A Multiplicative Bias Correction Technique for Estimating Quantile Function with an Application

Abstract

Smooth non-parametric quantile function estimators on basis of symmetric kernels exhibit boundary bias due to spill-over near the edges. An improved non-parametric estimator of a quantile function under simple random sampling without replacement is proposed, based on a multiplicative bias corrected distribution function. There is no spill-over around the edges with our new quantile estimator. The proposed quantile estimator's asymptotic properties are investigated. The suggested method is compared to existing estimators using real data set findings, demonstrating the improved performance.