Small sample performance of quantile regression confidence intervals

Abstract

Since the introduction of regression quantiles for estimating conditional quantile functions there has been ongoing research into how best to construct confidence intervals for parameter estimates. The three main methods are direct estimation, rank test inversion and resampling methods. Kocherginsky et al. [Practical confidence intervals for regression quantiles, J. Comput. Graph. Statist. 14 (2005), pp. 41-55] gave an overview of some of the available procedures. Five years on, the aim of this paper is to revisit and extend their analysis, evaluating additional techniques with a focus on smaller sample sizes and more extreme conditional quantiles. In particular, we find the percentile bootstrap (pbs) to be an eminently viable alternative for confidence interval construction. We show that it provides empirical coverage probabilities generally as good as, or better than, the other more complex resampling methods. Furthermore, pbs confidence intervals typically exhibit smaller average lengths across a variety of models than those based on the rank inversion methods which, like the pbs, avoids explicitly estimating asymptotic variances.

Who rebounds in the private transport sector? A comparative analysis between Beijing and Tokyo

“…However, because the main goal of this empirical analysis is to call attention to the heterogeneity of the rebound effect for vehicles, the strategy presented here to address the rebound effect could be applied to a larger data set in the future. In addition, despite the small sample size, the average rebound effect obtained from the OLS estimation is consistent with existing evidence, which may to some extent demonstrate the validity of the analysis here; in addition, many experiments have been performed that demonstrate that by using the resampling techniques (e.g., Markov Chain Marginal Bootstrap and percentile bootstrap), the estimation results of the quantile regression performed well, both in the large and small sample size scenarios, but with the recommendation of using larger sample sizes of at least 50-60 (Cade and Richards, 2006;Chernozhukov et al, 2009;Tarr, 2010). Second, vehicle efficiency is regarded as an exogenous variable to explain the vehicle kilometers traveled in this study, whereas it is very likely to be an endogenous variable.…”

Quantile Based Estimation of Scale and Dependence

“…This work demonstrates the competitive performance of our quantile based approach in a broad range of model designs with a focus on small and moderate sample sizes. These results were published in [5].…”

“…This work demonstrates the competitive performance of our quantile based approach in a broad range of model designs with a focus on small and moderate sample sizes. These results were published in [5].A reliable estimate of the scale of the residuals from a regression model is often of interest, whether it be parametrically estimating confidence intervals, determining a goodness-of-fit measure, performing model selection, or identifying …”

Spawning Locations within and among Tributaries Influence Flannelmouth Sucker Offspring Experience

“…We used the 90th quantile to first compare size at age among reaches across the entire sampling period and then size at age within sampling periods to determine critical growth periods (indicated by among‐ and within‐reach quantile slopes that differed from each other through time). To test responses for significant differences among each reach, we used 90% confidence intervals generated by the default rank inversion process in the quantreg package (Tarr ). We interpreted reaches to be significantly different at an age when the 90th quantiles diverged enough for their confidence intervals to not overlap.…”