Skip to contentScholarGate
LibraryBookshelfDeskReview StudioAssistant
Sign in
On this page
IntuitionHow it worksWhen to use itStrengths & limitationsCommon pitfallsApplicationsFrequently asked🔒 Read the full methodSourcesRelated methods
Cite this pageSpotted an issue on this page? Report or suggest a fix →
Home›Econometrics›Poisson and Negative Binomial Regression
Regression model

Poisson and Negative Binomial Regression

Also known as: count regression, log-linear count model, negative binomial regression, Poisson / Negatif Binom Regresyon

Poisson regression is a generalized linear model for count outcomes — events tallied as non-negative integers such as hospital admissions, accidents, or article counts. It models the log of the expected count as a linear function of the predictors, and is developed in the standard count-data treatment of Cameron and Trivedi (1998); when the counts are over-dispersed, the closely related negative binomial model (Hilbe, 2011) is preferred.

ScholarGate
  1. Regression model
  2. v1
  3. 2 Sources
  4. PUBLISHED
Cite this page →
Tools & resources
Download slides
Learn & explore

Read the full method

Members only

Sign in with a free account to read this section.

Sign in

Method map

The neighbourhood of related methods — select a node to explore.

Poisson Regression
Logistic RegressionOLS RegressionPanel Fixed EffectsQuantile RegressionAge-Crime Curve ModelingBayesian Negative Binomi…Bayesian Poisson Regress…Bayesian Zero-inflated m…Capture-RecaptureCroston's Method

+14 more

When to use it

Use Poisson regression when the outcome is a count of events (non-negative integers) with a sample of at least about 30 observations, and you want to relate the event rate to one or more predictors. The core Poisson assumption is equidispersion: the conditional variance equals the conditional mean. Check this first — if a dispersion test shows variance greater than mean, switch to negative binomial regression. Also screen for excess zeros (a zero share above roughly 30% suggests a zero-inflated model) and for multicollinearity (VIF below 10). It is well suited to cross-sectional, panel, and longitudinal count data.

Strengths & limitations

Strengths
  • Purpose-built for count outcomes: the log link keeps predicted counts non-negative and gives multiplicatively interpretable effects.
  • Coefficients exponentiate into incidence rate ratios (IRR), a directly interpretable change in the event rate per unit of a predictor.
  • Extends naturally to the negative binomial model for over-dispersed counts without changing the log-linear structure.
Limitations
  • The basic Poisson model assumes the variance equals the mean; real count data are often over-dispersed, which understates the standard errors.
  • Excess zeros beyond what the Poisson distribution predicts require a zero-inflated (ZIP/ZINB) model instead.
  • Like other regressions, it is distorted by strong multicollinearity among predictors and needs an adequate sample size.

Frequently asked

When should I use negative binomial instead of Poisson?

When the counts are over-dispersed — the conditional variance exceeds the mean. A dispersion test (variance/mean > 1) flags this. Plain Poisson then understates the standard errors; the negative binomial model adds a dispersion parameter and corrects the inference.

What is an incidence rate ratio (IRR)?

It is the exponentiated coefficient, exp(β). An IRR of 1.2 means a one-unit increase in the predictor multiplies the expected event rate by 1.2, i.e. a 20% increase. IRRs are the standard way to report Poisson and negative binomial effects.

What if my data have too many zeros?

If the zero share is very high (roughly above 30%), the Poisson and negative binomial models may underfit the zeros. A zero-inflated model (ZIP or ZINB) separates the process that generates structural zeros from the count process and is usually more appropriate.

How do I check whether the model fits?

Examine the Pearson and deviance residuals for goodness of fit, and use the Vuong test to compare the Poisson model against the negative binomial alternative. A poor fit or significant over-dispersion points to the negative binomial or a zero-inflated specification.

Sources

  1. Cameron, A. C. & Trivedi, P. K. (1998). Regression Analysis of Count Data. Cambridge University Press. DOI: 10.1017/CBO9780511814365 ↗
  2. Hilbe, J. M. (2011). Negative Binomial Regression (2nd ed.). Cambridge University Press. DOI: 10.1017/CBO9780511973420 ↗

How to cite this page

ScholarGate. (2026, June 1). Poisson and Negative Binomial Regression. ScholarGate. https://scholargate.app/en/econometrics/poisson-regression

Related methods

Logistic RegressionOLS RegressionPanel Fixed EffectsQuantile Regression

Which method?

Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.

  • Logistic RegressionResearch Statistics↔ compare
  • OLS RegressionEconometrics↔ compare
  • Panel Fixed EffectsEconometrics↔ compare
  • Quantile RegressionEconometrics↔ compare
Compare side by side →

Referenced by

Age-Crime Curve ModelingBayesian Negative Binomial RegressionBayesian Poisson RegressionBayesian Zero-inflated modelCapture-RecaptureCroston's MethodGamma RegressionGeneralized Linear ModelHurdle ModelMultinomial LogitNegative Binomial RegressionNonlinear Panel Data AnalysisOrdinal RegressionQuantile RegressionRecurrent Event ModelRobust Negative Binomial RegressionRobust Poisson RegressionSpatial Interaction ModelZero-inflated modelZero-Inflated Negative Binomial RegressionZero-Inflated Poisson Regression

Similar methods

Negative Binomial RegressionPoisson Rate RegressionZero-Inflated Negative Binomial RegressionBayesian Negative Binomial RegressionZero-Inflated Poisson RegressionBayesian Poisson RegressionRobust Negative Binomial RegressionZero-inflated model

Related reference concepts

Binomial and Poisson DistributionsLogistic RegressionCox Regression ModelsRegression and CorrelationCategorical Data AnalysisPoisson Processes

Spotted an issue on this page? Report or suggest a fix →

ScholarGate — Poisson Regression (Poisson and Negative Binomial Regression). Retrieved 2026-07-21 from https://scholargate.app/en/econometrics/poisson-regression · Dataset: https://doi.org/10.5281/zenodo.20539026
Quick facts
Originator
Cameron & Trivedi (textbook treatment); Hilbe (negative binomial)
Year
1998
Type
Generalized linear model for count data
Estimator
Maximum likelihood (log link)
Outcome
count (non-negative integers)
Related methods
Logistic RegressionOLS RegressionPanel Fixed EffectsQuantile Regression
ScholarGate

A content-first reference library for research methods — what each one is, how it works, and where it comes from.

Open data (CC-BY)

Explore

  • Library
  • Search the library…
  • Browse by field
  • Fields
  • Journey
  • Compare
  • Which method?

Reference

  • Subjects
  • Atlas
  • Glossary
  • Methodology
  • Philosophy

Your tools

  • Bookshelf
  • Desk
  • Chat

Company

  • About
  • Pricing
  • Contact
  • Suggest a method

Entries are compiled from published sources for reference. Verifying the accuracy and suitability of any information for your own use remains your responsibility.

© 2026 ScholarGate · A research-method reference library
  • Privacy
  • Cookies
  • Terms
  • Delete account