Comparison of conditional main effects analysis to the analysis of follow‐up experiments for separating confounded two‐factor interaction effects in 2 IVk−p fractional factorial experiments. (17th February 2020)

• Record Type: Journal Article
• Authors: Lawson, John
• Is Part Of: Quality and reliability engineering international. Volume 36:Number 4(2020)
• Page Start: 1454
• Page End: 1472
• Publication Date: 2020-02-17
• Subjects: aliased effects -- CME analysis -- confounding -- effect heredity -- effect sparsity -- hierarchical ordering; Reliability (Engineering) -- Periodicals; Quality control -- Periodicals; High technology -- Periodicals; 620.00452
• DOI: 10.1002/qre.2638
• Languages: English
• ISSNs: 0748-8017

Abstract

Abstract: Two‐factor interactions, where the effect of one factor depends on the level of another factor, are common, and understanding them is often the key to solving quality problems or making process improvements using designed experiments. Resolution IV 2 k − p fractional factorial designs are efficient and require fewer experiments or runs than resolution V or full factorial experiments. However, two‐factor interactions are confounded with other two‐factor interactions in resolution IV designs and their effects cannot be separated. Follow‐up experiments have been recommended in the literature to separate the effects of significant but confounded strings of two‐factor interactions in resolution IV designs. Recently, an analysis based on conditional main effects (or CMEs) has been shown to be useful in determining which interaction in a confounded string of two‐factor interactions is actually causing the significance without the need for follow‐up experiments. In this article, I investigate the value of this method of analysis by comparing its use with the analysis of follow‐up experiments using the data from three published experiments where follow‐up experiments were used to "de‐alias" confounded interactions.