Hi there,
I know I already posted some questions on this issue, but I still cannot perform this GLM according to expectations.
First, I have a dataset with multiple explanatory variables (e.g. nest temperature, nest measurements, location and species) and one skewed, proportional response variable (nest success).
Because it is a proportional response variable, my GLM + summary look as follows:
Call:
glm(formula = Success ~ Species + Location + `Average temperature` +
`emergence tunnel (cm)`, family = quasibinomial("logit"),
data = dd)
Deviance Residuals:
Min 1Q Median 3Q Max
-1.4768 -0.5145 0.2655 0.6588 0.8621
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -10.592625 20.056906 -0.528 0.600
SpeciesRicordii -0.015988 0.722221 -0.022 0.982
LocationPuente Arriba -0.221543 0.998854 -0.222 0.826
LocationTierra -0.550702 0.823761 -0.669 0.508
`Average temperature` 0.137862 0.223718 0.616 0.541
`emergence tunnel (cm)` -0.004118 0.008694 -0.474 0.638
(Dispersion parameter for quasibinomial family taken to be 0.4711331)
Null deviance: 20.175 on 43 degrees of freedom
Residual deviance: 19.569 on 38 degrees of freedom
(180 observations deleted due to missingness)
AIC: NA
Number of Fisher Scoring iterations: 4
Now I do get an output, but I just threw some possible explanatory variables in of which I don't know if they really contribute to the model (perhaps I need more or less variables).
Because I used a quasibinomial family, I do not get an AIC to see if this model is good. How can I check if my model is good then? And imagine this glm output is right, what conclusions can you make from it?!
Also when I try to check the normality of my residuals by performing...
hist(residuals.glm(model))
...the histogram shows skewed residuals towards 1.0.
In order to do a GLM I learned that the residuals MUST be normally distributed, but now it does not seem like it...
How should I solve this or am I doing something wrong?
I'm a real newbie to R, so I hope someone could help me by using understandable R-language ;).