Re: split-plot factorial design

Thanks alot for your comments Ray.

You are right about the N factor being a kind of observed dependent
variable, but the researcher's interest was in how this factor
potentially influences the main dependent variable of interest (an
ecological measurement), as a confounder. Unfortunately N wasn't
designed into the study and was only collected 'observation-style' at
the time of measuring the dependent variable. (Another consequence of
that is that it is quite unbalanced as well which will cause them a
problem given their sample size).

Regarding the D factor - if it is crossed with R then would a full
model also include interaction terms such as D*L, D*B, D*B*R, D*L*R
etc?

Thanks again,
Kylie.

.