## WHAT THE HELL IS AN EVALUATION STRATEGY?

### Statistics is not a land of milk and honey, but most people find what they are looking for.

18 November 2013: Stephan Conrad

Jean-Claude Riber, a Franco-German stage director of opera said that statistics was like a Spanish guesthouse – everybody found in it what they were searching for. What a wise saying! After all, statistics is not a land of milk and honey where we just have to open our mouth and a roast chicken flies to us. However, it is even a dangerous saying since everybody might find what they seek, whether that is good or bad…

Therefore, Carl Hahn, former chairman of the Volkswagen group, thought that statistics had to be taken with a pinch of salt and should be applied with reason. This is a very good approach. There is even a quote of Paul Schnitker, former president of the German Confederation of Skilled Crafts, which supports it. He once said that statistics was like a building made of figures. Like a good house, statistics needed a solid foundation, clear outlines and the proof that they would keep their value in the course of time if they should be good. Bad statistics, by contrast, would collapse like a house of cards.

However, how can we ensure that statistics are built on a solid foundation? The aim is to describe available data through statistics gained in statistical data analyses, such as capability analyses, and to evaluate whether they meet the specified requirements. In order to do so, you generally have to take the following steps. You have to check the systematic and random variations of the data, describe them by means of suitable distribution (time) models and calculate capability indices from the statistics of the distribution.

Since there are numerous common distribution models, test procedures and ways to calculate statistics, this is where we enter the Spanish guesthouse. You just have to know what you want in order to be able – sometimes a little bit forcefully – to deduce the desired results. That’s a piece of cake – of course, we will succeed somehow in retrieving a green smiley…

But that is certainly not our request! We would like to gain reliable results based on a firm and solidly built foundation and we still want to take them with a pinch of salt and apply them with reason. That is why we have to arrange and define which methods and procedures to apply and how to apply them. This is the only chance that each colleague reaches the same correct and reliable results that can be used in later evaluations...