I am a biologist in the middle of their PhD journey and I'm constantly learning new things about statistics but there are some things that are difficult for me to understand.

Until now I had dealt with linear mixed models and although it required some time for me to understand how they worked, the models I have worked with were not very complex. But right now, I am dealing with completely new stuff, specifically with a logistic regression with a random factor (GLMM) and I must say that I find these types of models very very difficult to understand because they are indeed complex and I don't have a strong formation on this topic.

I asked a question recently in this webpage about logistic regression, but doubts keep appearing in my mind. I don't mean to sound like I haven't searched in the internet, in the R packages documentation because I really have! I am writing this question to know if I can get a straight answer to resolve my doubts.

So, I work with a colony of wild birds, where some individuals are floaters (they dont have territory) and where some individuals defend breeding resources (territory, partner, nest etc...). I've built a logistic regression model with a link logit and 3 explanatory variables: body condition, degree of spotted plumage and length of ornamental feathers. Our goal is to know if these phenotypic features can predict an individual becoming a floater or not.

logistic <- glmer(status ~ condition + plumage + feather_length+(1|capture_year),
                   data = data, family = "binomial")


Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
 Family: binomial  ( logit )
Formula: status ~ condition + mot_suma + feather_length + (1 | capture_year)
   Data: data2

     AIC      BIC   logLik deviance df.resid 
   239.4    255.2   -114.7    229.4      171 

Scaled residuals: 
    Min      1Q  Median      3Q     Max 
-2.2089 -1.0149  0.5999  0.7930  1.8357 

Random effects:
 Groups       Name        Variance Std.Dev.
 capture_year (Intercept) 0.1558   0.3947  
Number of obs: 176, groups:  capture_year, 6

Fixed effects:
               Estimate Std. Error z value Pr(>|z|)   
(Intercept)     1.89966    2.01298   0.944  0.34532   
condition       0.50889    0.18690   2.723  0.00647 **
mot_suma        0.46022    1.10952   0.415  0.67829   
feather_length -0.04145    0.05569  -0.744  0.45667   
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Correlation of Fixed Effects:
            (Intr) condtn mot_sm
condition    0.179              
mot_suma    -0.081 -0.057       
fethr_lngth -0.991 -0.177  0.064

I checked the residuals with the package DHARMa and everything looks fine. But in this point I don't know how to continue.

I've read that the estimates that the summary gives are not really precise and that I have to perform a LRT test. But I get overwhelmed when looking for information to do it. I used the function drop1(logistic, test="Chisq") but if I must be sincere, I am not really sure about what I am doing. For this reason I am asking for help to know which should be the next step I should follow.

My mentor has never done logistic regression in R so I can say that I am alone on this and that I really need help, please.

Thank you.


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.