the only means of quality assurance. The real solution is to employ the techniques of SPC to avoid producing defectives in the first place.

Inspection, even 100% inspection, is a subjective and fallable procedure. Two different inspectors may give different results for the same batch, and even the performance of a single inspector may vary according to external factors. Such pressures include the length of time spent on shift, the difficulty of finding comparitively rare defects, and even fear of the consequences of reporting too many defective items.

A simple example that is often used to illustrate the difficulty of maintaining a 100% record in inspection is known as the 'F' test. (NB. this has nothing to do with the f-test used in statistical theory.) Clicking on the link below will open a new window containing a short passage of text. For the purpose of this example, the letter 'f' is to be considered as defective. From the time that the page loads, you will have a little over 1 minute to count every instance of the letter f in the passage. At the end of that time, the page will change to one which asks you to enter the number of f's that you have counted. (If the page does not change automatically, or you finish before time and want to move on, there will be a link to allow you to go to the next page manually.)