Response to the following:
Using the mixed command with an unstructured covariance matrix, fit 3 separate linear models to model the change in the measured distance with the age of the subject. At this stage do not include gender in your model. The first model should treat age as a categorical variable, the second should model age using a linear trend and the third should model age using a quadratic trend. Which of these models is most appropriate? Why?
(Hint: Although you should be using the mixed command, none of your models should use random effects. You should ‘center' the age variable before fitting the third model. AIC, appropriate hypothesis tests and/or residual plots may be useful to justify your answer.)