[R-sig-ME] declaring the variables

[Andrew Dolman]

lmer or glmm, the distinction (from a user point of view) is like lm
vs glm. If you are using a Gaussian model then either will work. If
you want to use a non-gaussian model then you will need to use glmm.



andydolman at gmail.com



On 17 November 2010 19:50, Michael Larkin <mlarkin at rsmas.miami.edu> wrote:
> Awesome!  All three of you have been extremely helpful.  Thanks!
>
>
>
> I have a couple of follow-up questions.
>
>
>
> When I log transformed my catch data to make it normally (gaussian)
> distributed.  Therefore, I do have normally distributed random error terms.
> I will log transform the catch before I apply the model.
>
>
>
> I will do the declaring steps suggested by Petar where I declare the
> variables (Angler = factor, Season=factor, and tide phase = numeric).
>
>
>
> So by using (1 | Angler) in the model for my angler variable (i.e. Joe
> Smith, John Doe, ect) it makes this variable have a random intercept?
>
>
>
> My last question, from reading your emails I should pursue the glmm instead
> of the lmer function because of the random nature of my angler variable?
>
>
>
> Mike
>
>
>
>
>
>