Experimental Design Details
At the beginning of the experiment, participants are randomly assigned to a role that they keep throughout the experiment: expert, client, producer A or producer B. The experiment consists of 15 rounds. In each round, participants are randomly grouped into 4-person teams, including 1 expert, 1 client, 1 producer A, and 1 producer B. The composition of each group changes in each round.
In each round, the expert must make decisions for the client in their group. The client needs either product A or product B. The expert does not know the client's actual need. However, they know which category their client belongs to and are aware of the probability ( p ) that an individual needs product A, within each category. The client's category and actual need can change in each period.
DESCRIPTION OF A ROUND
The experiment is composed of 15 rounds. Each round is composed of 3 steps.
Step 1
At the beginning of each period, one of the two producers (A or B) receives some tokens. The producer who receives the tokens (A or B) may vary from one period to another, based on a random draw. This producer can then choose to keep the tokens for themselves or to send them to the expert. If the tokens are transmitted, the expert knows which producer sent them. However, if the producer chooses to keep them, the expert does not know which producer could have transmitted them. Neither the client nor the other producer is informed of this decision.
Step 2
The expert is responsible for making an initial decision for the client within their group. The client needs either product A or product B, but the actual need is not known with certainty by the expert. However, the expert knows the category to which their client belongs, numbered from 1 to 9.
Each category provides information about the probable need of the client by offering statistical data: it indicates, among 1,000 clients in that category, how many need product A and how many need product B. For example, if a client belongs to category 3, it means that out of 1,000 clients in this category, 300 needed product A and 700 needed product B.The expert must then choose a product for their client.
Simultaneously, the client is informed of the category to which they belong during this period, but not of their actual need. They also have access to the statistical information and must indicate which product they would choose for themselves based on the available information.
Step 3
The expert can decide to try to better determine the client’s actual need during this period. To do this, they can choose to spend a certain number of tokens. The more tokens spent, the higher the chance of discovering the client’s true need (truth or noise signal). A table is given to the expert, that shows how the chances of uncovering the client’s actual need evolve depending on the number of tokens spent.
A draw is conducted based on the number of tokens spent:
• Either the expert discovers the client’s actual need during this period (product A or B);
• Or they do not obtain any additional information.
After the draw, whether or not the expert has gained extra information about the client’s need, they must once again choose the product to recommend to their client during this period. Neither the client nor the producers are informed of the number of tokens spent, nor of the final product chosen.
Subsequently to the main experiment, at the end of the 15 rounds, a belief survey is administered, tailored to the different roles, furthermore a survey is conducted to gather data about the participants’ social and risk preferences as well as socio-demographics.
TREATMENTS
We will consider 4 between-subject treatments:
1. Gift Treatment (Main):
Producers can send gifts in each round if they are randomly assigned the right to gift the expert. Experts may or may not receive a gift in each round depending on producer behavior. This creates variation in gift exposure across rounds for each expert.
2. Information Treatment (Main):
Producers may send messages about product match (optionally with a gift).
3. Gift Refusal (Control):
Experts have the possibility to refuse the gift sent by the producer.
4. Commission Treatment (Control):
Producers may offer a commission, but only if their product is selected.