2. Observing results

Today, observation remains an essential part of the engineer's art. On the quality of observation will depend the value of his decisions.

However, engineers have no a priori right to assume that observations are – or even should be – always identical. This is why he must approach the art of observation by adding the "statistical attitude" to his physicist's mental panoply.

By repeating the same measurement several times, the engineer will find results that are not identical, but fairly close to each other. The diameter and eccentricity measurements of shafts manufactured one after the other on the same machine by the same team are all different. But they are within a tolerance range that may be small, and they are naturally close to the target dimension. In other words, the engineer observes a dispersion around a central tendency. How can...