The performance of quasi-experimental methods applied to changes in non-market goods depends on the ability of reduced form models to accurately measure willingness to pay. When exogenous changes are non-marginal, the accuracy of the reduced form approximations is not well understood. Further complicating the performance of reduced form models is that the true representation of the non-market good in household utility functions may differ from the perceptions of that good as captured in the reduced form model. This paper evaluates a series of before/after quasi-experiments where the true model is known and examines the performance of these methods under a variety of conditions. We find that performance is impacted by the scale of the change and that differences in perceptions of the amenity between the reduced form model and the underlying utility function play an important role in determining the performance of quasi-experimental applications. For researchers interested in non-market goods where the true representations of changes in relation to the underlying utility function are unknown, the notion of perceived measures of the non-market good in reduced form models should receive considerable attention.