Bayesian Criteria for Non-Zero Effects Detection Under Skew-Normal Search Model

Abstract

Shirakura et al. [12] introduced search probability (SP) in order to compare search designs (SD). Afterwards, the SP-based and other related criteria were developed, all for the normal model. In the present study, we considered a general underlying skew-normal (SN) model and obtained new criteria in a simple explicit form using the Bayesian approach. These criteria are design[1]dependent and hence are able to rank SDs with respect to their search performance.

Keywords

• Bayesian approach
• Kullback–Leibler distance
• search design
• search linear model
• skew-normal distribution