Estimating Partitions of the Distribution of P-Values

• Activity Number: 356 - Contributed Poster Presentations: Business Analytics/Statistics Education Interest Group
• Type: Contributed
• Date/Time: Tuesday, July 30, 2019 : 10:30 AM to 12:20 PM
• Sponsor: Business Analytics/Statistics Education Interest Group
• Author(s): Robert Pavur* and Kellie Keeling
• Companies: University of North Texas and University of Denver
• Keywords: p-value; mixture distribution; false positives

Abstract

This study will examine the accuracy of estimating the occurrences of false positive, false negatives, as well at true-hypothesis conclusions by assessing an estimated mixture distribution of p-values. Extracting the uniform distribution from the estimated mixture distribution of p-values allows for insight into the occurrence of false positives and false negatives. Many proposed methods allow for control of false positives, which may distort conclusions. The proposed research focuses on the partitioning of estimated density functions for the p-value and assesses its reliability through a simulation study. A limiting factor in the study of p-values is that many p-values are from the same study and may not be independent. This study examines aspects of empirical studies that may limit the valid interpretations of the distribution of p-values and its partitions.