Comparing apples to apples: Evidence of convergent validity between contingent valuation and discrete choice experiments

To assess the reliability of Stated Preference (SP) techniques several studies have tried to establish convergent validity between different elicitation formats. However, many of these tests fail to adhere to best practices and inadvertently end up comparing apples and oranges. We design a SP experiment to facilitate a clean test of convergent validity and comparison of welfare estimates obtained from a set of Contingent Valuation (CV) questions and a Discrete Choice Experiment (DCE). We design the CV and DCE tasks to be identical and in line with best practice for both. To avoid ordering effects we randomize the order between and within elicitation formats. We use data from a large-scale, national SP survey eliciting the Norwegian population's willingness-to-pay (WTP) to reduce the spread of two invasive alien species (IAS) in the Barents Sea: The Red King Crab (RKC) and the Snow Crab (SC). The results show a convincing argument for convergent validity for both IAS considered. There is a significant difference based on the placement of the CV question. Importantly, we find that the apparent differences in welfare measures from the placement of the CV question in the sequence of choices is fully contained within the distribution of welfare measures derived from a mixed logit model in willingness-to-pay space. This suggests that not only do we have convergent validity between the two elicitation formats, but we show evidence that a properly conducted DCE adhering to best practices provides rich information about preferences.