Abstract - Quest for Global Competitiveness - Universidad de Puerto ...

Revealing VariationDr. Shewhart, studying variation as it affected the process of manufacturing telephones,identified two kinds of variation: common random variation that is inherent to the system beingstudied; and special cause variation. By way of illustrating the difference between the two typesof variation, consider a rifle, locked in a firing stand, fired at a distant target repeatedly byremote control. The bullets will not all pass directly through a single small bullet-sized hole;rather, the bullet holes from that rifle will describe a pattern on the target due to the inherentvariation of the system. A host of possible sources of variation exist in the system: tinyvariations in the sizes, weights, shapes, and balance of the bullets; slight variations in thescoring and temperature of the rifle barrel; the normal air currents in the room; the positioning ofthe bullet in the chamber; and so on. The effect of all these sources of variation is a randomdistribution of bullet holes in a group on the target. The less variation there is in the system, thetighter the pattern will be. This variation is known as common variation.Now, imagine the system being disturbed by some special event. A test technician might havebumped into the firing stand and slightly displaced it just before a test is conducted, or the testtechnician might have loaded the rifle with a batch of bullets made from a much heavier metalalloy than all the others. Either of these events would likely result in a group of bullet holes setapart from the others, which would indicate that something out of the ordinary had happened todisturb our system. The "something out of the ordinary" is special cause variation.Dr. Shewhart also identified two costly mistakes that could be made in dealing with variation.The first mistake (called "Type I") is treating common variation as something out of the ordinaryand trying to "fix" the system. Dr. Shewhart discovered that making this mistake actually leadsto increased variation in the system.The second mistake (called "Type II") is ignoring variation due to special causes, believing it tobe common variation. Making this mistake means missing two kinds of opportunities: theopportunity to fix something in the system that actually needs to be fixed; and the opportunity tolearn and benefit from something that went better than could have been expected.Making either mistake leads to loss. Type I mistakes lead to wider variation in the system,making it more difficult to identify and remove special cause variation or to identify changes inthe system's capability. On the other hand, Type II mistakes lead to missed opportunities forfixing or improving the system. Dr. Shewhart developed a statistical tool, known as a controlchart, to help minimize loss from the two mistakes.Let's say that we conduct an experiment by firing 20 bullets from our rifle and measuring thedistance from each bullet hole to the exact center of the target. If we do this 25 times andsample the results of each trial, we can draw a simple run chart as follows:21