The interval estimation of a binomial proportion has been one of the most important problems in statistical inference. The modified Wilson interval, Agresti–Coull interval, and modified Jeffreys interval have good coverage probabilities among the existing methods. However, as approximation approaches, they still behave poorly under some circumstances. In this paper, we propose an exact and efficient randomized plausible interval based on the inference model and suggest the practical use of its non-randomized approximation. The randomized plausible interval is proven to have the exact coverage probability. Moreover, our non-randomized approximation is competitive with the existing approaches confirmed by the simulation studies.
Three examples including a real data analysis are illustrated to portray the usefulness of our method.
Inferential model Binomial proportion Interval estimation Coverage probability Expected length
Mathematics Subject Classification
This is a preview of subscription content, log in to check access.
We thank the referees for the constructive comments that much improved the paper. The study is supported by grants from Guangdong Engineering Research Center for Data Science, Natural Science Foundation of Guangdong Province, China (2017A030313018), the Innovation Project of Graduate School of South China Normal University (2016lkwm73) and the National Institutes of Health (5UL1TR00108505 and P30 CA124435).
Agresti A, Coull BA (1998) Approximate is better than “exact” for interval estimation of binomial proportions. Am Stat 52:119–126MathSciNetGoogle Scholar
Agresti A, Gottard A (2005) Comment: Randomized confidence intervals and the Mid-P approach. Stat Sci 20:367–371CrossRefGoogle Scholar