Expected Monotonicity – A Desirable Property for Evidence Measures?