THE MALLEABILITY OF SOCIAL PREFERENCES

The treatment will consist in a set of 4 weekly “Reading lab” (RL) sessions, lasting one hour each. The RLs will be held in class by a unique experienced trainer, during winter (between January and March) 2026. The RL technique has been chosen after consultation with expert school teachers and trainers, in relation to the formative power of reading. The readings will address themes related to social preferences, such as generosity, altruism, and aversion to inequality. The stories and discussions will not provide explicit moral lessons or prescriptive guidance about how children should behave. Instead, their aim will be to offer stimuli for reflection and to foster interest and curiosity about issues related to generosity and fairness.

2026-01-07

2026-03-20

Donation in Dictator Game
Inequality Aversion
Update 6 February 2026: Idea of Social Justice

Donation (DG): the child has an endowment of 8 tokens and has to decide how many out of those to give to an anonymous other and similar-to-self child, and how many to retain. Donation will be a continuous variable quantifying the amount donated.
Inequality Aversion: among a set of two possible options, a child has to choose a preferred allocation of tokens to two other anonymous and similar-to-self children: equal split of (2-2) versus (3-5). We will construct a binary variable: whether chose the equal split.

Now we provide more details on underlying data collection for obtaining these measures:

Children will complete behavioral tasks in the field as part of the data collection process: in the experimental sessions aimed at measuring children’s social preferences, children will make incentivized choices. Measurements will be undertaken by a small team of experimenters that will conduct the class measurement event in a single day, in order to minimize disruption in school activities. The team will be supervised by a Researcher (PI or co-Author), who will not enter the experimental room where behavioural tasks are performed. Each measurement event will involve one child in turn. Children, in random order, will be called and, if giving oral consent to proceed, will be accompanied to the experimental room (where experimenters A and B will be present), and while being accompanied there they will pass by (and so have the chance to quickly have a look at) the experimental shop (where experimenters C and D will be placed) displaying items they might later be able to buy with the obtained tokens. In the experimental room, the child in Session 1 (baseline measurement) will play

- A Dictator Game (DG), e.g.: the child has an endowment of 8 tokens and has to decide how many out of those to give to an anonymous other and similar-to-self child, and how many to retain, while experimenters present in the room are turned and cannot see.
- An Inequality Aversion choice: among a set of two possible options, a child has to choose a preferred allocation of tokens to two other anonymous and similar-to-self children: equal split of (2-2) versus (3-5).

DG choices made by children will be implemented; the inequality aversion choice will be implemented with a probability of 1/n where n is class size.

Update on 6th February 2026: This update is registered before the Session 2 data collection begins (it will begin on the 9th of February 2026). Session 2 is the first "post-treatment". In Session 2 we are going to repeat the same measures undertaken in Session 1 (as described above and in that same order) and we are going to add, after those two, a third measure, where the child is proposed to take the “Sen’s flute” choice (A.Sen, The Idea of Justice, Cambridge MA: Belknap Press, 2009) as described in the template attached. This third measurement is meant to capture the child view of social justice. This third one is a hypothetical choice and will not be implemented.

Correlation between Parental and Child Social Generosity in DG

We will compute the correlation between a child donated amount and the amount parent reports would donate in validated survey question.

The research will involve about 1100 second grade children (i.e. aged about 7 or 8 years old) enrolled in public schools in the province of Venice. We invited all the (public) “Comprehensive Institutes” (IC) with primary schools in the Province. Institutes have been incentivised to participate: we pay the Institute 15 euros for each child-parent dyad consenting to participate and for whom the parental questionnaire is returned, to be used for educational activities of their choice or other projects they choose.

Irrespective of their IC participation decisions, Schools from all invited ICs will be randomized into a treatment and a control group. Treatment Schools will receive an in-class formative intervention dedicated to children, while control Istitutes will be offered, after data collections is completed (i.e. in the following school year) a different formative intervention on a topic to be later agreed in that school year.
In both treatment and control schools, children’s social preferences will be measured though incentivized choices, once before the intervention, and twice after the intervention, in the 2025/26 school year.
Also, at baseline, both in treatment and control schools, one parent per household will be requested to complete a questionnaire, which will collect background information on household environment, childcare arrangements, children’s and parents’ characteristics and a measure of parental social preferences. Parental responses will be incentivized with the payment of 15 euros to the Institute as mentioned above.

Not available

We will randomize across schools within the same Comprehensive Institute. This choice is motivated by the characteristics of the Italian education system and the rules and practices governing assignment to schools and classes. Each public primary schools is linked to a IC. ICs are unified educational entities that group schools of various levels (preschool, primary, and lower secondary) under a single administration, which implies a single headmaster overseeing the teaching staff across different schools, a unified school council, common school rules, joint “open days”, and so on. The schools within an IC serve a particular geographic area (e.g., a set of neighborhoods). In the Venice Province, each IC typically includes 2 to 6 primary schools. In a primary school, the number of classes for each grade usually ranges from one (increasingly common due to demographic trends) to three (more rarely found). Prospective pupils are assigned to one or more potential ICs based on their address of residence and are entitled to enroll in one of the schools linked to those ICs. Parents can select a specific school within the IC, and, in general, their child will obtain enrollment in the chosen school. If capacity constraints are reached, students may be redirected to another school within the same IC; however, this is uncommon.
Once enrolled, the school assigns students to specific classes. Assignment considers various factors, typically aiming for a balanced distribution of student characteristics (e.g., nationality, abilities observed in preschool reports, etc.) across classes. Occasionally, parents informally request that their child be placed in the same class as a friend, but this kind of request has no formal relevance and is generally not considered in the process.

While randomizing classes within the same school would maximize comparability between treated and control children (who would have chosen the same school, then being assigned to different classes by the school staff), we have chosen to randomize schools within each IC for three main reasons:
1. Minimizing spillover effects: if treated and control children were in the same school, but in different classes, there would likely be interactions. Children likely know each other (possibly coming from the same neighbourhood and preschool) and would spend time together during lunch, breaks, or even in activities organized by the school. Randomizing within a single class would be even more problematic, as children in the same classroom would have to be separated into treatment and control groups, which could generate clear contamination and confounding effects.
2. Expected comparability between schools within the same IC: children attending schools within the same IC generally live in close proximity and are subject to the same school policies. Additionally, through the parental questionnaire, we will be able to control for family background factors that may influence the specific school choice within an IC.
3. Feasibility: in many schools, there is only one class per grade; had we opted for randomizing classes, such schools and classes would end up being excluded (likely generating bias) from the study.

We will use a Stratified Randomization: Strata will be defined by Institutes. Within each stratum, Schools will be assigned a random number v in the interval (0;1) interval. If the number of schools per Institute (t) is even, the first (t/2) schools with largest v will be assigned to the treatment group. If the number of schools is odd, the ((t/2) -1) schools with largest v will be assigned to the treatment group. For budget reasons, the treatment group will plausibly be slightly smaller than the control group.

Stratified Randomization of schools will be conducted by the PI on the University-Laptop, after the preregistration is concluded, and before the fieldwork begins.

(Randomization of the order in which children will be called into the experimental room will be done in the field by teacher: the teacher will extract codes fishing from a textile bag, in the presence of the Research Team Leader supervising the Experimental team on that day. )

Randomization will be performed across schools belonging to the same comprehensive institute

Yes

34 Schools

About 1,100 pupils

11 Comprehensive Institutes: 34 Schools to be randomized (we expect about 17 (or slightly less) treated and 17 (or slightly more, see details on our randomization procedure) controls corresponding to about 1,100 pupils, . Actual numbers will depend also on parental consent to participation.