Library: | glm |
See also: | glmest glmstat glmll |
Quantlet: | glmlrtest | |
Description: | glmlrtest performs a likelihood ratio test of two nested GLM. |
Usage: | {lr,alpha} = glmlrtest(loglik0,dim0,loglik1,dim1) | |
Input: | ||
loglik0 | scalar, log-likelihood of the hypothesis model. | |
dim0 | integer, dimension of the hypothesis model. | |
loglik1 | scalar, log-likelihood of the alternative model, loglik1 >= loglik0. | |
dim1 | integer, dimension of the hypothesis model, dim1 >= dim0. | |
Output: | ||
lr | scalar, value of the lr test statistic. | |
alpha | scalar, significance level at which hypothesis is rejected. |
library("glm") n = 100 x = normal(n) eps= normal(n) y = x/3+eps one =matrix(n) mod0=glmest("noid",one,y) ; constant regression mod1=glmest("noid",one~x,y) ; linear regression {lr,alpha}=glmlrtest(mod0.stat.loglik,1,mod1.stat.loglik,2) alpha
The significance level (smallest level) at which the constant regression model is rejected.