Experimental studies on prosocial, divisive and discriminatory behavior

  • Chen ZHANG

Student thesis: PhD Thesis (Lingnan)


My PhD thesis contains three experiments: two of them studying prosocial and divisive behavior, guided by Tirole’s (2021) models, and the third studying discriminatory behavior. The first study investigates how people's prosociality changes as public spheres expand. Utilizing the advancement of information technologies, states have begun to introduce social scores to rate and publicize some behaviors in social interactions. With social scores, social interactions are thus divided into a public sphere, which is publicly observable, and a private sphere, which is not. Guided by Tirole’s (2021) model, we experimentally investigate whether social scores promote cooperation in a population with repeated, randomly-matched, pair-wise interactions. We find that in the presence of social scores, the subjects contributed more in the public sphere but less in the private sphere. Besides, with an expansion of the public sphere, the subjects reduced contribution intensity in both public and private spheres, after we controlled for reciprocity. Nevertheless, reciprocity induced positive feedback under a large public sphere, and consequently, an expansion in the public sphere led to more contributions.

The second study investigates divisive behavior, for which agents do not reach a consensus on the proper way to act, and to what extent the behaviors deviate from an agent’s authentic self when the behaviors are private/transparent. Privacy in this study means that an agent can freely select to whom to disclose behavior and costlessly hide behavior to others. In contrast, an agent’s behavior under transparency is always publicly observable. Tirole’s (2021) model predicts that the agent under privacy discloses her behavior to those who choose the same action as she does, but not to anybody else. This creates a “safe space”, and the moderates have to take sides. Behavior is thus not authentic. On the contrary, a transparent environment better promotes authenticity. We design an experiment to test his predictions. Subjects were asked to show their attitudes toward some news articles and nonprofit funds. A disinterested dictator game (DDG) follows to elicit the subjects’ image concerns. We find that subjects with privacy acted more frequently than those in the control group without the DDG. The subjects with transparency by contrast showed similar acting frequency as those in the control group. The results largely support Tirole’s predictions.

The third study examines whether hostility toward immigrants’ home countries spreads to immigrants. To provide empirical evidence for this question, we conducted a survey experiment and collected 1212 valid responses. Respondents were asked to read a news vignette about China expanding its nuclear weapons and told the vignette was taken from either mainstream or non-mainstream news outlet. A control group reading no news was also included. In what follows, a list experiment was employed to measure the attitudes toward Chinese immigrants. Our experiment finds the conservatives were encouraged to openly express discrimination against Chinese immigrants, especially those who were told the vignette was taken from a non-mainstream source and reported low trust in mainstream media.
Date of Award18 Jul 2022
Original languageEnglish
Awarding Institution
  • Lingnan University
SupervisorFuhai HONG (Supervisor) & Tianle ZHANG (Co-supervisor)

Cite this