Can courtship behavior explain mating succes (Generalized Linear Model?)

Mann

New Member
#1
Hi.

I'm trying to analyse some data about courtship behavior and mating succes in spiders.
I have two groups of male spiders, well fed (NF) and low fed (LF). Each male has been used in an experimental trial with a female, where the duration of 3 courtship behaviors of the male has been quantified in seconds and mating succes has been scored as "yes" or "no".

So I want to answer the question: "Can courtship behavior (either one of them or more than one combined) explain mating succes?"

I have been advised to use generalized liniar model to analyse this question, which I have tried to do in R.

Is the analysis as simple as what I have done?
I'm rusty at both statistics and R, so I'm afraid I'm missing something here. Don't i need to check for homogeneity of variance and normal distribution and "stuff like that" before I do the test?

Would you also use a GLM for this kind of analysis?

(I have attached a screenshot of my analysis, a plot of the behaviors in the 2 groups and a sample of the data).


Thank you very much for any advise.

Best regards
H. Mann
 
#2
Hey Mann.

Using a GLM is the right way to go if you are trying to estimate a probability (of Yes/No type).

What did you do for distribution assumptions with regards to the courtship behaviors? (It will not be normally distributed since you can't have negative time)
 

rogojel

TS Contributor
#3
hi,
this looks like a clear case of a binary logistic regression to me, with feed status as an indicator variable. The sample has a bit of a problem because the poor LF guys never once had a mating success - but if you have more samples that would be the way to go IMHO.

regards
rogojel