Dear Colleagues,

Unbelievably, inspecting an item considerably more times as against fewer times than required will both result in more errors being identified.

What are the Statistics here?
When inspecting two types of errors to avoid are a type 1 error where an actual part that is not defective is assessed defective (a Type 1 False Positive Error). Or a Part which is actually defective and which is assessed as not defective (a Type 2 False Negative Error).

In other words, in inspecting items of engineering equipment one can reject a good item (False Positive) or miss a defective item (False Negative).

A False Positive is not too Critical
This is perhaps not too bad apart from the cost and effort of replacing a defective item.

But a False Negative Could be Catastrophic
However, a false negative has to be avoided like the plague. Your critical piece of equipment has been inspected and signed off as acceptable when it is actually defective. This would result in failure during service with often horrible consequences.

Going Berserk with More Inspections to find errors is not the answer
Theoretically, doing a huge number of inspections could result in nearly every item being inspected being found defective for some reason or other. So, simply increasing the number of inspections is not necessarily the best answer. One has to look elsewhere for solutions here – such as improving (or even in changing) the manufacturing process for the equipment.

The optimum number of inspections is a balancing act between a high level of inspections (Costing a lot) against a minimal number where you may fail to uncover real errors (and thus resulting in a major failure of an item of equipment during service).

Thanks to 101 Things I learned in Engineering School by John Kuprenas with Matthew Frederick.

Although not necessarily good for equipment; this probably applies well to good engineering professionals – thanks to Jenni Rivera for a good quote: Thank you for accepting me as I am, with my virtues and defects.

Yours in engineering learning,