Inspection efficiency is made up of many factors, those contributing to the success of detection, the others contributing to noise. Juran’s estimation of 80% efficiency is based on the noise factor interfering with 100% detection in his personal experiences over the many years in industry. Think of Man, Machine/Equipment, Method, and Material. Now think of the possible noise contributing to the inefficiency of any one of these inputs. If Juran is right, then all the noise will reduce my ability to detect a defect by 20% regardless of the fraction defective in the lot. The lower the fraction defective, the lower my probability of detecting it. This is also an exponential function, not a linear regression.
James presents a good topic for discussion: the efficiency of private screening services at airports. Think about Man, Machine/Equipment, Method, and Material. Will the federal government do a better job in any one of these aggregates? Marc’s comments in correcting me are right on the money. What can be done to the process to poka yoke it (this will reduce the noise factor and increase detection)? What low cost/no cost modifications be done immediately to improve security? What more costly solutions can be instituted provided there is data to support the expenditure?
Suppose there is a 95% efficiency in the category Man, 95% in Machine/equipment, 95% efficiency in Method, and 95% efficiency in Material. What is the System’s overall efficiency (they are all interrelated)? Here is something posted by Don Winton (our Stats Wizard) on September 28, 1998 that explains System's Thinking:
“…Possibly. I prefer to think of this particular case as a lack of systems thinking. Each of the solid boosters had three joints. The specification read that each joint had to have a less than 2% probability of failure during any single firing. The testing verified this to be the case (satisfied the specified requirements). Now, let us look at the system requirement.
Each system contained two solid rocket boosters, each with three seals that had met the 98% probability of success. What is the probability of a system failure?
This is defined as (0.98)^6 = 0.886 or 88.6% So, for each system launch, there is a 88.6% probability of success!!!!!! By the time STS-56 was launched, it only had a 50% chance of success. That is a lack of system thinking personified!…”
Did this just muddy the water?
Kev