p_value.glm {glmglrt}

Computing p-values of hypothesis tests on coefficients of Generalized Linear Models and other

Description

This S3 method is a specialization of parameters::p_value for stats::glm models. By default, it computes Wald's P-values that are known to be more biased than LRT P-values, but the behavior can be overriden by the method="LRT" argument. This is for compatibility with the default method of parameters::p_value.

Usage

## S3 method for class 'glm'
p_value(
  model,
  method = NULL,
  parm = NULL,
  alternative = c("two.sided", "less", "greater"),
  H0 = 0,
  debuglevel = 1,
  force = FALSE,
  ...
)

Arguments

model

glm object; as obtained by calling stats::glm or MASS::glm.nb.

method

character value; may either be "LRT" (synonym "Chisq"), "Rao", "wald" (default value, synonym "Wald" and "F").

parm

integer or character vector or NULL; specify coefficients to test, by name or indexes. the default parm=NULL outputs all coefficients.

alternative

a character string specifying the alternative hypothesis, must be one of "two.sided" (default), "greater" or "less". You can specify just the initial letter.

H0

numeric vector of length 1 or of the same length as parm; the value of the coefficient under the null hypothesis. Zero by default.

debuglevel

integer value; set to 0 (default) to disable warnings, 1 to enable warnings and 2 to enable warnings and notes.

force

logical; if TRUE, force computation of P-values in case of convergence problems.

...

Ignored arguments. Allows compatibility with the generic parameters::p_value.

Value

a data.frame with two columns; the first column, Parameter represents the name of the coefficient and p (second column) represents the P-value of the hypothesis test against H0

Examples

require("parameters")
mod = glm(family="poisson", c(2,30) ~ c(0,1), offset=log(c(8,30)))
# Wald's tests (biased)
p_value(mod)
# Rao score tests (biased)
p_value(mod, method="Rao")
# LRT tests (less biased)
p_value(mod, method="LRT")

# only test slope (faster since only one test is performed)
p_value(mod, method="LRT", parm=2)
# is slope greater than log(2) ?
p_value(mod, method="LRT", parm=2, H0=log(2), alternative="greater")