Introduction/Need for the Innovation

Students need to practice statistical process control (SPC) in order to learn its mechanics and, perhaps more importantly, to understand its role.  Textbook problems reinforce the mechanics.  However many problems fall short in giving students an appreciation of how and why SPC is used.  Thus many students do not gain an appreciation for the true power of the techniques.   Typical textbook problems provide students with the characteristics of k number of samples and instruct students to construct and interpret a chart using just these points.  Many students dwell on the calculations themselves because of their complexity.   By contrast, the critical application of SPC in industry is monitoring a process on an ongoing basis—by taking samples and plotting them after the control limits are calculated and the initial k points are plotted.  One objective of the exercise described here is to create an experience that, instead of emphasizing chart-making, demonstrates SPC’s role in process monitoring. 

Additionally, the exercise described here is intended to underscore SPC’s ability to help businesspeople link cause and effect.  Many existing SPC exercises move students beyond “crunching the numbers,” and they can be excellent at demonstrating concepts such as the relationship between sample size and Type I and Type II errors.   However many of these, such as Deming’s red bead experiment (Walton, 1986), use steady state systems, such as a single bowl of beads.  Others, such as (Umble & Umble, 2000), model process changes (for example, by substituting one bowl of beads for another), but they announce to students when these changes occur.    By contrast, the objective of the game below is to challenge students to use SPC to determine for themselves if a change in the process has occurred. 

In general, a business process takes place in the context of constantly changing conditions and events—new suppliers, temporary help, changes in weather and so on.  Each of these has the potential to affect a process; however, only a few actually do so.  SPC, of course, helps identify these.  The exercise described here emphasizes this aspect of SPC.   In the game students monitor a process on an on-going basis--after constructing a chart.  Moreover, after some time passes, an event occurs (a new raw material supplier), and it is up to students to use SPC to detect whether or not that change significantly affects the production process.

Mechanics of the Game

The game can be played in about 30 minutes.  It is simple enough to be used in larger sections of the introductory operations management class.  The exercise is designed to be administered in class after (preferably next class period) students have been taught SPC mechanics and calculations, including the reasons a process can go out of control (e.g. change in raw material characteristics, new employee with inadequate training). Detailed instructions for students are in the appendix. 

The instructor divides the class into groups of 4 and announces that each group will be running a business process: The process is that of shuffling a deck of cards.  The shuffling process “produces” 4 different suits (spades, clubs, diamonds and hearts).  Hearts, students are told, are defective.  Shortly, students will begin shuffling the deck and then taking a sample from the process by looking at the first 20 cards after each shuffle.  Instructors may wish to tell students that the shuffling process represents a particular business process (and perhaps adapt the hand-out and what not to correspond to the particulars of that process).  The most direct analogies can be made to processes with one material input and one output, such as many conversion processes.  Good candidates are re-rolling steel, roasting peanuts, extruding aluminum, or molding plastic

First however, in order to save time, the instructor announces that the process has been operating already and that someone has already collected 10 samples over time (These data appear in the table in the appendix/handout).  Students are instructed to use this data to calculate control limits and plot the 10 points.  The instructor should make sure that, after completing this chart, students understand that the process is in-control with a mean of 25% defective.

A deck of cards is distributed to each team.  The deck is normal (25% hearts), but students are not told this.  Students are instructed to engage in the shuffling process, take samples, plot them, and determine whether the process is still under control.  Since all teams have a normal deck, and the 10 initial points from the appendix were from a normal deck, the chances of an out-of-control determination from shuffling and sampling from this deck are quite small. 

The New Supplier

Next, it is announced that management has replaced the current supplier to save cost.  So all teams will receive a new (replacement) deck of cards.  The supplier has guaranteed that the characteristics (i.e. proportion of hearts) of her decks are identical to the old supplier’s.  However, it is up to students to verify this by continuing to monitor the shuffling process using SPC (Step F). 

For this step, the instructor must prepare a second set of decks before class--one deck per team.  Half of the decks in this second set should be normal decks (25% hearts).  The other half should have more or less than 25% non-conforming units (hearts).  Therefore, take 6 hearts out of a quarter of the decks and swap them with 6 non-hearts out of another quarter of the decks.  (The number 6 is somewhat arbitrary, but it yields charts that show an out of control situation fairly reliably in less than 10 samples).  This results in 19 hearts in one quarter of the decks and 7 in another quarter with the other half being normal decks. (Students are not told whether or not their second deck is the same as the first.)  

When students have made a decision as to whether their deck/process is still in control, instruct them to count the number of hearts in their entire deck in order to see if they were right. 

Rules of interpretation from the text my students use are in the appendix.  Using these, the altered decks will most likely show out of control by the 4 of 5 rule or the 8 consecutive rule. 

However, the author suggests not instructing students to draw a specific number of samples from the new deck.  Rather let each team determine when to declare (to the instructor and the class), whether or not their process is under control.  This brings up the opportunity to discuss sampling after the game. 

Blaming Workers for Variation

During the game the author encourages student “supervisors” to berate workers who appear to be having a bad run (step E5).  This, of course, brings up an opportunity after the game to discuss the merits of attributing results to employees versus to work systems—the characteristics of the raw materials in this case.

Evidence of Effectiveness

            This section presents three types of evidence of effectiveness: (1) a 1-way ANOVA with the game as the experimental treatment, (2) student survey data, and (3) adoption by other faculty.  In the experiment, the author’s 2 sections (of approximately 35 students each) of the intro OM course were both given an SPC lecture on day one, and they were assigned some review problems to prepare for an announced quiz on day 2.  During class on day 2, both sections played the game.  One section (“control group”) took the quiz immediately before the game, while the other section (“treatment group”) took the quiz immediately after the game.  Therefore differences in quiz performance can be reasonably attributed to the game’s effect on student’s understanding.  As table 1 indicates, the treatment group outperformed the control group by 1.1  (p<.001, SAS 8.0 “proc ANOVA” procedure).  The difference in standard deviation, suggests that the exercise may also narrow gaps between top and bottom performers.

Table 1. One-Way ANOVA Results

After the game, the author administered a brief anonymous questionnaire to all students.  Results were very positive (table 2).  Perhaps more illustrative are the comments students provided on the optional comments section of the questionnaire.  Of the 21 comments received, 20 were positive and 1 was neutral to negative (“more emphasis is needed with regard to which situation a supplier should be notified”).  Representative positive comments include, “As a mktg. major, my career will never deal with q.c. in terms of mfg. This game got me listening and involved in class;” and “Helps provide examples of how real business uses it and where to go from after the data.”

Table 2. Questionnaire Data

Finally, the game has been adopted by one other faculty member at the author’s university and by one at another school (Northern Illinois University). 

REFERENCES

Umble, E. J., & Umble, M. M. (2000). Developing Control Charts and Illustrating Type I and Type II Errors. Quality Management Journal, 7(4).

Walton, M. (1986). The Deming Management Method. New York: Dodd, Mead.