Testing the linearity of negative binomial regression models

Research output: Contribution to journalArticle

Abstract

The negative binomial (NB) is frequently used to model overdispersed Poisson count data. To study the effect of a continuous covariate of interest in an NB model, a flexible procedure is used to model the covariate effect by fixed-knot cubic basis-splines or B-splines with a second-order difference penalty on the adjacent B-spline coefficients to avoid undersmoothing. A penalized likelihood is used to estimate parameters of the model. A penalized likelihood ratio test statistic is constructed for the null hypothesis of the linearity of the continuous covariate effect. When the number of knots is fixed, its limiting null distribution is the distribution of a linear combination of independent chi-squared random variables, each with one degree of freedom. The smoothing parameter value is determined by setting a specified value equal to the asymptotic expectation of the test statistic under the null hypothesis. The power performance of the proposed test is studied with simulation experiments.

Original languageEnglish (US)
Pages (from-to)1013-1025
Number of pages13
JournalJournal of Statistical Computation and Simulation
Volume85
Issue number5
DOIs
StatePublished - Mar 24 2015

Fingerprint

Binomial Model
Negative Binomial
Linearity
Splines
Covariates
Regression Model
Penalized Likelihood
B-spline
Null hypothesis
Testing
Knot
Statistics
Negative Binomial Model
Likelihood Ratio Test Statistic
Chi-squared
Count Data
Smoothing Parameter
Poisson Model
Null Distribution
Limiting Distribution

Keywords

  • B-spline
  • overdispersion
  • penalized likelihood ratio test
  • semiparametric negative binomial regression

ASJC Scopus subject areas

  • Applied Mathematics
  • Statistics and Probability
  • Modeling and Simulation
  • Statistics, Probability and Uncertainty

Cite this

Testing the linearity of negative binomial regression models. / Li, Chin-Shang.

In: Journal of Statistical Computation and Simulation, Vol. 85, No. 5, 24.03.2015, p. 1013-1025.

Research output: Contribution to journalArticle

@article{213970b17ce44f70955bb34bf4b50ffe,
title = "Testing the linearity of negative binomial regression models",
abstract = "The negative binomial (NB) is frequently used to model overdispersed Poisson count data. To study the effect of a continuous covariate of interest in an NB model, a flexible procedure is used to model the covariate effect by fixed-knot cubic basis-splines or B-splines with a second-order difference penalty on the adjacent B-spline coefficients to avoid undersmoothing. A penalized likelihood is used to estimate parameters of the model. A penalized likelihood ratio test statistic is constructed for the null hypothesis of the linearity of the continuous covariate effect. When the number of knots is fixed, its limiting null distribution is the distribution of a linear combination of independent chi-squared random variables, each with one degree of freedom. The smoothing parameter value is determined by setting a specified value equal to the asymptotic expectation of the test statistic under the null hypothesis. The power performance of the proposed test is studied with simulation experiments.",
keywords = "B-spline, overdispersion, penalized likelihood ratio test, semiparametric negative binomial regression",
author = "Chin-Shang Li",
year = "2015",
month = "3",
day = "24",
doi = "10.1080/00949655.2013.860138",
language = "English (US)",
volume = "85",
pages = "1013--1025",
journal = "Journal of Statistical Computation and Simulation",
issn = "0094-9655",
publisher = "Taylor and Francis Ltd.",
number = "5",

}

TY - JOUR

T1 - Testing the linearity of negative binomial regression models

AU - Li, Chin-Shang

PY - 2015/3/24

Y1 - 2015/3/24

N2 - The negative binomial (NB) is frequently used to model overdispersed Poisson count data. To study the effect of a continuous covariate of interest in an NB model, a flexible procedure is used to model the covariate effect by fixed-knot cubic basis-splines or B-splines with a second-order difference penalty on the adjacent B-spline coefficients to avoid undersmoothing. A penalized likelihood is used to estimate parameters of the model. A penalized likelihood ratio test statistic is constructed for the null hypothesis of the linearity of the continuous covariate effect. When the number of knots is fixed, its limiting null distribution is the distribution of a linear combination of independent chi-squared random variables, each with one degree of freedom. The smoothing parameter value is determined by setting a specified value equal to the asymptotic expectation of the test statistic under the null hypothesis. The power performance of the proposed test is studied with simulation experiments.

AB - The negative binomial (NB) is frequently used to model overdispersed Poisson count data. To study the effect of a continuous covariate of interest in an NB model, a flexible procedure is used to model the covariate effect by fixed-knot cubic basis-splines or B-splines with a second-order difference penalty on the adjacent B-spline coefficients to avoid undersmoothing. A penalized likelihood is used to estimate parameters of the model. A penalized likelihood ratio test statistic is constructed for the null hypothesis of the linearity of the continuous covariate effect. When the number of knots is fixed, its limiting null distribution is the distribution of a linear combination of independent chi-squared random variables, each with one degree of freedom. The smoothing parameter value is determined by setting a specified value equal to the asymptotic expectation of the test statistic under the null hypothesis. The power performance of the proposed test is studied with simulation experiments.

KW - B-spline

KW - overdispersion

KW - penalized likelihood ratio test

KW - semiparametric negative binomial regression

UR - http://www.scopus.com/inward/record.url?scp=84920654142&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84920654142&partnerID=8YFLogxK

U2 - 10.1080/00949655.2013.860138

DO - 10.1080/00949655.2013.860138

M3 - Article

VL - 85

SP - 1013

EP - 1025

JO - Journal of Statistical Computation and Simulation

JF - Journal of Statistical Computation and Simulation

SN - 0094-9655

IS - 5

ER -