Cooperation and punishment in a nationwide experiment

We investigate whether cooperation rates increase through the implementation of a third-party punishment (TPP) mechanism. Therefore, we plan to run a large-scale artefactual field experiment with a national representative sample of the German population, where participants play an incentivized prisoners dilemma (PD) game.
To do so, we run two experimental treatments. In the first treatment, there is no TPP, while in the second treatment, TPP is possible. We randomly divide all subjects into two groups. The first group is assigned to the first treatment and plays the PD game without TPP, while the second group is assigned to the second treatment and plays the PD game with TPP.

Subjects will conduct the experiment via an online platform.

Subjects will be randomly assigned to one of two experimental groups, and each subject takes decisions in four (Group 1) or three (Group 2) consecutive stages.

We inform them, that they have to answer several control questions, where they can earn additional 10 cents if they answer them correctly.

Experiment — [Question wording:]

Treatment 1 — Stage 0: Information & Control Questions

“You now make a decision in the so-called "sending game".
For this, you receive an amount equivalent to 1 euro and must decide whether to send the amount to your partner or not.
If you decide to "send", the amount will be doubled and your partner will receive 2 euros.
If you choose "don't send", you keep 1 euro.
Your partner also receives 1 euro and must decide whether to send the amount to you or not. If your partner sends the amount, the amount is also doubled and you receive 2 euros. If your partner does not send the amount, he keeps 1 euro for himself.
The decisions of you and your partner are made at the same time. This means that you do not know what decision your partner will make when you make your decision. Your partner also does not know which decision you make when he makes his decision.
[new screen]

Control questions: Remember the game just described. On the next page, you can decide whether you want to "send" the amount to your partner or not. Here you can see once again the consequences of your decision:
If you and your partner decide to "send", you will both receive 2 euros each. If you choose to "send" but your partner chooses to "not send", you will receive 0 euros and your partner will receive 3 euros.
If you and your partner choose "don't send", you both get 1 euro each. If you choose "don't send" but your partner chooses "send", you get 3 euros and your partner gets 0 euros.

Please answer the following control questions. If you answer these questions correctly, you will receive an additional 10 cents.

[Q1-00 …] How much do you and your partner earn in the following situations? Assume ...
[…i1] ... you and your partner decide for "send".
[…i2] ... you decide for "send", your partner decides for "don't send".
[…i3] ... you decide for "do not send", your partner decides for "send
[…i4] ... you and your partner decide for "don't send".
[Answer categories:] Open answer – numbers.


Treatment 1 — Stage 1: Cooperation

[Q1-01] Which decision do you want to make?
[Answer categories:]
(1) SEND
(If your partner chooses "send", you will both receive 2 euros each. If your partner chooses "do not send", you will receive 0 euros and your partner will receive 3 euros).
(2) DO NOT SEND
(If your partner chooses "send", you will receive 3 euros and your partner 0 euros. If your partner chooses "do not send", you will both receive 1 euro each).
[Note: Before decisions are taken, subjects have to answer a set of control questions that require typing in the payoffs for the decisionmaker and the partner in all possible choice constellations. Correct answers will be rewarded with 0,1 euros.]

Treatment 1 — Stage 2: Beliefs about the behaviour of the partner

[Q1-02]: What do you think your partner chooses?
[Answer categories:] 1: I think my partner chooses "send".
2: I think my partner chooses "do not send”
[Q1-03] “How confident are you that your answer is approximately correct?”
[Answer categories:] 7-point scale: (1) “Very uncertain” … (7) “Very certain”

Treatment 1 — Stage 3: Punishment Decision

Now we come to a new decision situation:
You now receive an amount equivalent to 0.5 euros and must decide whether or not to use this amount to take money from another player.
This player is not your partner from the sending game, but another randomly chosen participant in this survey. This player also plays the "sending game" with a different partner, and must choose between "send" and "do not send".
If you choose "do not send", you keep 0.5 Euro, and the player keeps the amount from the sending game.
If you choose "deduct", you do not keep the 0.5 euros, and the player will be deducted the full amount from the sending game.
Important: These rules are known to the player when he makes his decision in the sending game. There is also another person for the player's "partner" in the sending game who can deduct money from the partner.

[Q1-04] Which decision do you want to make? I would like to take the money from the player ...
[Answer categories:]
(1) … in any case (no matter if he chooses "send" or "do not send").
(you will receive 0.5 euros)
(2) ... only deduct the money if he chooses "do not send".
(you get 0,5 Euro if the player chooses "send", 0 Euro if he chooses "do not send")
(3) ... deduct only if he chooses "send".
(you will get 0 Euro if the player chooses "send", 0,5 Euro if he chooses "do not send")
(4) ... deduct in any case (no matter if he chooses "send" or "do not send").
(you will receive 0 euros)

Treatment 1 — Stage 4: Beliefs about the behaviour of the player

[Q1-05]: What do you guess, how does the player decide?
[Answer categories:]
(1) I think the player chooses "send".
(2) I think the player chooses "do not send”

