We need demographic parity with the prior study. We may need to coordinate with James around his metrics too.