Is Deming's Red Bead Experiment Misleading?

• Activity Number: 653
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Quality and Productivity Section
• Author(s): Charles Goldsmith*+
• Companies: Simon Fraser University
• Keywords: Red beads ; Rating employees ; Common cause variation ; Power

Abstract

Deming's Red Bead Experiment was introduced to show that workers are not responsible for the outcome of a process because it is management's responsibility. After using the red bead experiment in my quality improvement class for years, exactly once has an analysis of the data showed the worker was a possible factor to explain red bead variation. The data usually had a total sample size of around 30. Analyses suggested that the variation seen in the data is explained as common cause variation. Subsequent power analysis showed that workers would need many more replicates to find either replicate or worker to be important factors in the data. Deming and others would not attribute the output of a process to the workers or other factors, and as a result it is management's job to make the process better, not worker responsibility. As usually designed, the red bead experiment is destined to find no statistically significant effect of the two possible factors because the experiment has too little power to detect what might be considered to be important effects on process outcome. Examples from my courses will be used to illustrate these points in the talk.