Quality Control Pressures to Improve

Quality Control

Pressures to improve management in government have long been couched in terms of efficiency and economy. As a result, considerable progress has been made in the direction of increasing production and reducing costs. Comparatively little use has been made, however, of effective methods for controlling the equally important element of the quality of work (Walton, 2000).

Reduced costs and increased production are illusory gains if they are achieved at the expense of serious deterioration in quality. In any activity it is imperative to determine standards of quality as well as standards of quantity and cost. Although the relative importance of these three factors may vary in different situations, each of them must be considered in every case (Walters, 2007). The purpose of this article is to stress the importance of defining the degree of quality wanted in government operations and to point out that once these quality goals are set, management can use the relatively new technique of statistical quality control to see that these goals are met (Perez & Ziaja, 2008).

Quality cannot be controlled until a decision is reached upon the desired quality standards or goals. In most cases it is possible to design a procedure to attain almost any degree of accuracy. But the higher the degree of accuracy desired, the greater will be the cost. It is possible to approach perfection, but only at the expense of excessive checks to correct the errors, which inevitably appear in any process. The taxpaper will get the most for his dollar if quality goals are realistic enough so that expenditures to prevent errors are not greater than the costs resulting from the errors (Matthews & Peter, 2012).

Once quality goals have been set and operations stabilized, statistical quality control enters the picture (Walton, 2012). Statistical quality control is simply a method for determining the extent to which quality goals are being met without examining every item produced, and for telling management whether or not the errors or variations which occur are exceeding normal expectations. It was introduced into large-scale manufacture in the United States in the 1930's (Walters, 2007). During the war the technique spread rapidly in British and American war factories and resulted in tremendous savings. The Western Electric Company, for example, cut its rejects on some items up to 50% and saved millions of dollars in overhead. In another case, armor-plate rejection percentages were reduced from 33 to 3 (Matthews & Peter, 2012).

In industry, especially when dealing with manufactured items, quality goals are usually stated in terms of such characteristics as dimensions, weight, or durability; in terms of fraction defective (e.g., the ratio of broken panes of glass to the total number inspected); or in terms of defects per unit (e.g., the number of imperfections in a bolt of cloth, or the number of missing parts in an assembled item) (Perez & Ziaja, 2008). These types of goals are not appropriate, however, to the clerical operations so frequently found in government operations. In most government agencies, whether at the local, state, or federal level, quality goals can be set more effectively in terms of number of errors made (Walters, 2007).

In general, the quality of a given product can be determined in three ways: (1) by analyzing the complaints of those people who use or are affected by the product, (2) by surveying the opinions and attitudes of people familiar with the product, or (3) by some form of inspection, review, or test of the product itself (Walton, 2000). Although industry has made significant strides in the analysis of customer complaints and in surveying customer opinion, government has done little exploring in these areas (Matthews & Peter, 2012). Government seems to have concentrated on the third method, with the result that many a citizen's complaint of government delays and red tape can be traced to excessive inspections, checks, and reviews. Despite this emphasis, government offices have made relatively little use of the most modern version of inspection -- statistical quality control. This recently developed technique appears to offer the greatest possibilities for effectively and economically insuring that quality goals are met (Perez & Ziaja, 2008.

Statistical quality control employs two statistical techniques: the control chart and statistical sampling. Both of these techniques are based on the laws of probability (Box & Colleagues, 2009).

The Control Chart

The control chart has been developed in various forms. Essentially, however, it is a device for plotting data (such as dimensions, errors, weights, or similar pertinent figures) so as immediately to reveal the frequency and extent of variation from standards or goals. Control limits based upon the established tolerance limits for the data being dealt with are placed upon the chart. Variations that fall within the control limits may be considered as due to chance or unknown causes. These causes bring about what may be called the natural variability of a process (Perez & Ziaja, 2008. Variations that fall outside the control limits are danger signals and indicate that there is a definite, assignable cause at work helping to bring about the variations. The control chart tells the manager at a glance whether his process is in control (i.e., within the control limits); thus he need not dissipate his energies tracking down random variations, but can begin to act the moment an assignable cause appears (Matthews & Peter, 2012).

The control chart has been likened to a highway whose control limits are the shoulders on one side and the center line on the other. No car driving along the highway can maintain a perfectly straight path. Unevenness in the road, play in the steering wheel, gusts of wind, and a host of other factors cause slight variations in the path of the car. It would hardly be worthwhile to investigate the causes of these small irregularities. However, the moment the car swerves outside one of the limits, an assignable cause can be assumed to exist and an investigation should be begun. The cause may turn out to be a defect in the steering mechanism, a sleepy driver, a "one- armed" driver, or some similar specific correctable factor (Sherr & Teeter, 2010).

The primary value of control charts is that they tell the manager when assignable causes for variations are at work. They contribute an additional advantage, however, in that they publicize production results; thus they furnish a convenient way of stimulating competition, either among groups doing similar work, or within the same group by permitting comparison of present and past records (Mercer, 2003).

Statistical Sampling

The second technique involved in statistical quality control is statistical sampling. Statistical sampling attempts to insure a true picture of the whole by use of a random sample (i.e., one in which every item has an equal chance of being inspected) which is at the same time thorough (i.e., all variations in the sample are discovered) and regular (i.e., recurring consistently rather than at long and irregular intervals). Statisticians have worked out tables so that once the quality goal is determined (e.g., 1% errors allowed) and the percentage of errors made by the inspectors is known, the size of sample to be used to insure the quality goal can be determined. In certain cases the system of sampling permits the use of a larger sample if the variations in the sample taken exceed a specified amount. This method is frequently used when testing the relative acceptability of purchased items (Matthews & Peter, 2012).

