Generalised Linear models

#1
Do generalised linear models only apply where there is a causal relationship? Or can one apply use them to substantiate the results of a correlation as well?
 
Last edited:
#2
Hi,

I am not sure that I understand your question. I think you can use it to understand correlation between two variables. If the two variables are interval you can use a standard correlation analysis as well. If you arbitrarily assign the one variable to be the dependent variable and the other to be the independent then you can carry out the procedure and see if the coefficient is significant or not. This would be an indecation that the two are correlated.

If one wants to decide whether a categorical and interval variable are 'correlated' you can't just carry out a correlation. I think a GLM would be useful to decide whether there is a significant relationship between two variables of these types.

I don't know a great deal about GLM's but as I understand they are like regression but just more general so that they can be used to carry out ANOVA, regression, logistic regression and other stand-alone statistical techniques.