Testing Homogeneity in a Contaminated Normal Model with Correlated Data

• Activity Number: 653
• Type: Contributed
• Date/Time: Thursday, August 13, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Meng Qi* and Richard Charnigo
• Companies: University of Kentucky and University of Kentucky
• Keywords: Microarray analysis ; Mixture model ; Homogeneity test

Abstract

In this talk, we consider the problem of testing homogeneity in a contaminated normal model, when the data is correlated under some known covariance structure. To address this problem, we developed a moment based homogeneity test, assuming the data has a known compound symmetric covariance structure, and designed the weight for test statistics to increase power. We did simulations to assess size and power of the test and established asymptotic properties. In a case study, we applied our test to microarray about Down's syndrome caused by an extra copy of chromosome 21. By assuming different covariance parameters, we got a contour plot of p-values from our test, showing that failing to take into account correlation may massively understate the p-value.