2.4.2. Experiment

The experiment was run at LINEEX, the experimental economics lab of the University of Valencia, during the academic year 2015–2016. The programing was made using the software z-Tree [30]. During the MTG experiment, students took resource allocation decisions in a modified version of the original TG by Berg, Dickhaut, and McCabe [22]. In Table 1, we make a list of the main features that characterize the original TG and our MTG.


**Table 1.** Main characteristics of the Berg et al.'s trust game (TG) and our modified trust game (MTG).

> ExCU: Experimental Currency Unit.

In the MTG experiment, subjects were first separated into two groups and randomly assigned a permanent role: type A or sender, type B or receiver. The game was repeated in 10 rounds. In each round, subjects were re-matched randomly, preserving anonymity to prevent any trust-unrelated e ffect, such as reputation, pre-arrangement, or punishment.

At the beginning of each round, subjects were given an ExCU 50 endowment. In each round, the sender decided first how much of that amount to send to an anonymous receiver. In particular, each sender chose the amount x ∈ {0, 10, 20, 30, 40, 50}. The amount sent was then tripled in the receiver's hands, who then decided how much to return to the sender, by choosing the amount y ∈ {0, 10, 20, 30, 40, ... , 200} such that 0 ≤ y ≤ 3x + 50.

Two treatments were performed: a baseline treatment (T0) in which, at the end of each period, the subject received information just about own earnings. Any other information related to gender, education level, religion, wealth, etc., remained hidden. The amount of money the sender decided to send was treated as a measure of how much he trusted the receiver. Moreover, the amount returned by the receiver was interpreted as reciprocity. In this treatment, there were 9 males and 9 females participating.

In the information treatment (T1), we tested whether wealth inequality had an e ffect on trust and reciprocity. In this treatment, at the beginning of each round, the subject was informed about the other's cumulative earnings. Therefore, it may be the case that the subject chose considering the distance in earnings from their partner. In this treatment, the participants were 11 males and 21 females.

After the MTG experiment, students were asked to answer a questionnaire consisting of two blocks of questions. The first block related to sociodemographic characteristics (studies, gender, economic autonomy, and housing). The second one included a questionnaire based on the theory of multiple intelligences [31] concerning interactive, analytic, and introspective dimensions (see Figure 3).

**Figure 3.** Questionnaire based on multiple intelligences.

Students were incentivized with one extra-credit point added to the student module's overall grade. Specifically, the MTG mark was weighted at 15% of the total mark in the RED activity. Since we wanted to observe students' natural willingness to share [32], the reward depended on share decisions contributing to the common good. Camerer and Hogarth [33] were first in suggesting that the TG is not significantly a ffected by changes in the incentive mechanism. Luccasen and Thomas [34] confirmed this fact since they do not find significant di fferences between the rates of trust and reciprocity in TG experiments incentivized with cash or class credit.
