Trust and Reciprocity on and off the Spectrum – Evidence from school-age children in Vienna

2025-12-17

2026-12-31

For each student and round: Amount send (as sender), expected return value (as sender), return value for different amounts received (as receiver).

Students first respond in the role of sender. They report (i) the discrete amount (0-4 Euro) they send to the receiver and (ii) expected return value from the receiver (in Euro).
Then students respond in the role of receiver. For each potential amount received (they are shown the amount they would have) the students report the amount they are sending back to the sender.
The procedure is repeated a second time, after students have been assigned to the treatment or control group Amounts sent/returned are being used in levels as well as ratios.

Differences in the (i) amount send, (ii) the amount send back, and (iii) response to experimental treatment between neurotypical and neurodivergent students.

Between the two rounds, half of the students receive a treatment emphasizing potential gains from cooperation in the trust game. We are interested in initial differences in amount sent / sent back between neurotypical and neurodivergent students, as well as their differential reaction to the treatment.

The experiment is conducted in person in schools in Vienna, Austria. The experiment will be conducted in German. Students (age range 15-18 years) fill out a paper-based survey containing the description of the trust game, the treatment, and their decisions in the trust game. Part of the students are sampled from special needs schools for students on the neurodivergence spectrum. Another part of students is sampled from other secondary schools (Gymnasium and Handelsakademie) to represent an adequate control group in terms of age, sex and (if necessary) migration background.

Students eligible for participation are all students present, conditional on signed parental consent forms and their own consent to participate. Students can withdraw from the experiment at any time. Students are supervised by experienced teaching assistants throughout the experiment. Some questions on diagnoses (students and parents) as well as age, educational background (grouped) and labor market participation (full-time, part-time, not employed) are answered on a voluntary basis by parents when signing the consent form.

Students without main consent form for participation (we allow non-response to pre-experiment survey questions), or who fail comprehension checks (non-integer response to integer question, amount send / send back higher than endowment) are excluded from the main analysis.

The experiment in the classroom has the following steps:

Step 1:
The game is explained to students in an easy, neutral, and accessible way. The questionnaire contains three pages describing the trust game and the students’ tasks in detail.
The trust game itself has the following steps: (i) The sender starts with an initial endowment of 4 Euro; the receiver has an initial endowment of 0 Euro. (ii) The sender can then send any integer number of Euro (0-4) to the receiver. The sent amount is tripled. (iii) The receiver can then send any desired amount of Euros back to the sender.

Step 2:
Students play the trust game for the first time. Students start in the role of sender and report the amount sent (0-4 Euro, discrete) and the expected return amount (in Euro). Students then continue in the role of receiver and report the Euro amount sent back, conditional on Euro amount received. Both roles are on separate pages of the questionnaire, the questionnaire is printed single-sided.

Step 3:
Before the treatment, the questionnaire includes a page stating “Please wait. Your teacher will tell you when to continue” (in German). All students start receiving the treatment at the same time.
The treatment group (50 %) receives a statement that cooperation is beneficial, followed by example calculations of the total amount available in the second stage (sending back decision for receiver) when 0 Euro are sent and when all 4 Euro are sent. The control group receives the previous game explanation again.
After the treatment, the questionnaire includes a page stating “Please wait. Your teacher will tell you when to continue” (in German). All students start playing the second round of the trust game at the same time.

Step 4:
Students play the trust game for the second time. Students start in the role of sender and report the amount sent (0-4 Euro, discrete) and the expected return amount (in Euro). Students then continue in the role of receiver and report the Euro amount sent back, conditional on Euro amount received. Both roles are on separate pages of the questionnaire, the questionnaire is printed single-sided.

Step 5:
After both rounds of the trust game are played, students fill out additional questions on trust and their number of social interactions in their daily life. These control variables include the number of social interactions (proxied by siblings and participation in a sports team), social preferences regarding fairness, risk preferences, trust in a set of individuals (adolescents, adults, teachers), and schooling.

Payout for each student is determined for one randomly selected round (in either the role of sender or receiver) by randomly matching the participating students in pairs of two players. Earnings are paid out some weeks after the experiment by teachers handing out sealed envelopes containing the participant code.

Not available

Questionnaires are printed and receive an ID. Randomization is done on the student's desk level within class. The ordering (randomization between treatment and control) of questionnaires is determined in office by a computer. Questionnaires are distributed in class in order from the top of the stack.

Randomization is done on the student's desk level. There are (in most cases) two students per denk in schools. All questionnaires are being distributed. If desks are only partially occupied, questionnaires are distributed regardless (and collected after distribution is finished), to not distord the randomization

Yes

approximately 10 schools. Schools for students with special needs have small class sizes. This may increase the number of schools required.
Within each class, we aim to cluster by class (if possible, given the implied cluster size)

approximately 300 students

Intended sample size per arm is 63 students (Treatment vs. Control x Neurotypical vs. Neurodivergent students) for 80% statistical power. For 95% statistical power, the students per arm increase to 105. Sample size per treatment arm will be adjusted following the pilot (see Power caluclation below).

Sample size calculation is based on a comparison of 2 independent samples via t-test. We aim for a statistical power of 80% (z_(1-β)=0.86 for 95%; for 95%: z_(1-β)=1.65 ) and 95%-confidence intervals (z_(1-α/2)=1.96). For a medium sized effect (Cohen’s d = 0.5), we require 63 observations per group (105 for 95%). That is, for our hypotheses concerning the comparisons between neurodivergent and neurotypical individuals, we require 63 (105) observations of each group. For the analysis of the difference in the reaction to the treatment between the two groups, we require 126 (210) individuals of each neurotype. The power calculation is based on best guesses, especially concerning the standard deviation in the sample. We run a pilot in three schools (1 with neurodivergent students and 2 with neurotypical students) December 17-19, 2025 to better calibrate the power analysis, and re-calculate the number of students required to estimate our effect sizes. These updated calculations will be added to the pre-registration as an addendum.