R help - Aug 2010 - reading lmer table

Dear all,

I'm quite new in R and especially with linear mixed effects models and 
I'm not completely sure to read the lmer table in the right way.
for example:

 head(march.f)
   fam subjID Cond      Code   reg     total             first        
second         log.total     log.second   cat
3    f     30      an         fDan1   3 1.2304688 0.6679688 0.56250000  
0.20739519 0.44628710   f
7    f     30      an         fDan2   3 0.9414062 0.9414062 0.00000000 
-0.06038051 0.00000000   f
11  f     30      an         fDan3   3 0.4179688 0.4179688 0.00000000 
-0.87234861 0.00000000   f
15  f     30      an         fDan4   3 0.6015625 0.3906250 0.21093750 
-0.50822484 0.19139485   f
19  f     30      an         fDan5   3 1.5625000 1.4726562 0.08984375  
0.44628710 0.08603434   f
23  f     30      an         fDan6   3 0.3632812 0.2968750 0.06640625 
-1.01257795 0.06429435   f


m7 <- lmer(log.second ~ Cond + cat + (1|subjID) + (1|Code), data = march.f)
 >  summary(m7)


Linear mixed model fit by REML
Formula: log.second ~ Cond + cat + (1 | subjID) + (1 | Code)
   Data: march.f
   AIC   BIC logLik deviance REMLdev
 279.6 312.9 -132.8    243.2   265.6

Random effects:
 Groups   Name        Variance  Std.Dev.
 Code     (Intercept) 0.0067216 0.081986
 subjID   (Intercept) 0.0116881 0.108111
 Residual                 0.0677396 0.260268
Number of obs: 860, groups: Code, 143; subjID, 38

Fixed effects:
                  Estimate Std. Error  t value
(Intercept)  0.32217    0.02888  11.154
Condle      -0.09536    0.02915  -3.271
Condme    -0.12683    0.02840  -4.465
catb           -0.05677    0.02361  -2.405

Correlation of Fixed Effects:
                 (Intr) Condle Condme
Condle   -0.493             
Condme -0.498  0.498      
catb        -0.357  0.000  -0.041
 

As we can see in the formula in this model I used 2 predictors as fixed 
factor: 'Cond' e 'cat'.  Cond has 3 levels ('an', 
'le', 'me') while
'cat' has 2 levels ('b', 'f')
My problems start in the 'fixed effects' table:
in my case the intercept is the  condition  with 'Cond = an' and 
'cat=f'; what about the fourth line of the table ('catb')? does
it show
the global effect of the skip from level 'f' to level 'b' of the
predictor 'cat' regardless predictor 'Cond' or the effect of the
skip
from the intercept ('Cond = an' and 'cat=f') to the condition
'Cond =
an' and 'cat=b'? In few word: does this row indicate a global effect
of
the predictor 'cat' or a more specific passage?

Please help me!

Thanks in advance,

Nicola

Hi Nicola,
>> In few word: does this row indicate a global effect of the predictor
>> 'cat' 
>> or a more specific passage?
It indicates a more specific passage.  Use anova(m7) for global/omnibus.
Check this for yourself by fitting the model with different contrasts. The
default "contrasts" in R are treatment contrasts.

##
m7 <- lmer(log.second ~ Cond + cat + (1|subjID) + (1|Code), data = march.f, 
                        contrasts=list(Cond=contr.treatment,
cat=contr.treatment))
m7s <- lmer(log.second ~ Cond + cat + (1|subjID) + (1|Code), data = march.f, 
                        contrasts=list(Cond=contr.sum, cat=contr.sum))
summary(m7)
summary(m7s)
anova(m7)
anova(m7s)

Regards, Mark.
-- 
View this message in context:
http://r.789695.n4.nabble.com/reading-lmer-table-tp2329521p2330809.html
Sent from the R help mailing list archive at Nabble.com.