A Likelihood-Based Approach for Multivariate One-Sided Tests with Missing Data

• Activity Number: 554
• Type: Topic Contributed
• Date/Time: Wednesday, August 3, 2016 : 10:30 AM to 12:20 PM
• Sponsor: Scientific and Public Affairs Advisory Committee
• Author(s): Guohai Zhou* and Lang Wu and Rollin Brant and J Mark Ansermino
• Companies: University of British Columbia and University of British Columbia and University of British Columbia and
• Keywords: Likelihood method ; Missing data ; Multivariate data ; Order-restricted inference

Abstract

There has been extensive research on multivariate one-sided or order-restricted hypothesis testing in the past few decades. In practice, multivariate data often contain missing values since it may be difficult to observe all values for each variable. However, although missing values are common for multivariate data, statistical methods for multivariate one-sided tests with missing values are quite limited. In this article, we develop two likelihood-based methods for multivariate one-sided or order-restricted tests with missing values, where the missing data patterns can be arbitrary and the missing data mechanisms may be non-ignorable. We derive some asymptotic results, evaluate our new tests using simulations, and illustrate the methods through a recent dataset and obtain new findings which are previously unavailable.