I have an experiment in which I have 9 levels of one factor. Ultimately my goal is to test whether there's a quadratic relationship between the hemodynamic response in a given region and the levels of my factor. I'm not entirely sure how to do this...

If I was testing a linear relationship than I believe I could simply use AM regression to model this by 'marrying' each level to an equally spaced number (e.g. 1,2,3,4,5,6,7,8,9). This would create two regressors, one that would model the mean effect of the factor and another that would model linear changes around the mean for each level.

Can AM regression be used to model a quadratic trend? Or should I treat each level as a separate regressor and then use a GLT to test for a quadratic relationship?

thanks,
andrew