*Best of the Best vs. Worst of the Worst.
Rather than measuring total production, we take the a few of the best parts and the a few of the worst, and we look for what sets them apart.
The key is efficiency: by exaggerating the contrast, we highlight the important elements using very few parts.
We are forming pairs BOB / WOW, and for each one, we enter the variables to compare, quantitative or qualitative. Ellistat provides a p-value and a conclusion : Does this variable distinguish between the good and the bad?
💡 Method of roughing, to be used at the beginning of an investigation when it is not yet clear where to look. It identifies suspects; it does not establish causality. Once the factors have been identified, a design of experiments confirms and quantifies them.
⚠️ Selecting outliers creates a deliberate bias: the differences observed in the variables are amplified compared to normal production. Do not directly apply the same ranges.