1
$\begingroup$

I have computed GLMM using glmer in R. My response variable is species richness and my explanatory variable is grazing treatment (with three categories: cattle, sheep and ungrazed). In the model I have included site as a fixed variable and also a new object with the same number of variations as I have to attempt to account for underdispersal (obs):

model2<-glmer(VegRichness~Grazing+(1|Site)+(1|obs),family="poisson",data=veg.rich)

My output is below and the questions I have about it are:

How do I interpret the fixed effects section? Cattle grazing seems to be missing in the oputput, is this because it is somehow incorporated into the intercept?

> summary(model2)

Generalized linear mixed model fit by maximum likelihood (Laplace
  Approximation) [glmerMod]

Family: poisson  ( log ) 

Formula: VegRichness ~ Grazing + (1 | Site) + (1 | obs)
   Data: veg.rich


     AIC      BIC   logLik deviance df.resid 
    178.8    185.2    -84.4    168.8       22 

Scaled residuals: 

Min..........           1Q............           Median....       3Q.........        Max

-1.4936...      -0.5698.....       -0.1928...      0.4923...   1.3646 

Random effects:

 Groups  ... Name......        Variance..... Std.Dev.

 obs.........      (Intercept).. 0.00000....  0.0000 

 Site.........     (Intercept).. 0.03596....  0.1896

Number of obs: 27, groups:  obs, 27; Site, 3

Fixed effects:
                .......Estimate.... Std. Error..... z value... Pr(>|z|)    
(Intercept)............      3.55358.......    0.12309.......  28.869.....  < 2e-16 ***                                                                 
GrazingSheep......     0.01242......    0.07876........   0.158.......  0.87467    
GrazingUngrazed -0.27526.....    0.08503........  -3.237......  0.00121 ** 

---
Signif. codes:  0 x***x 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Correlation of Fixed Effects:
            (Intr)....GrzngS

GrazingShe......................... -0.322       
GrzngUngrzd...................... -0.298...  0.466
$\endgroup$
  • $\begingroup$ @Sven Hohenstein Great thanks . Is there a way to specify the contrast for 'Grazing'? $\endgroup$ – Ashley Jul 8 '15 at 12:59
  • 1
    $\begingroup$ If you are not specifically interested in a particular contrast, you could also do all comparisons for the factor Grazing: library(multcomp) summary(glht(model2, mcp(Grazing="Tukey"))) Unfortunately, I cannot comment yet, otherwise I wouldn't have pasted this as an answer but rather as a comment. $\endgroup$ – Stefan Nov 5 '15 at 16:42
2
$\begingroup$

Since you didn't specify the type of contrast for Grazing, R creates treatment contrasts by default. The first level, Cattle is the reference category. The remaining the levels are compared with the reference category.

  • The fixed effect GrazingSheep is the difference in the dependent variable between Sheep and Cattle.
  • The fixed effect GrazingUngrazed is the difference in the dependent variable between Ungrazed and Cattle.

Finally, the intercept represents the mean value of the dependent variable where Grazing = Cattle.

$\endgroup$
  • $\begingroup$ Great thanks. Is there a way to specify the contrast for Grazing? $\endgroup$ – Ashley Jul 8 '15 at 12:08
  • $\begingroup$ @Ashley Have a look at the help page of ?contrasts. $\endgroup$ – Sven Hohenstein Jul 8 '15 at 13:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.