A Dictator Game Experiment with Face Mask Wearers and Non-mask Wearers in Japan

We set the following seven treatment conditions in the dictator game experiment:

A) Anonymous: A recipient is anonymous for an allocator.
B) Japanese mask wearers: A recipient wears a mask. The recipient nationality is Japanese.
C) Japanese non-mask wearers: A recipient does not wear a mask. The recipient nationality is Japanese.
D) East Asian mask wearers: A recipient wears a mask. The recipient nationality is non-Japanese and East Asian.
E) East Asian non-mask wearers: A recipient does not wear a mask. The recipient nationality is non-Japanese and East Asian.
F) Non-East Asian mask wearers: A recipient wears a mask. The recipient nationality is non-Japanese and non-East Asian.
G) Non-East Asian non-mask wearers: A recipient does not wear a mask. The recipient nationality is non-Japanese and non-East Asian.

2023-09-08

2023-09-15

Allocation in the dictator game experiment (More specifically, within-difference between the allocated amounts in the second dictator game and the first anonymous dictator game. Please see the details in the pre-analysis plan.)

Attitudes toward the COVID-19-related policies (Agree or disagree for the policy of easing restrictions related to infectious disease control)

We conduct dictator game experiments in the following seven conditions:

A) Anonymous: A recipient is anonymous for an allocator.
B) Japanese mask wearers: A recipient wears a mask. The recipient nationality is Japanese.
C) Japanese non-mask wearers: A recipient does not wear a mask. The recipient nationality is Japanese.
D) East Asian mask wearers: A recipient wears a mask. The recipient nationality is non-Japanese and East Asian.
E) East Asian non-mask wearers: A recipient does not wear a mask. The recipient nationality is non-Japanese and East Asian.
F) Non-East Asian mask wearers: A recipient wears a mask. The recipient nationality is non-Japanese and non-East Asian.
G) Non-East Asian non-mask wearers: A recipient does not wear a mask. The recipient nationality is non-Japanese and non-East Asian.

Each participant in the five dictator game experiments as an allocator. We randomly set the order of the above conditions to create seven groups in each of the mask wearers and non-mask wearers samples. The order of five dictator games as follows:

Group 1. A⇒A⇒A⇒A⇒A
Group 2. A⇒B⇒C⇒F⇒G
Group 3. A⇒C⇒B⇒E⇒D
Group 4. A⇒D⇒E⇒B⇒C
Group 5. A⇒E⇒D⇒C⇒B
Group 6. A⇒F⇒G⇒B⇒C
Group 7. A⇒G⇒F⇒C⇒B

First, we set up six groups: the Japanese condition (B and C), the East Asian condition (D and E) and the Non-East Asian condition (F and G). In all groups, the first among the five games is played in the anonymous condition (A). In the second and third games, for each nationality condition, we randomly set either the in-group condition (B, D and F) or the out-group condition (C, E and G). In Groups 2 and 3, the fourth and fifth games will be played in the East Asian or Non-East Asian condition. In Groups 4, 5, 6 and 7, the fourth and fifth games are played in the Japanese condition. The order of the in-group and out-group conditions is set to correspond to the order of the in-group and out-group conditions in the second and third games. Consequently, each of the mask wearing and non-wearing samples have six groups. Furthermore, we create Group 1 as a control group, where we present the anonymous condition (A) five times, to address potential order effects caused by successive participation in the dictator games. Totally, each of the mask wearing and non-wearing samples have seven groups, including the control group.

We conduct the between-analysis to test the existence of ingroup favoritism, ingroup bias, and outgroup bias in each of face mask wearers and non-mask wearers. We also investigate whether the mask wearers’ ingroup favoritism change between when the nationality of the recipient pair is Japanese and when the nationality of the recipient pair is non-Japanese. Our experimental design allows for both between- and within-individual comparisons. We further conduct the within-analysis to check the robustness of the between-analysis.

We explain more details in the attachment of the pre-analysis plan.

First, we conduct a screening survey through MyVoice.com Ltd., which offers online surveys and experiments, and collect a sample of around 20,000 Japanese people from its registered monitors to match the proportions of a national representative sample in terms of sex (female, male), age (20-69), and residential area (Tokyo metropolitan area, Kansai metropolitan area, and Chukyo metropolitan area). In this survey, we set up questions to ascertain their age, sex, residential area, nationality, and mask-wearing status. We define as “mask wearers” Japanese people who wore a mask when going out during the recent week. We also define as “non-mask wearers” those who did not wear a mask when going out during the recent week.

Second, we sample 1,680 mask wearers and 1,680 non-mask wearers from the respondents of the screening survey and conduct the main survey experiment, including financially incentivized dictator games. In the experiment part, we first present, as priming, the COVID-19-related questions, including whether they have worn a mask. We then present a dictator game five times as in the procedure above. In each game, participants are given an endowment of 100 Japanese yen in addition to the participation fee. They are also asked to decide how much of the 100 yen they give to a paired other person. Participants are informed that the paired person is not participating in this same survey, that the participant solely determine the allocation, that they are the only one who can give a share of the money to the paired person, and that after the experiment, one of the five experimental responses will be randomly selected to carry out that allocation.

Furthermore, before and after the experiment, we set up questions to ascertain psychological and behavioral economic characteristics, health status, the COVID-19-related behavioral characteristics and attitudes, and socio-economic attributes. Concretely, the COVID-19-related questions include attitudes toward the COVID-19-related policies (Agree or disagree for the policy of easing restrictions related to infectious disease control).

Stratified randomization by a survey company. The strata are based on sex, age, and region.

Individuals

No

N/A

3,360 individuals (1,680 mask wearers and 1,680 non-mask wearers)

1,680 mask wearers include seven treatment groups (6*240 mask wearers = 1440 mask wearers) and one control group (240 mask wearers).

Similarly, 1,680 non-mask wearers include seven treatment groups (6*240 non-mask wearers = 1440 non-mask wearers) and one control group (240 non-mask wearers).

In the meta-analysis (Balliet et al., 2014), d of the ingroup-favoritism is 0.32. When we calculate the necessary sample size under the condition of power=0.8 and alpha=0.05, it becomes 155 for each group. To test multiple hypotheses, we finally set a sample size of 240 participants per group, which is the largest size as far as our budget goes. Reference: Balliet, D., Wu, J., & De Dreu, C. K. (2014). Ingroup favoritism in cooperation: a meta-analysis. Psychological bulletin, 140(6), 1556.