How much is enough?

I thought it was an interesting article, where the author Blanton Godfrey, pointing out that in Six Sigma, changes are made without sufficient proof of validity. The advantage in manufacturing and servicing environments is the verification of benefits from such changes; through ways of collecting data, strengthening the evidence. But, data will not always give positive results. Instead, the way to have significant results is to increase the power of detecting changes and reducing the scope of experiment.

The common phrase used in Six Sigma training and practice “failed to reject the null hypothesis” shows the presence of flaws, hindering in achieving significant results. The reasons: small sample size, correlations in variables, choosing the wrong model and the inability to detect changes.

During a speech at the Institute for Healthcare Improvement National Forum, Don Berwick (CEO) raised a critical question; when do you have sufficient proof of success to implement what you feel is a positive change?

In response, the author mentioned there is a need for additional statistics tools; do more real-time or unplanned experiments to gain insight and information of the organization. Generally, organizations fail in collecting and analyzing data carefully. The changes rarely get implemented 100% by all employees, at the same time. For example; at hospitals, a new drug is introduced – some doctors’ start administering it, while others continue to prescribe the older, more familiar drug. The implementation of process change in some organization also results in some employees immediately applying – while others take time to adapt.

An efficient and thorough study of the collected data and information will gain beneficial change in well designed and planned experiments.

ASQ Six Sigma Forum Magazine; Feb 2008; 7, 2; Retrieved on May 31st from ABI/INFORM Global