2014/04/29 by Asaf Weinstein, Daniel Yekutieli, Weinstein, Asaf +1
Agricultural and Biological Sciences · Computer Science · Mathematics · #Bayesian Modeling and Causal Inference #FOS: Computer and information sciences #Methodology (stat.ME) #Sensory Analysis and Statistical Methods #Statistical Methods and Inference
paper · pdf · doi:10.48550/arxiv.1404.7403
openalex publication_date 2014/04/29 · openalex created_date 2022/10/04 · openalex updated_date 2026/07/28
Given m unknown parameters with corresponding independent estimators, the\nBenjamini-Hochberg (BH) procedure can be used to classify the sign of\nparameters such that the expected proportion of erroneous directional decisions\n(directional FDR) is controlled at a preset level q. More ambitiously, our\ngoal is to construct sign-determining confidence intervals---instead of only\nclassifying the sign---such that the expected proportion of non-covering\nconstructed intervals (FCR) is controlled. We suggest a valid procedure which\nadjusts a marginal confidence interval in order to construct a maximum number\nof sign-determining confidence intervals. We propose a new marginal confidence\ninterval, designed specifically for our procedure, which allows to balance a\ntrade-off between power and length of the constructed intervals, and, in fact,\noften enjoy (almost) the best of both worlds. We apply our methods to detect\nthe sign of correlations in a highly publicized social neuroscience study and,\nin a second example, to detect the direction of association for SNPs with\nType-2 Diabetes in GWAS data. In both examples we compare our procedure to\nexisting methods and obtain encouraging results.\n