[R-sig-ME] interactions in fixed effects

[Joshua Wiley]

Tests of the significance of random effects are notoriously
problematic, but for your question on glm(), try:

anova(GLM0, GLM1, test="LRT")

see the docs for anova for more details.

Cheers,

Josh

On Tue, Feb 5, 2013 at 8:58 AM, Gabriela Agostini
<gabrielaagostini18 at gmail.com> wrote:
> Hello
>
> I am working whit GLMM (binomial family) in lme4. In order to test
> differences between amphibian parasitic infections in two study sites, I am
> using a model consisting of two fixed effects ("sa" and "sp" ) and a random
> effects ("sdy"). After testing the significance of the random effect, I
> chose to use GLM  based on the same error structure and variables. My
> problem is: when I try to explore the interaction between two fixed
> effects, anova () does not show the values of Chi and the probability value.
>
> I am sorry for my question, maybe this is a bit basic. But I could not get
> these results.
>
> Thanks!
>
>  names(data)
> [1] "sa"  "sdy"  "sp"  "inf"  "noinf"
>> ### sa(study area)
>> ### ps(species)
>> ### sdy(sample day)
>> ### inf(individual infected)
>> ### noinf(individual non infected)
>
>> class(data$sa)
> [1] "factor"
>> levels(data$sa)
> [1] "cult" "ref"
>> class(data$sp)
> [1] "factor"
>> levels(data$sp)
> [1] "hpa" "lla" "llj" "rfa"
>
>> library(lme4)
>
>> data$Ymat<-cbind(data$inf,data$noinf-data$inf)
>> GLM0<-glm(Ymat~sa+sp+sa*sp,data=data,family=binomial)
>> summary(GLM0)
>
> Call:
> glm(formula = Ymat ~ sa + sp + sa * sp, family = binomial, data = data)
>
> Deviance Residuals:
>     Min       1Q   Median       3Q      Max
> -2.2910  -0.8011   0.0000   0.6543   2.5593
>
> Coefficients:
>             Estimate Std. Error z value Pr(>|z|)
> (Intercept) -2.18122    0.19937 -10.941  < 2e-16 ***
> saref       -1.43460    0.36483  -3.932 8.42e-05 ***
> splla        1.82021    0.33694   5.402 6.58e-08 ***
> spllj        2.09878    0.20676  10.151  < 2e-16 ***
> sprfa        2.10572    0.24216   8.696  < 2e-16 ***
> saref:splla -0.08470    0.57874  -0.146    0.884
> saref:spllj -0.01143    0.37498  -0.030    0.976
> saref:sprfa -0.47644    0.43437  -1.097    0.273
> ---
> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
>
> (Dispersion parameter for binomial family taken to be 1)
>
>     Null deviance: 1013.00  on 363  degrees of freedom
> Residual deviance:  355.21  on 356  degrees of freedom
> AIC: 935.48
>
> Number of Fisher Scoring iterations: 5
>
>> GLM1<-glm(Ymat~sa+sp,data=data,family=binomial)
>> anova(GLM0,GLM1)
> Analysis of Deviance Table
>
> Model 1: Ymat ~ sa + sp
> Model 2: Ymat ~ sa + sp + sa * sp
>   Resid. Df Resid. Dev Df Deviance
> 1       359     358.80
> 2       356     355.21  3   3.5908
>
>
> Thanks!!!
>
>         [[alternative HTML version deleted]]
>
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>



-- 
Joshua Wiley
Ph.D. Student, Health Psychology
Programmer Analyst II, Statistical Consulting Group
University of California, Los Angeles
https://joshuawiley.com/