PERCEIVED TRUTHS, RISKS AND STATISTICAL TESTS
PERCEIVED TRUTHS, RISKS AND STATISTICAL TESTS
Why it is not that easy to make decisions
14 March 2014: Roman Wenig
Our modern world is no longer imaginable without electronic data processing. You generate information from data. These pieces of information help you make decisions. We virtually face the decision-making process everywhere and permanently. Decisions control our processes and create our services and material products. However, this is not just about making decisions; it is also about how good these decisions are. You reach a decision on the basis of your experience or your personal assessment of the respective situation, but you may also come to a decision by applying a well-structured method based on hard figures, data and facts.
For economic reasons, companies normally control processes based on sample inspections, especially in industrial production. This fact compels us to accept risks. A sample inspection always incurs the risk to make a wrong decision. In order to calculate this risk, you necessarily have to master statistical methods. The basic methods include statistical tests.
There is currently no simple method dealing with the variation of our processes in an elegant and easier way but statistical tests. These tests offer a binary result provided with a calculated risk. Statistical tests separate chance from changes affecting a process systematically, they distinguish between signal and noise, they sort the wheat from the chaff. It is not possible to imagine a single step of the product life cycle without these tests. So they should be a standard tool for anyone making decisions based on figures, data and facts.
In order that the statistical test provides the desired result reliably, users have to master the methods on the one hand and must even meet certain requirements for the preparation, implementation and finishing process on the other hand. The following article describes this process step by step for parameter tests, i.e. tests determining whether the parameters of the population are significant. Each step includes specific questions that you have to answer and tasks you have to perform before taking the next step. Finally, an example illustrates this approach...