now I put all 3 questions in one trial and skip mid-trial to the next if the response of the participant determines that the following questions should not be presented.
So I can randomize via $ and not via BGR.
This should work.
But thanks for thinking.