Recent research suggests that, in some situations, the use of numeric information in a choice experiment may lack (what psychologists term) the ‘evaluability’ of visual representations of the same data, resulting in individuals relying upon heuristics rather than underlying preferences to formulate choices. This in turn generates preference anomalies such as the gains/loss asymmetry problem where willingness to accept (WTA) values are far higher than corresponding willingness to pay (WTP) measures. A split sample CE is described in which standard numeric approaches for conveying a coastal zone land use change scenario are contrasted with an alternative treatment in which objectively identical information is presented in visual form via virtual reality (VR) visualisations. A third treatment combines both formats. Results show that a conventional numeric CE design yields WTA values which are much higher than those elicited from the VR treatments while valuations elicited from the latter treatments also exhibit significantly lower variability.