Y. Benjamini, Y. E. Kling (Tel Aviv University, Israel)

Aspects of Multiplicity in Statistical Process Control (SPC)

The "Multiplicity Problem", is reviewed in the context of Statistical Process Control (SPC). Disregard of this issue, as is common practice, results in an inflated rate of false alarms. We examine the appropriateness of two overall error measures: the Familywise Error Rate (FWE) and the False Discovery Rate (FDR). We discuss a few selected SPC configurations that give rise to multiplicity: Multiple criteria on the same chart, controlling several aspects of a process, controlling multiple attributes of a product, and controlling the quality of the final product