I'm working with captive jaguars behavioral data to answer how several independent variables affect the incidence of a certain behavior "E". My dependent variable is the number of times the behavior E was observed. My independent variables are both quantitative (like area of the enclosure, age, and number of zoo visitors) and qualitative (like sex, kind of enclosure substrate, and origin of the animal).

I've been advised to use a generalized model or a mixed model to asses how these variables explain my dependent variable and to then use it to test a few hypothesis, but I have a few questions:

1- Do I have to convert all my independent variables to a single type (quantitative or qualitative)? If so, how? I'm using R by the way.

2- What tests should I perform on my data to fulfill the model's conditions?

3- I've noticed that the model uses different "link functions" like poisson or binomial and I'm really lost on that. Any advice as to which of these would work better?

If it helps my n is 21 individual jaguars. Each of them has one measure for every variable. Also, any advice on how to do all this in R will be greatly appreciated.

Thank you in advance.


1 Answer 1


A mixed model doesn't sound like it would be appropriate, because (so far as I can tell) you measured each jaguar only once on each independent variable (IV), and neither the jaguars nor the IVs are grouped.

  1. No.

  2. I guess that by "conditions" you mean "assumptions". Don't use null-hypothesis significance tests to make judgments about whether modeling assumptions are met. Instead, use what you know about the scales of the variables and the study design that the data came from. Also, plotting the data is always helpful.

  3. It sounds like you're trying to fit a generalized linear model with R's glm. Presumably you want a Poisson link, because your DV is a count.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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