Strategic Ignorance with Third-Party Information Provision

The present study aims to explore how third parties with the option to inform dictators of the potential consequences of their choices affect the behavior of dictators, particularly those who would otherwise wish to avoid that information. We will use the moral-wiggle-room game of DWK as the fundamental decision because of its simplicity and to maintain comparability with other studies. In particular, we will look at whether the presence of such third parties affects dictators’ information and allocation choices, whether the dictator is willing to reward the behavior of the informer, and whether such potential reward deters bystanders from providing information. The goal is to shed further light on the strategic nature of information avoidance, the conditions under which it may manifest as an image-management strategy, and how social outcomes are determined in environments with richer social interactions.
However, various secondary aspects of the environment may encourage or discourage information avoidance. To best be able to examine our primary questions, we first must identify an environment in which this information avoidance is prevalent at the outset. Thus, before embarking on our primary experimental data collection, we must first conduct a preliminary experiment replicating the original DWK study in our decision environment to establish whether the environment is suitable (Stage 1). If we fail to find sufficiently high levels of information avoidance, we will not continue with the rest of the experiment. While we will have failed at this attempt to answer our main questions, will have learned something about the conditions under which information avoidance does or does not manifest.
However, conditional on finding sufficient levels of information avoidance in the preliminary experiment (see Hypotheses 0a-0c below), we will conduct the main experiment (Stage 2) to answer our main questions, both of which concern the dictator’s behavior. First, how does the presence of a third party affect ignorance rates and allocation choices? If allocation choices are affected, is this caused by the mere presence of the bystander – through giving the dictator a feeling of being watched, for example – or is it caused by the bystander’s actual provision of information?
Second, what behaviors or characteristics are correlated with the dictator’s behavior when it comes to rewarding the third party informer? Is the third party rewarded less when they impose information on a dictator who wanted to remain ignorant? Do dictators who try to avoid information reward less than those that seek it? If the informer provides information, does it matter whether the news is good (aligned interests) or bad (conflicting interests)?
Our third main question concerns the third party’s behavior. If the dictator would really have preferred to stay ignorant, i.e., the information is in fact unwelcome, a bystander choosing to provide information might trigger negative emotions on the dictator’s part. Our third question is thus whether informers’ provision of information is deterred when the dictator has an option to reward the informer’s behavior.

2023-05-11

2024-12-31

See project The experimental set-up described above provides the following data:
• Dictators:
o Information choice yes/no (all but Baseline)
o Initial choice of allocation A/B (all treatments)
o Final choice of alternative (A/B) (Hypothetical Informer, Informer, Informer—Bonus)
o Bonus yes/no (Informer—Punishment)
o Justification for Bonus– open answer (Informer—Punishment)
• Informers:
o Choice of information provision yes/no (Hypothetical Informer—Separate Screens, Informer—Separate Screens, Hypothetical Informer, Informer, Informer—Punishment)
• Questionnaire
o Demographics
o Explanation for choices made
o Party sympathies: Democrat/Republican/Independent
o “How worried are you about global warming?” (Very, rather, slightly, not at all)
o In “August 2020, how worried were you about the Covid-19 pandemic?” (Very, rather, slightly, not at all)
plan

NA

See above

NA

2. Treatments
We will conduct eight treatments, numbered according to the corresponding subsections below.
2.1 Baseline
In the Baseline of their original study, Dana, Weber and Kuang (2007) (DWK) ran a binary dictator game experiment with student participants in which dictators had to choose between two alternatives, A and B. In alternative A, the dictator would earn 6 and the recipient 1; in alternative B, each would earn 5. About three quarters chose the fair alternative B. In DWK’s Hidden information treatment, however, dictators did not initially know the payoffs of the recipient in each alternative: they knew that dictators themselves would get 6 in A and 5 in B, but the payoffs to the recipient was either, with equal probabilities, 1 in A and 5 in B (conflicting), or 5 in A and 1 in B (aligned). This uncertainty, however, could be costlessly resolved by the dictators simply by clicking a button on the screen. About half of dictators did not click the button, thus remaining ignorant about the recipient’s payoffs – and among those who were actually in the state of conflicting interests, all chose the selfish option A. Thus, the option to stay ignorant substantially reduced the resulting generosity.



Conflicting Aligned
Dictator Recipient Dictator Recipient
A 6 1 6 5
B 5 5 5 1

Table 1. Payoffs in DKW’s Hidden Information treatment.

Our Baseline treatment replicates the incentives and main design of DWK’s baseline treatment. All instructions are given on-screen. After an initial explanation (see Figure 1) and a simple comprehension check, dictators make their choice between options A and B.




Figure 1. First instruction screen to Dictators (Person 1) and Recipients (Person 2) in all treatments. In treatments with a third-party Informer (Person 3), the Informer sees a different first screen.
After a quick attention check (entering a given number, time limit of 45 seconds), dictators are faced with the following screen:

Figure 2. Second screen, dictators, Baseline.
Our alternative treatments are Hidden Information; Hypothetical Informer; Informer; Informer—Punishment; Hidden Information—Separate Screens; Hypothetical Informer—Separate Screens; Informer—Separate Screens.
2.2 Hidden Information
The Hidden Information treatment is like the Hidden Information treatment of DWK.

