Experimental Design
For the experiment, we have selected real products that participants can win. We have selected T-shirts produced by 4 different firms and which have 4 different ethical qualities. We have selected soaps which are produced by 4 different firms and which have 4 different environmental qualities.We have selected boxes of muesli produced by 4 different firms and which have 4 different nutritional qualities.
The experiment has three independent parts. In each part, the participants have the possibility to win either a T-shirt, a soap, or a muesli. The three parts are the same, except for the product itself, and their order is randomized for each session.The timing of each part goes as follows:
1. Participants are randomly matched by pair. For each pair, we randomly select a product of one of the 4 possible qualities. In each pair, the two participants compete – in a 2-minutes real-effort counting task – to win the product. Participants know whether they can win a T-shirt, a soap or a box of muesli. After the 2 minutes, the participants learn whether they won the product or not.
2. Participants are told that the product they have won / have not won can be of 4 different qualities: very good, good, medium or low, with equal probabilities. We explain how we computed these qualities (ethical, environmental or nutritional) for each product.
3. Participants are the told that they may see a label that is informative about the quality of the product they have won / have not won. We explain the labeling strategy: 1-star label for very good or good products, no label for medium or low quality products. We also explain that the 1-star label is sometimes lost, which also result in no label on the screen. We give the probability with which the label can be lost: either 20% or 80%.
4. Once they saw a label or no label, participants have to state what they believe is the quality of the product they have just won / have not won. They get 1.5 euro if they guess the quality right and 0 otherwise.
Trial of the counting task and final questionnaire: Before the 3 parts, we also match participants by pair once and make them try the counting task. The participant who performs best in each pair wins 4 euros. This “trial” allows us to control for participants’ ability in the counting task independently of their desirability for the product at stake. After the 3 parts, the participants complete a questionnaire about their preferences regarding ethics, environment and nutrition. We also ask the participants when they plan to look for the products they won (if any). Finally, participants complete a demographic questionnaire (age, gender, etc).
Treatment variations: Our experiment manipulates 2 dimensions. We vary (1) whether participants either win or do not win each product, and (2) whether labels have a high or low chance to be lost for exogenous reasons, and therefore are not visible on the screen. Labels are lost either with probability 0.2 or 0.8, between participants.