They are predicting in car response including cabin gain. It won't match what you see in WinISD which is simulating anechoic response.