Sample testing or inspection has two primary advantages. In the first place, it saves time and money. The size of the sample can be calculated so as to assure the desired degree of quality (Walton, 2000). To the extent that the sample size represents less than 100% review, there is a saving in inspection time and cost. In the second place, sample testing often results in improving the quality of work. A worker who knows that only a portion of his work is to be reviewed feels an increased sense of responsibility and exercises greater care. Experience has shown that the work of the inspector also will be more reliable when he concentrates his attention upon only a selected portion of the items (Perez & Ziaja, 2008).

Quality Control in Office Operations

The statistical quality control system described above is usually said to have originated with Walter a. Shewhart of the American Telephone Company in the early 1920's (Walters, 2007). During the last war, the great need for speedy production, the enormous increase in actual production, and the shortage of qualified inspectors made it imperative that effective methods of quality control be adopted. As a result, the use of statistical quality control spread rapidly in both the United States and Great Britain in ordnance factories (including Army and Navy ordnance plants) and in industries producing such items as electrical equipment, steel, automobiles, and photographic equipment (Walton, 2000).

The application of statistical quality control to clerical operations (i.e., mass paper work activities) as opposed to manufacturing is of more recent vintage. Observers and practitioners of public administration are especially interested in this particular use of quality control, for much government work consists in the routing and processing of a huge volume of paper work. One of the best examples of the use of statistical quality control in clerical operations is found in Aldens' Mail Order House in Chicago. Statistical quality control was begun at Aldens' early in 1945 by the installation of sample inspection and the control chart in one of the order-picking departments (Mercer, 2003). Within two months, the error ratio in this department fell from 3% to less than 1% while efficiency increased from 82% to 107% (Walton, 2012). Since then, use of the system has been extended throughout the organization (23 departments by June, 1947) with such outstanding success that it has gained the complete support of top management. Over a two-year period, statistical quality control brought about a reduction in errors of 25.4% as indicated by customer adjustments (Box & Colleagues, 2009).

With few exceptions, statistical quality control in the federal government has been confined to engineering or construction activities (e.g., Army and Navy ordnance). The Bureau of the Census and the National Office of Vital Statistics of the Federal Security Agency have experimented with dif Quality circles (QC) are organizational interventions that seek to increase an organization's productivity and the quality of its products through direct employee participation (Walton, 2012). The underlying assumption is that such participation will result in useful suggestions for improving work methods and quality control, and for increasing employee commitment to implement these changes (Sensenbrenner, 2005). A quality circle is composed of a small group of employees, doing similar work, who volunteer to meet periodically to discuss production, quality, and related problems, to investigate causes, recommend solutions, and take corrective actions to the extent of their authority (Sherr & Teeter, 2010).

Normally, a company-wide steering committee of both union and management representatives decides where in the organization quality circles should be introduced and what types of problems are appropriate for the quality circles to work on. Once initiated, a quality circle (consisting of about ten employees from a work unit and their immediate supervisor) holds a weekly one-hour meeting to discuss ways of improving productivity and related issues (Walton, 2000). To aid their effectiveness, the group and its leader are trained in group dynamics, problem solving, data analysis, quality control, and the presentation of information and recommendations to management. Circle leaders usually receive about three days of training prior to the circle's first meeting (Sensenbrenner, 2005). Circle members receive their training during the first eight to ten circle meetings. These meetings are held on company time and at company expense, and the decision to implement any of the group's suggestions remains ultimately with management. External facilitators, who have received about five days of training in the use of quality circle techniques and are usually company employees, guide and assist the quality circle during the meetings (Walters, 2007).

Within the federal sector, the Navy was the first to implement a quality circle program in 1979 in its Norfolk Naval Shipyard. By 1980 the shipyard claimed to have achieved a four-to-one cost-benefit ratio. The Navy has since expanded its QC program to a number of its bases and shipyards. In addition, a variety of other federal agencies (including the Air Force, the Veteran's Administration, and the Public Health Service) have all begun to experiment with their own quality circle programs. Interest in the QC process among federal agencies appears to be rapidly growing (Mercer, 2003).

The Bureau of the Census, for example, in processing 1940 census figures for housing and population used statistical sampling in the verification of card punching and maintained quality control charts on each individual puncher (Walton, 2000). Great care was taken when setting up the sampling system to develop criteria for selecting the punchers whose work should be sample verified. Length of experience, average error rate, and fluctuations in error rate were determined to be the controlling factors (Sensenbrenner, 2005). Over 90% of the qualified punchers stayed within the upper control limit plotted on their respective charts. Investigation of the reasons for errors in the case of those who exceeded the permitted limit revealed such assignable causes as (1) schedules poorly filled out by the enumerator, (2) a puncher who had returned to work too soon after a siege of measles, and (3) sickness in the family of a puncher. With this knowledge as to the causes of errors, management was able to take intelligent action to remedy situations. This statistical quality control system was estimated to have saved $263,000 in direct labor costs; indirect savings were estimated to have paid for the cost of the system. In addition, speedier service in the preparation of the final statistical tables was obtained (Walters, 2007).