Logistic Quantile Regression in Stata

Abstract

We present a set of Stata commands for the estimation, prediction, and graphical representation of logistic quantile regression described by Bottai, Cai, and McKeown (2010, Statistics in Medicine 29: 309–317). Logistic quantile regression models the quantiles of outcome variables that take on values within a bounded, known interval, such as proportions (or percentages) within 0 and 1, school grades between 0 and 100 points, and visual analog scales between 0 and 10 cm. We describe the syntax of the new commands and illustrate their use with data from a large cohort of Swedish men on lower urinary tract symptoms measured on the international prostate symptom score, a widely accepted score bounded between 0 and 35.