1
$\begingroup$

I have data of time budget behaviors in form of relative frequencies (Nr specific behavior/nr of total behavior observed). The best would be to perform a multinomial analysis, but it seems too hard to interpret results for me. So I am going with GLMM (which is not simple but I try to!). I decided for a binomial distribution with the rate of foraging (for example, then I will use other behaviors as responses) as response (numerical variable) and other 4 factorial variables as fixed factors, Id as random. My first model is then like this

mod1<-glmer(rate.for~PERIOD+DAY+AGE+ZONA+(1|ID),
            data=timeb,weights=totale.attivit_,family="binomial")

if I calculate overdispersion then I got 2547.6/720 =3.5 which is high, right?

So now..how do I proceed? I saw that I cannot use negative binomial or beta; I also saw to ad a random term for each observation: this point, which I do not fully understand, is not working in the model, with problems of convergence.

This is how my data look like

$ ID.OBS         : int  NA 1947 1943 1945 1954 1959 63 64 1956 1960 ...
 $ FAMILY      : Factor w/ 12 levels "ALTE","ATOLIMPO",..: 1 1 1 1 1 9 9 9 9 
 $ id       : Factor w/ 67 levels "N","b",..: 54 58 51 57 25 17 19 26 
 $ NSCAN          : int  3 3 3 3 3 4 4 4 4 4 ...
 $ DAYCONT          : int  72 72 72 72 72 72 72 72 72 72 ...
 $ PERIOD        : Factor w/ 3 levels "A","B","C": 1 1 1 1 1 1 1 1 1 1 ...
 $ DAY         : Factor w/ 3 levels "D","E","F": 1 1 1 1 1 1 1 1 1 1 ...
 $ SEX          : Factor w/ 2 levels "F","M": 2 2 2 2 2 2 1 1 2 2 ...
 $ AGE            : Factor w/ 3 levels "0","1","2+": 3 3 3 3 2 2 2 2 3 2 ...
 $ STATUS         : Factor w/ 3 levels "DOMINANTE","GIOVANE",..: 3 3 3 1 2 2 2 2 
 $ ZONA           : Factor w/ 2 levels "ALTA","BASSA": 1 1 1 1 1 1 1 1 1 1 ...
 $ rate.for       : num  0.2 0.5 0.5 0.4 0 ...
 $ rate.play      : num  0.6 0.5 0.5 0 0 0 0 0 0 1 ...
 $ rate.gree      : num  0 0 0 0 0 0 0 0 0 0 ...
 $ rate.groo      : num  0 0 0 0 0 ...
 $ rate.fight     : num  0 0 0 0 0 0 0 0 0 0 ...
 $ rate.look      : num  0 0 0 0.6 0 0 0 0 0 0 ...
 $ rate.l.burrow  : num  0.2 0 0 0 1 0 0 0 0 0 ...
 $ rate.move      : num  0 0 0 0 0 0 0 0 0 0 ...
 $ rate.hay       : num  0 0 0 0 0 0 0 0 0 0 ...
 $ rate.scream    : num  0 0 0 0 0 0 0 0 0 0 ...
 $ rate.dig       : num  0 0 0 0 0 0 0 0 0 0 ...
 $ rate.mark      : num  0 0 0 0 0 0 0 0 0 0 

Actually my rate of foraging behavior (nr of foraging event/total behavior) appears like this in the plot plot of rate of foraging

I also was thinking to use only the number of foraging event instead of rate. In this case foraging would be a counts and I could use Poisson distribution; anyway the overdispersion remain high.

Thanks to any help!

$\endgroup$

Your Answer

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

Browse other questions tagged or ask your own question.