Skip to main content
Article
Testing homogeneity in semiparametric mixture case-control models
Communications in Statistics: Theory and Methods (2016)
  • C Z Di, Fred Hutchinson Cancer Research Center
  • G KC Chan, University of Washington
  • C Zheng, University of Washington
  • KY Liang
Abstract
Recently, Qin and Liang (Biometrics, 2011) considered a semiparametric mixture case-control model and proposed a score test for homogeneity. The mixture model is semiparametric in the sense that the density ratio of two distributions is assumed to be of exponential form, while the baseline density is unspecified. In a family of parametric admixture models, Di and Liang (Biometrics, 2011) showed that the likelihood ratio test statistics, which is equivalent to a supremum statistics, could improve power over score tests. We generalize the likelihood ratio or supremum statistics to the semiparametric mixture model and demonstrate the power gain over the score test. We also provide another justification of the proposed test statistics from an empirical likelihood perspective.
Publication Date
Summer June 1, 2016
Citation Information
C Z Di, G KC Chan, C Zheng and KY Liang. "Testing homogeneity in semiparametric mixture case-control models" Communications in Statistics: Theory and Methods (2016)
Available at: http://works.bepress.com/di/13/