Estimating the impact of between- and within-subject designs on consumer valuation experiments: the case of positive vs. negative information

Introduction Valuation experiments are an important technique for measuring preferences for products, product attributes, and the effect of information on consumer valuation. While these experimental valuation techniques are widely used, there is little empirical evidence about an important design decision: the choice of a between-subject vs. within-subject design. Within-subject designs offer clear value in terms of providing multiple observations per participant, which increases statistical power. However, there are long-standing preferences for between-subject designs. This stems, in part, from concerns about the potential for biases or heuristics, such as the anchoring and adjustment heuristic, to decrease the external validity of results generated in within-subject experiments. Methods We test for differences in consumer willingness to pay (WTP) for products that were elicited in between-subject designs versus within-subject designs. We examine the impact of information on product valuation in two experiments. The first examines the introduction of negative information --about the economically motivated mislabeling of extra virgin olive oil, while the second provides positive information: information about the health benefits of zinc-biofortified rice. In both cases, we examine the impact of information on the products linked to the information provided, as well as spillover effects on products that are not discussed in the text. Results Our findings show that consumer product valuation estimates elicited from within-subject designs do not significantly differ from estimates elicited using between-subject designs. This result holds in the case of positive and negative information. Discussion These findings indicate that the impacts of information on product valuation are measured consistently in both between-subjects and within-subjects designs, suggesting that researchers can take advantage of the additional statistical power of within-subject designs to examine response to information. JEL codes C9, C91, D12, Q18.