Age and gender effects in neuropsychological tests

Dear friends,

I will try to ask it simple:
Dependant variable: scores of a cognitive test (scale)
independant variables: brain atrophy (yes,no), family history (yes,no), gender (m,f),

I am searching if brain atrophy group performed worse in cognitive tests. However atropy group is older than no-atrophy group and there are more males (so males are older than females).

How can I eliminate gender and age effects and truly understand if atrophy really has a relationship with the test results?
thank you all.


Super Moderator
Multiple regression including brain atrophy, gender, and age as predictors seems the most obvious solution. The coefficient for atrophy will then be the expected difference in score between individuals with and without atrophy, while holding age and gender constant.


Cookie Scientist
In case it helps the OP, note that traditionally this would be called an ANCOVA model (although as CowboyBear notes it is equivalent to multiple regression).

Another approach would be to match subjects on age/gender but it's not clear this would buy you much over the ANCOVA.