Generalized Score Test for Zero Mixture in Count Data

• Activity Number: 127
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Wei-Wen Hsu*+ and David Todem
• Companies: Michigan State University and Michigan State University
• Keywords: Mixture models ; Score test ; Zero inflation/deflation ; Poisson model ; exponential family ; Binary regression

Abstract

In a 1995 Biometrics paper, Van Den Broek(1995,738-748) developed a score test for zero inflation in a Poisson distribution. Others have extended this test to independent and correlated discrete data with density function from the exponential family. But the existing testing procedures rely on restrictive assumptions. One such restriction is that the mixing weights are unrelated to covariates under alternative. Another limitation is that it is often unclear whether these tests are conducted under the hierarchical or the marginal representation of the mixture distribution. In this paper, we propose a generalized score test for homogeneity of zeros against alternatives from the marginal mixture model. The technique is based on a decomposition of the mixing weights into terms that have an obvious statistical interpretation. We exploit this decomposition to lay the foundation of the test.