How to control for depression when sample was measured using two different tests? I am looking at pre and post test depression scores in a 2x3 design with treatment and genotype as the IVs and depression score as the DV.  However, I have to use one depression test for younger people and a different depression test for older people, because I would like to covary out age.  I think I need a repeated measures ANCOVA but am not sure how to compare the depression data from 2 different tests.  Any suggestions?
 A: A simple approach in this case would be to convert depression scores to a common scale (e.g., convert to z-scores within groups).
In general, this should work well if the groups are a random sample. Otherwise, such a standardisation process will remove group differences.
In your case, you seem to both want to make the test comparable across groups and control for age, which as I understand it is conceptualised categorically and is synonymous with the choice of depression test that you have used with your groups.
Thus, by converting depression test scores to within age-group z-scores, you will both standardise the test and covary for age. This seems to be what you want to do.
Admittedly, because you've used different tests, it wont be clear (at least not without more information) the degree to which score differences between older and younger participants are due to age related differences in depression of differences in the scaling of the two tests. However, perhaps if age is somewhat irrelevant to you, then this might not be a huge concern.
More broadly, you would want to present an argument as to the comparability of the two tests. e.g., present previous research showing the correlation between the two depression measures.
Also, if you do have additional information from test manuals and norm tables, then you may be able to standardise the depression measures to a common scale using norm tables. This would permit the approximate assessment of age related differences in depression.
