The Great M&M’s Experiment
by Deborah Rindfuss Ellis

hen organizations with multiple locations are faced with budget reductions, they are often torn between providing essential training and exceeding budget limits due to travel costs. Training developed at the corporate level can be expensive as it often requires a trainer to visit multiple sites or employees from several locations to travel to a common site.

The Great M&M’s Experiment is a training model that allows multisite organizations to use e-mail, phone conferencing and a networked computer system to provide effective and efficient machine capability training.

Machine capability is a statistical technique used to analyze the variability of a machine in relation to the blueprint specifications. Capability analysis is generally applied in situations in which variable characteristic measurement—of length, thickness or diameter—takes place.

In an effort to enhance the user friendliness of this training, the number of candies in a 1.69-oz. bag of M&M’s is used to represent products. In the real world, however, the variable characteristic of weight would be analyzed instead of the count.


Planning for the Great M&M’s Experiment should take place about one month before the training. During this phase, the trainer should:

1. Obtain commitment from participant supervisors.

2. Determine participant names, locations and training date.

3. Schedule phone conference date, time and duration. Approx-imately two hours should be set aside to conduct the training and answer questions.


Every effort should be made to create an aura of excitement for the training. Each participant should:

1. Purchase at least one 1.69-oz. bag of plain M&M’s. A total number of 30 bags is needed to conduct the experiment, so if there are 15 participants, each should purchase two bags.

2. Count his or her M&M’s and record the total on a pre-established spreadsheet located in a shared network file. When participants become curious about the number of M&M’s obtained by other participants, the trainer should explain the concept of variation.

The trainer should e-mail the material to all participants one week before the training so they can become familiar with it. He or she should also consider including a photo of a bag of M&M’s in the participant materials because, in one case, a foreign-born participant was unfamiliar with M&M’s.

Training Package Development

Once the participants provide the sample data (number of M&M’s per bag), the trainer can use them to develop the training material. He or she should focus on mean, standard deviation, normal curve properties, Cr (capability ratio), Cp (capability index) and Cpk (centering capability index), while keeping in mind the audience’s diverse education levels.

The trainer should also include the use of statistical software to generate capability parameters and graphs.


On the day before the training, the trainer should remind participants of the conference call time. In a multisite organization, the possibility of time zone misunderstandings is common and needs to be prevented.

The participants should be seated near a networked computer during the conference call; they will use the computer to view the material and obtain access to previously loaded training examples.

Once training begins, each participant should introduce him or herself to help create a classroomlike atmosphere. The initial discussion should focus on the M&M’s data spreadsheet.

Next, the trainer should use the 30-piece M&M’s data to teach the following concepts:

1. Sample vs. population mean, median, mode, range and standard deviation.

2. Normal curve properties, such as 69.3% of all 1.69-oz. bags contain [mean +/-1 standard deviation] M&M’s, and 95.5% of all bags contain [mean +/-2 standard deviations] M&M’s.

3. Cr, Cp, Cpk: To teach these concepts, the trainer should develop fictitious specification limits, such as 56+/- 4 M&M’s.

At the end of the training, the participants should perform a capability study on an operation at their facility. The trainer can store the results in a shared folder so all participants can view the others’ input.


One month later, the trainer should conduct a brief follow-up conference call to review the capability results. The trainer can discuss the concept of continuous improvement and ways to improve low Cpk values, such as centering the process, reducing variability or increasing specification limits.

DEBORAH RINDFUSS ELLIS is the director of quality at Balzers Inc. in Amherst, NY. She earned a doctorate in higher education leadership and policy from the State University of New York at Buffalo. Ellis is an ASQ Senior Member, certified quality engineer and recipient of the ASQ Century Club Award for sponsoring more than 100 new ASQ members.

104 I JUNE 2004 I www.asq.org

A model for multisite machine capability training.


If you would like to comment on this article, please post your remarks on the Quality Progress Discussion Board at www.asq.org, or e-mail them to editor@asq.org.


Average Rating


Out of 0 Ratings
Rate this article

Add Comments

View comments
Comments FAQ

Featured advertisers