On 10/16/12 8:27 PM, MAHINDA YOGARAJAH wrote:

Dear Doug,

Many thanks for reply. Given the age-group interaction, and the fact that even when age is demeaned it simply means a difference in thickness will be tested for at age=mean age (as opposed to age=0) can I at least interpret the findings when using DODS model with age as a nuisance factor by saying that the difference in thickness between groups represents the difference between groups at the mean age across both groups.

Yes.

Beyond that limited interpretation, is there any other way to analyse the data ? Is there a another model I could use ? My age-group interactions are in a different location to the group differences - does this make any interpretation of group differences more robust ?

You can reanalyze using DOSS and look for group differences. If those differences are in a different location than the interaction, then you are good to go. doug

Thanks.

Mahinda