In the majority of choice experiments (CEs) the attributes of non-market goods are conveyed to respondents as a table of numeric and/or categorical data. Recent research suggests that respondents may have difficulties evaluating data in this format. In the context of a CE eliciting preferences for changes in coastal land use, this study uses a split-sample experiment to compare standard presentations with virtual reality (VR) visualisations conveying objectively identical information. We find that compared to the standard presentation, preferences elicited in VR treatments are less variable and exhibit a significant reduction in asymmetry between willingness to pay (WTP) for gains and willingness to accept (WTA) for corresponding losses. We conjecture that the greater 'evaluability' of the VR presentation reduces respondent judgement error and moderates reliance on the loss-aversion heuristic.