[R-sig-ME] The role of the Null Model

[Quentin Schorpp]

Hello,

This is probably the wrong forum for such a question, but at the moment i
don't know who to ask instead.

When analysing data with glmm, at one point it comes to the question of
model selection.
I wrote a paper and got it back with major revisions. For the analysis in
this paper I used Multimodel-Averaging to determine the "best" model as
well as the Average Model of the best model subset within delta 4 AIC
points difference to the "best" model.

Anywhere I read that, if the Null model (fit with intercept only) defined
as, i.e. lmer(response ~ 1 + (1|ranef), data) is incuded within the top
model subset, it means that the data were insufficient for unveiling an
effect of any of the chosen explanatory variables. Hence I wrote it in my
paper. Now the reviewer wants an explanation on the reasons for that and
asks if we made too less sampling efforts in our investigation.

Can someone tell me if I was rigth with my statement, and probably provide
literature for citation?

Thank you very much and kind regards,
Quentin