Includes a number of common goodness of fit measures to compare different GRT models.
Usage
GOF(grtMod, teststat = "X2", observed = NULL)
Arguments
grtMod
a grt object
teststat
a string indicating which statistic to use in the test.
May be one of the following:
'X2'for a chi-squared test
'G2'for a likelihood-ratio G-test
'AIC'for Akaike information criterion score
'AIC.c'for the AIC with finite sample size correction
'BIC'for Bayesian information criterion score
observed
optional, to provide a matrix of observed frequencies if no fit conducted.
Examples
data(thomas01a)
fit1 <- fit.grt(thomas01a)
fit2 <- fit.grt(thomas01a, PI = 'same_rho')
# Take the model with the lower AIC
GOF(fit1, teststat = 'AIC')
GOF(fit2, teststat = 'AIC')