exactLRT {RLRsim} | R Documentation |
This function provides an exact likelihood ratio test based on simulated values from the finite sample distribution for simultaneous testing of the presence of the variance component and some restrictions of the fixed effects in a simple linear mixed model with known correlation structure of the random effect and i.i.d. errors.
exactLRT(m, m0, seed = NA, nsim = 5000, log.grid.hi = 8, log.grid.lo = -10, gridlength = 200, print.p = TRUE, return.sample = FALSE)
m |
The fitted model under the alternative; of class lme , lmer or spm |
m0 |
The fitted model under the null hypothesis; of class lm |
seed |
Specify a seed for set.seed |
nsim |
Number of values to simulate |
log.grid.hi |
Lower value of the grid on the log scale. See exactLRT . |
log.grid.lo |
Lower value of the grid on the log scale. See exactLRT . |
gridlength |
Length of the grid. See exactLRT . |
print.p |
print table with observed variance ratio, observed test statistic -2log LR and p-value? |
return.sample |
return simulated sample? |
The model under the alternative must be a linear mixed model y=X*beta+Z*b+epsilon with a single random effect b with known correlation structure and error terms that are i.i.d. The hypothesis to be tested must be of the form
H0: beta_1=beta0_1,..,beta_q=beta0_q, Var(b)=0
versus
H0: beta_1 neq beta0_1,..or..,beta_q neq beta0_q ot Var(b)>0
The exact finite sample distribution of the likelihood ratio test statistic that is simulated from was derived by Crainiceanu & Ruppert (2004).
If return.sample = FALSE
, the p-value.
Else a list with components
p |
p-value for the observed test statistic |
sample |
the simulated sample of test statistics under the null |
Fabian Scheipl
Crainiceanu, C. and Ruppert, D. (2004) Likelihood ratio tests in linear mixed models with one variance component, Journal of the Royal Statistical Society: Series B,66,165–185.
LRTSim
for the underlying simulation algorithm;
RLRTSim
and exactRLRT
for restricted likelihood based tests
library(nlme); data(Orthodont); ##test for Sex:Age interaction and Subject-Intercept mA<-lme(distance ~ Sex * I(age - 11), random = ~ 1| Subject, data = Orthodont, method = "ML") m0<-lm(distance ~ Sex + I(age - 11), data = Orthodont) summary(mA) summary(m0) exactLRT(m = mA, m0 = m0) library(SemiPar); data(janka); attach(janka); mA <- spm(I(log(hardness)) ~ f(dens, basis = "trunc.poly", degree = 2), spar.method = "ML") m0 <- lm(I(log(hardness))~ dens) ####test for linear trend vs. smooth alternative ## Not run: exactLRT(m = mA, m0 = m0)