*3.1. Experimental Results*

Our experimental data from students' decisions in the MTG experiment were classified by roles: trustor and trustee. Trustors' decisions were classified in three levels: Low, Moderate, and High, corresponding to transfers of up to 1/3, up to 2/3, and more than 2/3 of the ExCU 50 initial endowment. Table 2 reports the percentages of these decisions in treatments T0 and T1. Interestingly, at the High level, a di fference of more than 17 percentage points between the treatments was found. This means that students with information about the other's "wealth" were less likely to transfer high amounts compared to students without such information.


**Table 2.** Classification of Trustors' decisions.

Similarly, trustees' decisions were classified into three categories: Selfish, Egalitarian, and Altruistic. A decision was considered Egalitarian when the returned amount is *y*\* = 2*x*, so higher (lower) amounts were classified as Altruistic (Selfish) decisions. The minimum selfish amount is 0, and the maximum altruistic amount is 3*x* + 50. In turn, selfish and altruistic intervals were divided into three equal-length subintervals: low, medium, and high.

According to this classification, a first view of the decisions of trustees is reported in Table 3. An overall information e ffect was found in all categories. When information about the "wealth" gap was available, selfish and altruistic decisions increased by 7 and 2 percent, respectively, whereas egalitarian ones reduced by 8 percent. More specifically, high selfish decisions increased by 9 percent when students had information about their partner's accumulated earnings. Likewise, high altruistic decisions decreased by 1 percent, which was overcome by the increase in medium altruistic ones.


**Table 3.** Classification of Trustees' decisions.

Table 4 reports the descriptive statistics of the trustor's sharing decision and the trustee's transferring decision. Parameter λ represents the percentage of the endowment ExCU 50 sent by the trustor (λ = *x*/50). Similarly, we denote by *r* the rate of return paid back by the trustee (*r* = (*y* − *x*)/*x*). As already mentioned, in treatment T0 subjects had the same initial endowment and did not have information about the other's accumulated earnings at all. This situation was labeled as "Equality". In treatment T1, subjects were additionally given information about the other's accumulated earnings. Thus, subjects know *the gap in the accumulated earnings* before deciding. We defined that *gap* as the trustor's accumulated earnings minus the trustee's. Therefore, we differentiated among positive inequality (*gap* > 0), negative inequality (*gap* < 0), and equality (*gap* = 0).


**Table 4.** Sharing and Transferring decisions.

Comparing both treatments, the difference between the median λ in T0 and T1 was not statistically significant (Mann–Whitney or MW test, *p*: 0.2529). Therefore, in general, the trustor was not observed to modify his sharing decision when knowing the trustee's accumulated earnings in the game. Even when the trustor knew that the trustee had a higher accumulated gain, he kept the median sharing amount (MW test, *p*: 0.2571). Specifically, the probability of sharing a lower λ was 0.544. Therefore, at first glance, our data do not allow us to confirm H1.

We also investigated any gender effects through H3 and H4. Figure 4 presents the box and whiskers plots of students' decisions grouped by gender and treatment. In line with H3, we first observe that female trustors' decisions were significantly different from those of male trustors in both treatments. Compared to T1, in T0 females transferred a higher median amount than males. One interpretation of this result is that males and females react in a different way when they are provided with information on their partner's earnings.

In the trustee role, females' decisions were similar to those of males in T0 (*p*-value > 0.05 for the median test), but not in T1 (*p*-value < 0.01 for the median test) since females paid back a higher median amount than males. Therefore, at this stage of our analysis, the data do not support our H4 regarding a gender effect in trustees' decisions.

**Figure 4.** Distribution of the amounts sent/returned, by gender and treatment.
