Simultaneous Estimation of Percentile Curves with Application to Salary Data

Abstract

A nonparametric iterative method is proposed for the simultaneous estimation of percentile curves. The method provides good descriptive statistics for salary data for which the assumptions of normality and equal variance are usually not valid. The method, a modified counting procedure based on minimum chi-squared estimation, is related to ordinary least squares through grafted polynomials and works even for nonlinear percentile curves.

Key Words:

• Percentile curves
• Grafted polynomials
• Minimum chi-squared estimation
• Least squares
• Linearization technique