automultinomial (version 1.0.0)

significanceTest: Significance tests for autologistic/automultinomial models

Description

Performs two-sided hypothesis tests for each coefficient in a centered or uncentered autologistic model. The tests use the asymptotic normality of pseudolikelihood estimates with the Godambe sandwich variance estimator. A test for the spatial correlation eta is also performed.

Usage

significanceTest(model)

Arguments

model

a centered or uncentered autologistic fit

Value

a list of p-values and z-values

Details

For multinomial models, tests are performed for each row of the coefficient matrix, as each row corresponds to a single predictor. The matrix of eta coefficients is tested against the null hypothesis of no spatial correlation.

Examples

# NOT RUN {
#model=MPLE(...)
#significanceTest(model)
# }