[R] Linear model without an intercept

Gang Chen gangchen at mail.nih.gov
Tue Jul 31 22:05:52 CEST 2007


I would like to run a regression analysis without a constant  
(intercept) or a special one-way within-subject (repeated-measures)  
ANOVA. I'm not sure if the following command lines are correct or not:

m1 <- lme(Resp ~ Cond - 1, random = ~  Cond - 1 | Subj, TestData)

or,

m2 <- lmer(Resp ~ Cond - 1 +( Cond - 1 | Subj), TestData)

Also I notice that both lme and lmer provide the correlation matrix  
of the fixed effects. So if there are more than 2 levels in the fixed  
effect of condition (Cond), how can I correct for sphericity  
violation in lme and lmer if I want to get an F test for condition  
effect?

Thanks,
Gang



More information about the R-help mailing list