Abstract:

Survey research has long demonstrated that small changes of the survey instrument with respect to wording, context or order can significantly influence the outcome. In this regard, split samples can contribute to improving the validity and robustness of stated-preference data by observing the influence of small changes in the design on welfare estimates. Using a choice experiment survey conducted in Central Sulawesi, Indonesia, the effects of attribute order on choice cards on parameter estimates and implicit prices were investigated. The results suggest the occurrence of recency effects, i.e., in the choice experiment context the likelihood of an attribute to be chosen is higher if its position is at the bottom of the choice card. There is indication that the magnitude of recency effects depended on the relative importance respondents ascribe to the attributes on the choice cards.

KEYWORDS: ordering effects, attribute order, recency effects, splitsamples, cost attribute, implicit prices, Indonesia