The impact of accessibility on residential choice - empirical results of a discrete choice model