New tests for null hypothesis of non unity ratio of proportions

J Biopharm Stat. 2007;17(2):229-45. doi: 10.1080/10543400601177426.

Abstract

Testing for noninferiority and equivalence between an experimental therapy and a standard therapy in terms of the ratio of binomial proportions is considered. New tests based on the Fieller-Hinkley distribution of the ratio of random variables are proposed. Restricted maximum likelihood estimates of the null variances are used to derive the tests. Sample size determination is discussed. The proposed test procedure is extended to multiple tables. The tests are applied to numerical examples.

MeSH terms

  • Confidence Intervals
  • Controlled Clinical Trials as Topic / methods
  • Controlled Clinical Trials as Topic / standards
  • Controlled Clinical Trials as Topic / statistics & numerical data
  • Data Interpretation, Statistical*
  • Humans
  • Models, Statistical
  • Research Design / standards
  • Research Design / statistics & numerical data
  • Sample Size