Quality Management: Statistical Sampling In Quality Control

Fallacy of Inspection Total Quality Control The recent North American approach to quality emphasized that bad quality was the fault of workers. Because quality was a "worker problem," the only way to achieve acceptable outgoing quality was to "inspect it in" by adding more inspectors and more inspection stations. Thus, statistical sampling of items was performed to ensure (with, say, a 95% confidence level) that the number of outgoing defective items did not exceed a specified percentage. Occasionally, if demands to meet production targets and sales orders were high, marginally defective items were passed through final product inspection. After all, that's why the company had a field service force and a warranty reserve; failures, if they occurred, could be ixed in the field. In the meantime, production targets could be met and fixed costs absorbed into production (of both good and defective items). This casual approach to quality had predictable consequences. Factories had high levels of inventory at each production stage to protect against shortages caused by uncertain quality from previous stages; incurred large costs for inspection, rejects, rework, scrap, and warranty repairs; and produced general confusion on the factory floor as production schedules were interrupted by parts and material shortages due to defective items, and by numerous production schedule changes required in order to reprocess defective items. Large areas of the factory were used to house defective items that were waiting to be reprocessed (and were even given a name, the "rework area"). Such activities were not a problem for traditional standard-cost sys-tems. A "normal" allowance for scrap, waste, and rework was estimated and built into the standard costs for the process. Thus, as long as the production process did not deteriorate drastically from year to year, the costs of defective production could be assigned to the good items actually produced and unexpected variances would therefore be avoided. Also, some of the costs incurred to produce defective items could be recovered by higher prices in the marketplace. The philosophy of emphasizing production volume over quality worked fine for U.S. manufacturers from the mid-1940s through the 1970s. Few of their domestic competitors were emphasizing manufacturing qual-ity, and companies in the rest of the industrial world were rebuilding their factories from the destruction wreaked during World War II, so that they could not compete effectively in the market for manufactured goods. During this time, however, several innovative Japanese companies and some West European manufacturers, especially in Germany, Switz-erland, and Sweden, adopted a total quality control philosophy in which the only acceptable quality level was zero defects. The zero defect goal was not expected to be achievable, but it provided a target so that a company would never be satisfied unless it was reducing the incidence of defects by substantial amounts year after year. For these companies, defects were measured not as a percentage of outgoing items, but as a parts-per-million (PPM) ratio of defects to items produced. Many activities had to be done correctly in order to achieve continual reductions in PPM defect rates. Most important was to recognize that quality had to be designed in, not inspected in. For example, products designed with fewer parts generally place lesssevere demands on the quality of each component in order for the entire product to achieve a given reliability level. Also, when product designers thoroughly understand the manufacturing process, they will be less likely to design products that make stringent requirements on fabrication and assembly processes. The IBM Pro-Printer was designed so that it could be assembled by robots, without manual intervention. The parts were designed so that they fit together with only one orientation. As a consequence of the care taken during the design stage, the product can also be reliably assembled even by an unskilled worker without any mechanical tools, thereby greatly reducing the errors that might otherwise be introduced in the assembly process. Many manufacturing problems arise because of the difficulty of reliably producing a product that has been designed for performance, but not for manufacturability.