Properties of a one-sided likelihood ratio test as applied to pool screening

Academic Article

Abstract

  • Copyright © Taylor & Francis Group, LLC. In the control of tropical disease, data are typically collected using pool screening design. When eradication programs have been in place for a period of time, researchers are interested in testing whether the disease prevalence is below a certain target level using one-sided likelihood ratio test (LRT). We will investigate the finite and largesample properties of this test statistic. Based on simulations, if the number of pools is not large enough, the LRT has inflated type I error rate. Consequently, researchers will make the critical mistake of stopping the treatment when the target level has not been reached.
  • Authors

    Digital Object Identifier (doi)

    Pubmed Id

  • 25733157
  • Author List

  • Gao H; Aban IB; Katholi CR
  • Start Page

  • 4546
  • End Page

  • 4565
  • Volume

  • 43
  • Issue

  • 21