Confidence Intervals for Kernel Density Estimation

Abstract

This article describes asciker and bsciker, two programs that enrich the possibility for density analysis using Stata. asciker and bsciker compute asymptotic and bootstrap confidence intervals for kernel density estimation, respectively, based on the theory of kernel density confidence intervals estimation developed in Hall (1992b)and Horowitz (2001). asciker and bsciker allow several options and are compatible with Stata 7 and Stata 8, using the appropriate graphics engine under both versions.