On 21 May 2013 01:52, Rebecca Pepper <[log in to unmask]> wrote:
>
> Any help would be greatly appreciated :)
Interactions are kind of hard, and multinomial logistic regression is
kind of hard, so you have hard-squared.
Have you plotted them? You plot the probabiliity of being in each
group, with your first predictor as the x-axis, and groups defined by
high and low scores on the second predictor. (Also, I wouldn't worry
about doing the Johnson-Neyman thing, you just want a description of
what's happening, and the graph will do that).
I (and no one else, AFAIK) can interpret your parameter estimates
unless you give them all. And given that it's multinomial logistic
regression, it's going to take a little while to work it out.
J
|