Confidence Intervals Construction for Relative Difference Under Binomial Sampling Based on MOVER Method
Relative difference,also known as relative risk reduction,mainly measures the additional impact of risk factors or treatment factors on individuals,and has im-portant clinical significance in epidemiological research.This paper presents nine in-terval estimation methods for relative differences using traditional interval estimation method and MOVER method under independent binomial sampling.The advantage of the MOVER method proposed in this paper is that by borrowing the MOVER method,the confidence interval of the relative difference can be constructed using the confidence intervals of two independent binomial distribution ratios.Compared with traditional interval estimation method,this method does not need to calculate the asymptotic variance of the relative difference in the process of constructing con-fidence intervals,and does not require the Fisher information matrix and its inverse matrix,which greatly simplifies the calculation.In addition,this paper investigates the performance of nine interval estimation methods under different parameter set-tings through Monte Carlo data simulation.The data simulation results show that the MOVER method can provide a more accurate confidence interval compared to traditional methods.Finally,this paper demonstrates the practical application of the proposed nine interval estimation methods through actual data cases.
NETL
NSTL
万方数据
