Supporting social dilemma experiments with App Inventor

Abstract
    In the study of the problem called "social dilemma", various experiments targeting human behavior have been conducted. I created the following app to efficiently implement one of those kinds of experiments. The feature of this app is that it can automatically reflect in real-time the effects of individual actions on others.

Simple examples of social dilemmas and experiments
    According to Toshio Yamagishi (Professor Emeritus, Hokkaido University, Japan) [1], the social dilemma means that everyone can benefit if they cooperate with each other, but everyone suffers if they act only in their own interests. Research (consideration and experiment) about this has been done for a long time. Here we try to reproduce the famous experiment in this book. However, the following is an explanation based on my own understanding.

A) A group consists of four students who are strangers to each other.

B) Do not touch each other during the experiment.

C) Participants fully understand the mechanics of this experiment.

D) 100 yen will be given to each student in each round.

E) Participants can donate d yen arbitrarily.

F) In that case, the donor's money will be (100 - d) yen.

G) Based on the donated d yen, (2d / 3) yen will be given to each of the 3 people other than the donor.

    The experiment was repeated 12 times with the above items (D) to (G) as one set. The part-time job for this experiment was about 1 hour. In fact, the participants took it seriously because they could get that amount of money. In a round, if only one student donates 100 yen and no other students donate, the donating student gets 0 yen and the other student gets 167 yen. That is, only the donated student will lose. If no one donates at all 12 times, each person will get 1,200 yen. On the other hand, if everyone cooperates and donates 100 yen in all rounds, each person can finally get a total of 2,400 yen (this is the maximum amount). This is a very expensive hourly wage!

    It turned out that each person's donation behavior depends on the "interpersonal trustworthiness" that a person has. That is, it was confirmed that one's donation behavior is determined by "whether or not other people can be expected to contribute." Participants in this experiment include people with high confidence and those with low confidence. (This reliability depends on the response scores of the five questionnaires conducted separately.)

    According to the experimental results by Professor Yamagishi, for the 100 yen given in each round, those with high interpersonal reliability donated 55 yen on average, and those with low interpersonal trust donated 33 yen on average. This result seems convincing. In many cases, it seems that there is an awareness that "I don't know how much other people will donate, so I want to keep my donation at this level".

Smartphone app for a social dilemma experiment
    The calculation of the amount of money in the above social dilemma problem is easy, but it is convenient to have an app to proceed with the experiment smoothly. Reference [1] states that a PC was used, but the details are unknown. Here, we have created an app to perform this experiment using smartphones. Finally, I made a smartphone app with a design like Fig.1. This will be installed on the participant's smartphone. Data exchange between users is fully automated as described below. We thought that it would be desirable for each person to use a smartphone in order to proceed with the experiment without touching each other.

 

Fig.1 A support tool for social dilemma experiments

    An example of running this app with 4 students (4 smartphones) is shown below. Fig.2 shows the beginning of the first round, and each student decides how much to donate. Fig.3 is the execution result of the first round, reflecting those donations. 

Fig.2 Beginning of the first round (determining the donation amount)

Fig.3 Results of the first round (reflecting the distribution of donations)

Use of CloudDB (real-time database)
    The feature of this smartphone app is that each person's donation can be immediately reflected in other people's income without explicit communication or administrator intervention, as shown in Fig.3. That was easily accomplished by using CloudDB (real-time DB) built into MIT App Inventor. The main part is only the following figures. Fig.4 stores the donation amount in CloudDB together with the donor ID. When the donation is stored, the block of Fig.5 of the application of each smartphone is automatically called in real-time. Then, the distribution amount is calculated from the donation amount and added. In this way, the program has become very simple thanks to the real-time database.

 

Fig4. Blocks for registering donations to CloudDB

Fig.5 Blocks automatically called when registering donation to CloudDB

Notice
    By default, anyone can use CloudDB immediately with a common id and token. However, you cannot guarantee how long your data will live. Therefore, you can apply for your own id and token. Although the capacity of the free tier is small, it is stable and useful if you use it.

References
[1] Toshio Yamagishi : Social Dilemma ,PHP Laboratory, July 2000. (in Japanese)