Term: Generalized Linear Model (GLM)

Glossary Definition

Last Updated: 2006-10-20

Definition:

A unified class of models for regression analysis of independent observations of a discrete or continuous response. A characteristic feature of generalized linear models is that a suitable non-linear transformation of the mean response is a linear function of the covariates. Generalized linear models provide a unified method for analyzing diverse types of univariate responses (e.g., continuous, binary, counts). Generalized linear models are actually a collection of regression models and they include as special cases the standard linear regression for normally distributed continuous outcomes, logistic regression models for a binary outcome, or Poisson regression models for counts (Fox J., 1997). Estimate statements in generalized linear models provide both a parameter estimate based on the regressions and a statistical test of linear combinations of fixed effects. Contrast statements can be used to produce statistical test results.

Related concepts 

Related terms 

References 

Term used in