A Power Calculation Method of Closed Testing Procedure Based On Weighted Bonferroni Tests to Each Intersection Hypothesis by Integrating Joint Multivariate Distribution  

Oleh:

Okayama, Akifumi ; Nagata, Yasushi

Jenis: Article from Proceeding
Dalam koleksi: 12th ANQ Congress in Singapore, 5-8 Agustus 2014, page 1-6.
Topik: Multiple comparison; Power calculation; Closed Testing Procedure; Modified Bonferroni test; Process Inspection
Fulltext: QP1-1.3-P0239.pdf (450.85KB)

Isi artikel

In sampling inspection, when a product has n-characteristics (n > 1) to be inspected, each inspection for the characteristic can be regarded as a hypothesis test and multiple comparison problems arise. One of the multiple comparison problems is the inflation of the overall Type I error rate [i.e. Familywise Error Rate (FWER)]. Alpha inflation increases producer’s risk in manufacturing (e.g. Supplier rejects a batch though the batch complies with standards.). It is important to control the FWER for managing producer’s risk. In addition, it would be required to calculate the overall power for setting each inspection appropriately. Many multiple comparison methods that control the FWER have been proposed and applied. Especially, attention has come to be paid to closed testing procedure based on weighted Bonferroni tests to each intersection hypothesis due to its strict FWER control, comparatively high power and flexible adaptation. When characteristics to be inspected are correlated, the power of closed testing procedure based on weighted Bonferroni tests to each intersection hypothesis was calculated by simulation using multivariate pseudorandom numbers in previous studies. Contrastingly, this study provides a new power calculation method that integrates non-null joint density function by applying computational meshes. The improvement in accuracy at high efficiency is shown by comparing with the previous power calculation method

Opini Anda

Klik untuk menuliskan opini Anda tentang koleksi ini!

Kembali