Mode shift from car to public transport (PT) can be obtained by imposing restrictive policies on car use, or by improving PT. However, the state of knowledge of modal substitution is weak. This paper documents a preference survey of Norwegian urban dwellers who were asked to consider their mode choice for a reference trip in hypothetical scenarios where their current mode becomes less attractive or unavailable, or substitute modes improve. This allows for analysis of aspects that have rarely been studied before, like directional asymmetry, different mode choice impacts of different attributes (travel time, price etc), and the impacts of magnitude (or intensity) of the attribute change. The analysis focuses on diversion factors (δ), i.e. the proportion of the change in demand for one mode which diverts to another mode. We present empirical evidence of δ and show how it differs with different attributes, with size/intensity of attribute change, and, importantly, how it depends on whether the attribute change accrues to their current transport mode (stick) or to an alternative mode (carrot).