Six Sigma: the difference between significance and relevance

Six Sigma can do magic tricks with statistics. Statistical tools are used and applied in a hyper intelligent way and with statistics everything can be significantly proven. Differences can be investigated, compared and concluded. Right? Yes, but only from a statistical point of view. A proper use of Six Sigma however, requires a deep understanding of the business and the business challenges ahead.

First, you have to identify potential root causes together with the process owners. From a Six Sigma perspective, the hypotheses have now been identified. The root causes need to be translated into a specific data definition and these data must be gathered by the business and be verified. If you are sure that it represents the process and its variance and that you have the real transaction data of the process, the proper statistical tools can be chosen and applied.

Relevance over significance

Once you have the conclusion based on statistical analysis, the relevance should be confirmed by the business. At this point it is important to distinguish ‘significance’ from ‘relevance’. If you notice a statistical relation between the root cause and the process variation, it doesn’t mean that you have found the real reason. For example, there is a significant relation between accidents in a swimming pool and the outside temperature. However, this is not relevant. The wrong argumentation would be: the warmer the weather, the more people visit the swimming pool, leading to higher amount of accidents. Furthermore, lowering the outside temperature (if ever possible) would reduce the turnover and income of the swimming pool. Wrong argument, wrong solution.

Root cause analysis

It would be more relevant to argue that running kids are leading to accidents. An effective solution might be to improve the sliding resistance of the floor. So, by selecting the relevant root cause, the variation will be reduced. Only the business itself can confirm and assure the relevance of the defined and confirmed root cause. Once a full analysis has been conducted, the real improvement project can be started.

Data-based improvement

If you find a relevant relation between the process variation and the root cause, you can achieve both data-based and fact-based improvement. And this improvement is a strong basis for success. Because in the end, if the business is supported with a strong confirmed improvement, this will lead to enhanced business results.

Frank Oudshoorn is business process consultant at R&G Global Consultants.

Lean Six Sigma – DMAIC