The Decision Sciences Journal of Innovative Education

Serving Up the Red Beads Experience

Christine M. Wright

Assistant Professor of Management

Western Carolina University

Cullowhee, NC 28723

Michael E. Smith

Assistant Professor of Management

Western Carolina University

Cullowhee, NC 28723

Introduction:

The “Red Beads” exercise popularized by W. Edwards Deming (1986, 1992) is useful when teaching about quality management, statistical process control and the management of people within a manufacturing context. However, students often fail to generalize the lessons to the service setting, which is frequently the venue of their eventual employment.  In order to aid in the development of more general understanding, the following exercise was devised using an airline example, and incorporating many of the principles from Deming’s 14 Points.

Student Instructions:

You are a customer service representative for Cullowhee Airlines.  Company policy requires regular monitoring of customer satisfaction.  One day is randomly selected out of each week, and five randomly selected customers that you serve on that particular day will be surveyed regarding the extent of their satisfaction or dissatisfaction with their service encounter. The results from this regular monitoring will be used in your yearly performance evaluation. During the class period, you will simulate the monitoring of your performance as well as the performance of four of your co-workers for a 5-week period.

In order to conduct this simulation, you will randomly draw five customer cards from the top of the deck. This essentially simulates the randomness of a first come, first served one-queue waiting line system where the next available server is automatically assigned to the next waiting customer. After each card is drawn, the quality control manager will read each card (customer survey) and announce whether that customer was satisfied or dissatisfied with the day’s service. The quality control manager’s secretary will record the number of dissatisfied customers for each day in Table 1 as well as sum and average those results as appropriate; you should do the same.

Instructor Instructions:

Give each student the student instructions and results chart. Choose five students to be customer service employees, a quality manager and a quality secretary.  Make copies of the customer satisfaction/dissatisfaction cards shown in Exhibit 1 in such number as ensures that 20% of the total number of cards indicate customer dissatisfaction and 80% indicate customer satisfaction. In conducting the session, follow the student instructions.  For more details relating to the original “Red Beads” exercise, a review can be found in Latzko and Saunders (1995).

Discussion:

The authors use the lessons and concepts from the “Red Beads” made famous by Deming in this adaptation of the exercise, which was developed for a quantitative business analysis course that focuses on management of service-delivery systems.  Instructional opportunities supported by this activity include creation of statistical process control p-charts, and discussions of topics including appropriate methods for collecting data, the importance of monitoring and measuring activities, and the issues involved in making decisions about employees based upon these data, as well as many points in Deming’s approach to transformative management.

It is common to hear dismissive remarks made about the simplicity of many training simulations of the type described in this article.  In fact, simplicity often highlights critical concepts since complexity does not obscure them.  Obviously, the system accounts for the variation in evaluations in our simulation, this may not be as evident in restaurants where tips may vary, but many of the variables affecting customers’ perceptions are not under the server’s control.  The servers may not have control of where patrons are seated, rarely have any control over the preparation of the meal or the number of servers working, and certainly do not have control over the construction and decoration of the physical facilities.  Customers rarely separate these elements from those over which the server has control when reporting their satisfaction with their server.  For all of the situations outside the control of the server, evaluation either by the patrons or by restaurant management essentially represents the luck of the draw.

Beyond the question of realism, the experience described here also effectively serves as a platform for discussions of the presence of variation in every work process.  Only reasoning guided by statistical knowledge recognizes variation when we use numeric information to manage work processes.  Constructing a p-chart, as shown in the Appendix (directions for which can be found in most operations management textbooks, e.g., Fitzsimmons and Fitzsimmons, 2000), can help to demonstrate the value of such knowledge. Furthermore, with regard to quantitative analysis, ties could also be made to probability distributions, the central limit theorem, sample size and variability.

The control chart allows us to predict that continued operation of our airline will result in employee performance that falls between the control limits, which can be illustrated by subsequently conducting the simulated business for additional weeks.  Often, students will mistake such predictive ability for indicating adequacy of performance, instead of correctly regarding performance within the limits as indicating that the causes for current performance rest within the system.  The inadequacy of common supervisory behavior under such circumstances becomes apparent when we discharge those employees who have the largest number of dissatisfied employees after the third week of the game, and “motivate” employees with exhortations and posters bearing messages such as “work smarter not harder; do it right the first time; zero defects; quality is up to you; quality counts” (Latzko and Saunders, p. 90).

