Communications for Statistical Applications and Methods

The Korean Statistical Society (한국통계학회)
권호 표지

On the Interval Estimation of the Difference between Independent Proportions with Rare Events

  • im, Yongdai (Hankuk University of Foreign StudiesYong-in) ;
  • Choi, Daewoo (Hankuk University of Foreign StudiesYong-in)
  • Published : 2000.08.01
Download PDF
⟨ Previous Next ⟩

Abstract

When we construct an interval estimate of two independent proportions with rare events, the standard approach based on the normal approximation behaves badly in many cases. The problem becomes more severe when no success observations are observed on both groups. In this paper, we compare two alternative methods of constructing a confidence interval of the difference of two independent proportions by use of simulation. One is based on the profile likelihood and the other is the Bayesian probability interval. It is shown in this paper that the Bayesian interval estimator is easy to be implemented and performs almost identical to the best frequentist's method -the profile likelihood approach.

Keywords

References

  1. Mathmkatical Statistics: Basic Ideas and Selected Topics Bickel, P.J.;Doksum, K.M.
  2. Journal of the American Statistical Association v.78 Binomial confidence intervals Blyth, C.R.;Still, H.A.
  3. BayesianData Anaysis Gelman, A.;Carlin, J.B.;Stern, H.S.;Rubin, D.B.
  4. The American Statistician v.35 Confidenceintervals for a binomial parameter after observing no successes Louis, T.A.
  5. Statistics in Medicine v.17 Interval estimation for the difference between independent proportions: comparison of eleven methods Newcombe, R.G.
  6. Journal of the Royal Statistical Society, B v.55 Bayesian intervals with good frequentist behavior in the presence of nuisance parameters Nicolaou, A.