ARM, ANOVA AND ALL THE REST
ARM, ANOVA AND ALL THE REST
Expert insights into this complex topic
21 November 2011: Stephan Conrad
The days when calculation methods were explained with the help of a thumbnail sketch and some ranges are long gone. Back then, the magic of the truth was hidden in two k factors. Ever since the release of the 4th edition of the AIAG MSA manual, the method of ANOVA has always been recommended whereas ARM is only allowed in case of an “emergency”.
This fact seems to be justified by the fact that ARM only determines EV and AV components (repeatability and reproducibility) whereas the method of ANOVA also evaluates interactions IA. So this is enough information to enter into the discussion, maybe even enough to end it, but what is really behind it?
First of all, here is a brief description of type-2 study. Typically, 2-3 operators take 2-3 repeated measurements on ten different workpieces. This is an ordinary full factorial design of experiments.
What is the scope of ARM?
This method tries to answer the following questions:
- How precisely can the operator repeat his measurement?
- To what extent are the operators‘ measurements comparable to one another?