Posters (easily produced by printing slogans and enlarging them on a copier) can be placed in conspicuous locations prior to the class session, and the “workers” can be directed to pay attention to them.  We also make it clear to the employees that continuation of their jobs is contingent upon their adequate performance.

At the conclusion of the first week, the instructor may wish to tell the employees that their performance is disappointing, and institute a “zero defects” campaign (an opportunity to provide additional encouragement through more posters and slogans).  After the second week, those workers with satisfied customers can be told that they have received a bonus, and those with dissatisfied customers can be told that they have been placed on probation.  This provides the context for firing those with continued poor performance after the third week (we often suggest that we cannot risk the jobs of other employees by tolerating poor performance, and that there are plenty of people eager to work).  The instructor may also want to point out the use of “progressive discipline.”  You can use exhortations to supplement or replace the posters, and the order of these misguided attempts to improve the performance of the employees is not critical, however, the more creative your efforts, the more readily you can illustrate the damage wrought by poor management.

The exhortations and firings are obviously nonsense (in our experience, this is a point that students grasp more readily in this adaptation than with the “Red Beads”), and they engender fear which inhibits worker contributions to improve the system, a line of reasoning implied in Point 8 (Drive out fear) of Deming’s (1986) 14 Points.  Deming’s Point 5 states that systemic improvement is a constant requirement, and management needs to be proactive in obtaining worker participation in such improvement efforts by instituting training (Point 6).

Deming’s discussion of points 3, 7 and 10 is also applicable to this exercise. 

Point 3: cease dependence on inspection, and instead build quality into the system.  The inspection efforts in this simulation do not provide quality.  While one can argue that, at least in principle, reworking or rejecting manufactured goods can be marginally effective, such an opportunity rarely exists in the service setting.

Point 7: institute leadership.  Mangers need to take responsibility for the systems for doing work, and engage with workers in their redesign.  As illustrated in the scope of the problems engendered in this simulation, leaders need to break down the barriers between departments in order to accomplish effective system redesign (Point 9).  Points 1, 2, and 12-14 undergird discussion of the importance of leadership in providing the context for better management of Cullowhee Airlines.

Point 10: eliminate slogans and targets. This experience shows that slogans and targets are ineffective and harmful, as are quotas and management by numerical goals (Point 11), a point that can be further illustrated by explaining the funnel experience (Deming, 1992).

We have included this summary with respect to Deming’s 14 Points to aid you in preparing to facilitate discussion.  You may want to focus on particular points, or leave identification of examples to students as an exercise.

Finally, we find that students readily identify issues raised by the present customer satisfaction instrument.  This realization provides the opportunity to initiate a discussion of designing and responsibly using data from a survey that could guide the efforts of this work group.

References:

Deming, W. E. (1986).  Out of the Crisis.  Cambridge, MA:  MIT CAES.

Deming, W. E. (1992)  The New Economics for Education, Government and Industry.  Cambridge, MA:  MIT CAES.

Fitzsimmons, J. A. and Fitzsimmons, M. J. (2000). Service Management, Operations, Strategy, and Information Technology 3^rd Ed. New York, NY:  McGraw-Hill, Inc.

Latzko, W. J. and Saunders, D. M. (1995). Four Days with Dr. Deming, A Strategy for Modern Methods of Management. Reading, MA:  Addison-Wesley Publishing Co.

Table 1

Exhibit 1.

Appendix.  P-chart example

      Figure 2. Dissatisfied customers per day per agent

*Clearly, five samples of five observations each is not enough data to establish valid control limits. However, for illustration purposes, this data works well.

In order to make a percentage defectives control chart (P-chart) one must find the center line (CL), the upper control limit (UCL) and the lower control limit (LCL) as well as the ps.

CL =  = d / (s*n) = 25/(5*25) = .2

Where: d is the total number of dissatisfied customers (total number of defectives)

s is the number of samples = five weeks

n is the sample size = five employees x five customers for each employee per week

p₁ (Week 1)  4/25 = .16,   p₂ (Week 2)  5/25 = .20,   Etc.