[R] mixed effect model (multilevel)

[franco salerno]

Hi, I have a problem with mixed effect model (multilevel).

My model that is written in following formulation using the lme4
package and Zelig package:

mylogit<- lmer(OVERFLOW ~ ALTEZZA + INTENSITA + ( 1 | CODICE), family=
binomial(link = "probit"),data = dati)

OVERFLOW can be 0 or 1 and represents the activation of a combined
sewer overflow

ALTEZZA is integer and reppresents the total precipitation occured
during an event

INTENSITA is integer and reppresents the maximun precipitation
intensity occured during an event

CODICE are different (13) combined sewer overflows

I am analysing 480 OVERFLOWs.

Summary

Generalized linear mixed model fit by the Laplace approximation
Formula: OVERFLOW ~ ALTEZZA + INTENSITA + (1 | CODICE)
   Data: dati
   AIC   BIC logLik deviance
 509.2 525.9 -250.6    501.2
Random effects:
 Groups Name        Variance Std.Dev.
 CODICE (Intercept) 0.73159  0.85533
Number of obs: 480, groups: CODICE, 13

Fixed effects:
             Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.818969   0.264502  -3.096  0.00196 **
ALTEZZA      0.023416   0.004184   5.596 2.19e-08 ***
INTENSITA    0.069759   0.021693   3.216  0.00130 **
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Correlation of Fixed Effects:
          (Intr) ALTEZZ
ALTEZZA   -0.155
INTENSITA -0.201 -0.399


I would like to have:

1) an indicator of performarce of my model (is it good the fitness of my model??

2) a sort of cross-validation

3) information on relative importance of predictors (marginal effects???)


Please, help me!!!!!

Thanks