Bhisham C. Gupta - Statistical Quality Control

Quality control helps an organization to create products that, simply put, are of better quality. Continuous quality improvement makes operators, engineers, and supervisors more focused on customer requirements, and consequently, they are less likely to make any “mistakes.”

Quality control may use a technique called acceptance sampling to improve quality. An acceptance sampling plan is a method for inspecting a product. Acceptance sampling may inspect only a small portion of a lot or 100% of the lot. In some cases, inspecting 100% of the lot means all products in that lot will be destroyed. For example, if we are testing the life of a new kind of bulbs for a particular type of projector, then inspecting 100% of the lot means all the bulbs in that lot will be destroyed.

But acceptance sampling plans increase quality only of the end product or service, not of what is still being manufactured or of services that are still being performed, which means any defects or errors that occurred during the production process will still exist. In certain service industries, nothing can be done until the service has been fully provided or after it has been provided. For example, if a patient is receiving treatment, then nothing can be done during or after the treatment if the treatment was bad. Similarly, if a dentist has pulled out the wrong tooth, then nothing can be done after the dentist has completed the service. Thus quality improvement is extremely important in such situations. In manufacturing, acceptance sampling very often requires rework on defective units; after rework, these units may turn out to be acceptable or not, depending on what kind of defects these units had in the first place. All of this implies that acceptance sampling is not a very effective method for quality improvement. We will study acceptance sampling plans in more detail in Chapter 9.

We turn now to process control. In process control or statistical process control , steps are taken to remove any defects or errors before they occur by applying statistical methods or techniques of five kinds: define, measure, analyze, improve, and control. We discuss these techniques in Chapter 2. Deming describes statistical quality as follows: “A state of statistical control is not a natural state for the manufacturing process. It is instead an achievement, arrived at by eliminating one by one, by determined effort, the special causes of excessive variation.” Another way of describing statistical quality is as an act of taking action on the process based on the result obtained from monitoring the process under consideration. Once the process‐monitoring tools (discussed in detail in Chapters 5–8) have detected any cause for excessive variation (excessive variation implies poor quality), the workers responsible for the process take action to eliminate the cause(s) and bring the process back into control. If a process uses statistical control, there is less variation; consequently, quality is better and is continuously improved. If the process is under control, then it is more likely to meet the specifications of the customer or management, which helps to eliminate or significantly reduce any costs related to inspection.

Quality improvement is judged by the customer. Very often, when a customer is not satisfied with quality improvement, they do not bother to file a complaint or demand compensation if the product is not functioning as it is expected to. On the other hand, if there is significant quality improvement, the customer is bound to buy the product repeatedly. These customers we may define as loyal customers . So, quality improvement is best judged by loyal customers, and loyal customers are the biggest source of profit. If there is no significant improvement in quality, then not only do we lose dissatisfied customers but we also lose some of the loyal customers. The loss due to dissatisfied customers or losing loyal customers usually is not measurable – but such a loss is usually enormous, and sometimes it is not recoverable and can cause the collapse of the organization. Thus, quality control and quality improvement are the best sources of good health for any company or organization.

Deming's 14‐point philosophy helped Western management transform old‐fashioned “business as usual” to modern business, where concern for quality is part of the various problems that face any business. Note, however, that there is a danger that these concerns may spread like wildfire, to the detriment of the business as a whole. Further, some problems are what Deming calls “deadly diseases” and become hurdles on the way to fully implement the transformation (Deming 1986, Chapter 3). Deming describes the deadly diseases as follows:

1 Lack of constancy of purpose to plan products and services that have a market sufficient to keep the company in business and provide jobs.

2 Emphasis on data analysis, a data‐based decision approach, and short‐term profits. Short‐term thinking that is driven by a fear of an unfriendly takeover, and pressure from bankers and shareholders to produce dividends.

3 Performance evaluations, merit ratings, or annual reviews without giving sufficient resources to accomplish desired goals.

4 Job hopping by managers for higher ranks and compensation.

5 Using only visible data or data at hand in making decisions, with little or no consideration of what is unknown or unknowable.

6 Excessive medical costs.

7 Excessive liability cost that is jacked up by lawyers who work on contingency fees and unfair rewards given by juries.

Deadly diseases 1, 3, 4, and 5 can usually be taken care by using a total quality approach to quality management, but this topic is beyond the scope of this book. However, deadly diseases 2, 6, and 7, add major costs to the organization without contributing to the health of the business. They are more cultural problems, but they pressure companies to implement quality improvement and consequently compete globally.

Numerical quotas for hourly workers to measure work standards have been a common practice in America. This is done by the Human Resources (HR) department to estimate the workforce that the company needs to manufacture X number of parts. While doing these estimates, it could be that nobody bothers to check how many of the manufactured parts that have been produced are defective, or how many of them meet the specifications set by customers or will be rejected/returned. HR normally does not take into account the cost of such events – which, of course, the company has to bear because of rework on defective or nonconforming parts or rejected and trashed parts. All of this adds to the final cost.

In setting up numerical quotas, the average number of parts produced by each worker is often set as a work standard. When we take an average, some workers produce a smaller number of parts than the set standard, and some produce more than the set standard. No consideration is given, while setting the standard, to who produced a small or large number of parts that meet customer specifications. Thus, in this process, workers who produce more parts than the set standard – regardless of whether the parts meet the specifications – are rewarded, while other workers are punished (no raises, no overtime, etc.). This creates differences between workers and, consequently, chaos and dissatisfaction among workers. The result is bad workmanship and more turnover, and workers are unable to take the pride in their workmanship to which they are entitled.