Experimental Design
We conduct an online survey experiment through CAWI (Computer-Assisted Web Interviewing). In this experiment, we will elicit trust toward other individuals with different characteristics through a non-simultaneous trust game (Berg et al., 1995). This means that participants will play the trust game against partners who have already made their decisions.
Partners
Our approach involves a sequential process rather than simultaneous play. We have initially recruited 12 individuals (partners), stratified by gender, migrant status, and disability status. We have gathered their informed consent to participate in the trust game, collected few simple demographic information and explained the trust game rules and setting. We have then asked them to determine their preferred allocation if a hypothetical participant were to send 0, 1, 2, 3, 4, ... 10 euros. Differently from the participants, the partners did not see any pictures but were informed that their opponent is an individual residing in Italy.
Participants
In the subsequent phase, a survey company specializing in data collection, recruits and administers a longer questionnaire and the trust game through CAWI to a representative sample of the Italian population, after having obtained their informed consent.
Each participant is explained the trust game setting. In the initial screen of the trust game, participants will see an AI-generated photo and will be informed that their partner in the trust game shares the same appearance as the individual shown in the picture and lives in Italy.
A randomly selected group of participants will not see any photo and will simply be informed that their trading partner is an individual living in Italy (pure controls).
The participants' decisions in this part of the experiment will serve as our primary outcome variable, while the characteristics of partners presented to participants will be our main dependent variable.
Before and after the trust game, participants complete a short questionnaire, including standard socio-demographic information, a question about their self-reported general level of trust, a question to assess their risk attitudes, and a self-assessed measure of political orientation, and a short set of questions inquiring their gender norms, beliefs about disability, and views on ethnic minorities. This information will be crucial as control variables or margins of heterogeneity in our analysis.
The Trust Game
Both sets of individuals (participants and partners) will be explained the rules of the trust game and informed that their final compensation will be determined at the end of the game based on their strategic decisions. Participants will play as player 1 and partners will play as player 2.
Dictator game
Participants will also be asked to make decisions in a standard dictator game setting.
Determining Payoffs
For the participants, compensation will be determined through a random draw, with an average expected compensation of 10 euros for the trust game. We will first randomly draw 15% of the participants and then randomly draw one of the two games (trust or dictator) to use to determine the compensation. If the trust game is selected, the responses from the partners with the same characteristics as the AI generated photo that appeared on the screen will be used to compute the payoffs for the participants, ensuring transparent and honest interaction without deception. While if the dictator game is selected, the payoffs will be solely computed based on the decision made by participants.
For the partners - We reward them based on their choices and on the choice of one randomly picked participant who played against them. Partners will receive their compensation as vouchers within three weeks from the end of data collection.