[Q1-06] “How confident are you that your answer is approximately correct?”
[Answer categories:] 7-point scale: (1) “Very uncertain” … (7) “Very certain”


Treatment 2 — Stage 0: Information & Control Questions

Same as in Treatment 1.
After answering the control questions participant in Treatment 2 get additional information on a new screen:

IMPORTANT: You have an "observer" in this game. Your observer is not your partner, but another randomly chosen participant in the survey. Your observer can deduct money from you: He has to decide, depending on your decision in the sending game, whether to give up 0.5 euros in order to deduct the entire amount from the sending game from you.
The observer has four choices:
1) He does not deduct any money from you in any case (no matter if you choose "send" or "do not send").
2) He deducts money only if you choose "do not send".
3) He deducts money from you only if you choose "send”
4) He deducts money from you in any case (no matter if you choose "send" or "do not send").
Your partner also has his own observer who can deduct money from your partner.


Treatment 2 — Stage 1: Cooperation

[Q2-01] Which decision do you want to make?
[Answer categories:]
(1) SEND
(If your partner chooses "send", you will both receive 2 euros each. If your partner chooses "do not send", you will receive 0 euros and your partner will receive 3 euros. In any case, it is possible that the money will be deducted from you or your partner if your observer or your partner's observer chooses to do so).
(2) DO NOT SEND
(If your partner chooses "send", you will receive 3 Euro and your partner 0 Euro. If your partner chooses "do not send", you will both receive 1 euro each. In any case, it is possible that the money will be deducted from you or your partner if your observer or your partner's observer chooses to do so).

[Note: Before decisions are taken, and before subjects see the box informing them about the option to be punished, subjects have to answer a set of control questions that require typing in the payoffs for the decisionmaker and the partner in all possible choice constellations without punishment. Correct answers will be rewarded with 0,1 euros.]


Treatment 2 — Stage 2: Beliefs about the behaviour of the partner

[Q2-02]: What do you think your partner chooses?
[Answer categories:]
(1) I think my partner chooses "send".
(2) I think my partner chooses "do not send”

[Q2-03] “How confident are you that your answer is approximately correct?”
[Answer categories:] 7-point scale: (1) “Very uncertain” … (7) “Very certain”

Treatment 2 — Stage 3: Beliefs about the behaviour of the observer

[Q2-04]: How do you guess your observer decides? I think my observer chooses ...
[Answer categories:]
(1) ... not to deduct in any case (no matter if I choose "send" or "do not send").
(2) ... only to deduct if I choose "do not send”
(3) ... to deduct only if I choose "send”
(4) ... to deduct in any case (no matter if I choose "send" or "do not send")

[Q2-05] “How confident are you that your answer is approximately correct?”
[Answer categories:]
[Answer categories:] 7-point scale: (1) “Very uncertain” … (7) “Very certain”
 
[Treatment Variation]
As explained above, two different groups play two different versions of the game
[Treatment 1] — PD without TPP
[Treatment 2] — PD with TPP

2021-04-09

2021-05-20

Cooperation rate in the PD game, that is, the percentage of participants who decide to send their endowment to their partner.

We will investigate effect heterogeneities by age, gender, and socioeconomic status, as well as whether respondents answered the control questions correctly. Furthermore, we will investigate the correlation between cooperation and punishment decisions and survey-based measures of real-life cooperative and punishment decisions in the context of Covid-19 vaccination decisions.

We plan to sample a minimum of 2,000 adults aged between 18 and 69 years (up to a maximum of 3,000 — conditional on pending supplementary research funding; also see “Experiment characteristics”).
The experiment is conducted in cooperation with a survey institute. The recruitment and polling are managed by the survey institute, which collects the data via an online platform. That is, our participants participate in the experiment autonomously on their own digital devices. Randomization is carried out by the survey institute at the individual level, using a computer. Payment of incentives is carried out anonymously by the survey institute.
Our experiment is structured as follows:
Respondents will be randomly assigned (between subject) to Treatment 1 (p=0.5), or Treatment 2 (p=0.5).

Treatment 1:
Stage 1: Cooperation
Stage 2: Beliefs about the behaviour of the partner
Stage 3: Punishment Decision
Stage 4: Beliefs about the behaviour of the player

Treatment 2:
Stage 1: Cooperation
Stage 2: Beliefs about the behaviour of the partner
Stage 3: Beliefs about the behaviour of the observer

(e.g. public lottery, randomization done in office by a computer, coin flip): Randomization is carried out by the survey company, using a computer.

at the individual level

No

2,000 individuals (adults aged between 18 and 69 years). (maximum: 3,000 – conditional on pending supplementary research funding )

2,000 individuals (1,000 per treatment). (maximum: 3,000 (1,500/1,500) – conditional on pending supplementary research funding ). According to our conservative cost calculations, we have sufficient funding for sampling 2,000 individuals. In case that we are granted supplementary research funding, we envisage to continue sampling up to a maximum number of 3,000 individuals.

1,000 per treatment. (maximum: 1,500 – conditional on pending supplementary research funding).