Control calls for a measurement system to generate information on which to base control actions. In industry, this is the inspection function. Inspectors make measurements that are called for by the quality standards, thereby separating acceptable from non-acceptable units. In addition, the final product must be tested to be sure it performs to standards. However, no control or corrective action is implied. When we link measurement, investigation to determine why an acceptable product has been produced, and corrective action, we have completed the control loop.
The measures of quality in productive systems are diverse. These measures are perhaps most obvious in manufacturing systems where quality characteristics can be related to objective standards of dimensions chemical composition and actual performance tests. Standards for these measures can then be established and regular procedures can be used to measure the critical characteristics to determine if standards are maintained. In service oriented systems, measures of quality are often not as objective. The personal contact required may emphasize the way service is given, even though the service is technically adequate. Waiting time is often a criterion for service quality. The following are typical quality measures of the outputs of productive systems:
TYPE OF SYSTEM MEASURE OF OUTPUT
Tolerances on dimensions
Medical service False positive diagnosis
False negative diagnosis
Postal service Waiting time at post office
Errors in delivery
Banks Waiting time at windows
Liability for poor product quality has been well established. Although negligence on the part of the manufacturer is central to liability, the concept in legal practice extends to include foreseeable use and misuse of the product. Product warranty includes both that expressed by the manufacturer (written and oral) and the implied warranty that the product design will be safe for the user. The uses are not legally restricted to those specified in warranties but include those that may be foreseen. The concept of ‘foreseeable usage’ has often interpreted to mean that if the product was misused, then such use was foreseeable. These legal doctrines place a particularly heavy responsibility on the quality control function, for liability suits can have an important bearing on enterprise survival.
Medical malpractice liability has become an important factor in health care costs Insurance premiums have skyrocketed, and physicians’ fees have reflected the increases. Controversy over the control of health care quality has resulted with emphasis on establishing standards.
All process exhibit variation and the manager’s task is to distinguish between tolerable variation that is representative of the stable system and major changes that result in an unacceptable product. The manager must be aware of the system’s inherent capability in order to know when system behavior is abnormal. Thus, because we are dealing with systems that exhibit variation, the manager’s control model must be a probabilistic one.
Because of the ongoing nature of processes and their inherent variability we must usually base quality control decisions on samples. First, we cannot usually examine all the data because the process is continuous and, at best, we have access to a sample at a particular point in time. (In some instances, 100 percent automatic inspection can be incorporated into the process). Second, even if the entire universe of data were available, it might be uneconomical to analyze it. Third, measurement and inspection sometimes require the destruction of the unit. Fourth, with some products, any additional handling is likely to induce defects and should therefore be avoided. Thus, the sampling of information about the state of incoming raw materials and the control of process is the common approach on which decisions and control actions are based.
The amount of sampling justified represents another managerial choice. As the amount of sampling increases, approaching 100 percent inspection, the probability of passing defective items decreases, and vice versa. The combined incremental cost curve is again dish shaped. The manager’s range of choices is near the minimum cost, but it is a range because un-quantified factors must influence the decision.