Date of Award

1-1-2011

Document Type

Campus Access Dissertation

Department

Epidemiology and Biostatistics

Sub-Department

Biostatistics

First Advisor

Matteo Bottai

Abstract

We describe and compare methods for constructing confidence intervals for quantile regression coefficients. We consider methods based on resampling, sparsity estimation, and test-inversion. In the latter group, along with the popular rank-score, we include methods based on linear and logistic regression that exploit the direct relationship between quantile function and probability functions. These might prove practical alternatives to other more popular approaches and can be applied to dependent data, as those that arise in longitudinal, cluster, spatial, and complex survey designs. Results of a simulation study seem to indicate that they may have correct coverage and similar or sometimes narrower confidence intervals than the other methods considered.

Share

COinS