The first screen is identical to that in Baseline. The second screen, identical for Dictators and Recipients, looks like this:

Figure 3. Second screen, all subjects, Hidden Information.

When pressing “continue”, dictators are sent to a decision screen where they are asked to choose either A, B, or revealing information about which table applies:

Figure 4. Dictators’ initial choice screen, Hidden Information.
If the Dictator chooses either A or B, the game is over and the payoffs are realized. However, if the dictator instead chooses `Reveal’, then she is directed to the allocation decision screen with the payoffs shown, as in Figure 5.

Figure 5. Dictators’ allocation choice after revealing payoffs, Hidden Information. The aligned-interests payoffs are shown in this screen shot.


2.3 Hypothetical Informer
For Dictators, the Hypothetical Informer treatment is identical to Hidden Information until and including the screen shown in Figure 4. However, after the screen shown in Figure 4 (and before getting to see the correct table, if choosing to reveal information), dictators get to see an additional screen with the following instructions:
“You chose [option A/option B/to reveal information about which table is being used for your pair].
A third participant, Person 3, has been matched with your pair. Person 3 answers the following question: “In the case that Person 1 did not reveal which table is being used for their pair, would you prefer to overturn their decision, providing this information to Person 1?” The question is a purely hypothetical one, however; Person 3 is not able to provide any information to you. Thus, you will get to know which table is used only if you requested this information yourself. Person 3’s payoff is independent of their choice.
On the next screen, you will be informed about Person 3’s decision. You will also receive information about the correct table if you requested it. You are then asked to make your choice between options A and B.
Note that we will ask you to choose between options A and B even if you already did so on the previous decision screen. Your previous choice will then be filled out by default, and you can confirm it by pressing the “OK” button. If you wish to change your choice at this point, however, you are free to do so. Payoffs to both you and Person 2 will be based on your final choices on the next screen.”
After this, the dictator is informed about the informer’s decision and also, if the dictator has chosen to reveal information, about the correct table. On the same screen, the dictator makes their final allocation decision and the game ends.

2.4 Informer
The Informer treatment is like the Hypothetical Informer treatment except that the Informer’s choice is actually implemented. The text appearing after Dictators’ information-choice screen is now:
“You chose [option A/option B/to reveal information about which table is being used for your pair].
A third participant, Person 3, has been matched with your pair. Person 3 answers the following question: “In the case that Person 1 did not reveal which table is being used for their pair, would you prefer to overturn their decision, providing this information to Person 1?” If Person 3 answers yes, we will show you which table is being used, regardless of whether you chose yourself to reveal this information. If Person 3 answers no, we will show you which table is used only if you requested this information yourself. Person 3’s payoff is independent of their choice.
On the next screen, you will be informed about Person 3’s decision and then make your choice between options A and B.
Note that we will ask you to choose between options A and B even if you already did so on the previous decision screen. Your previous choice will then be filled out by default, and you can confirm it by pressing the “OK” button. If you wish to change your choice at this point, however, you are free to do so. Payoffs to both you and Person 2 will be based on your final choices on the next screen.”
The Informer then makes their choice; Dictators are informed about this choice, get information about the correct table if either party chose to reveal it, and make their final choice between options A and B.

2.5 Informer—Bonus
The Informer—Bonus treatment is similar to Informer, except that the Dictator now has a third choice to make (i. information, ii. allocation, iii. bonus): After getting to know whether the Informer chose to impose information on him or not, the Dictator can now choose to increase the Informer’s payoff by 1 ECU, covered by the experimenters.
The text appearing after the information choice screen is now:
“You chose [option A/option B/to reveal information about which table is being used for your pair].
A third participant, Person 3, has been matched with your pair. Person 3 answers the following question: “In the case that Person 1 did not reveal which table is being used for their pair, would you prefer to overturn their decision, providing this information to Person 1?” If Person 3 answers yes, we will show you which table is being used, regardless of whether you chose yourself to reveal this information. If Person 3 answers no, we will show you which table is used only if you requested this information yourself. Person 3’s payoff is independent of their choice.
On the next screen, you will be informed about Person 3’s decision and then make your choice between options A and B.
Note that we will ask you to choose between options A and B even if you already did so in the previous decision screen. Your previous choice will then be filled out by default, and you can confirm it by pressing the “OK” button. If you wish to change your choice at this point, however, you are free to do so. Payoffs to both you and Person 2 will be based on your final choices on the next screen.
Finally, we will ask you whether you want to increase Person 3’s payoff by 1 ECU. If you use this option, the extra payoff to Person 3 is covered by the experimenters, not by yourself.”
Informers get a similar instruction screen providing them with the same information.
After Dictators make their final A/B choice, they face a new screen, where they are reminded of their own and the Informer’s information choices, and are asked whether they prefer to increase Person 3’s payoff by 1 ECU. It is specified that Person 1’s own payoff is not affected by this choice. When both have made their choices, the Dictator is asked for their reasons to increase Person 3’s payoff or not (open-ended question), before the game ends.

2.6 Hidden Information—Separate Screens
To examine whether the very knowledge that others may impose information cause dictators to seek information themselves, we include the three separate-screens treatments.
The Hidden Information—Separate Screens treatment is like Hidden Information, except that the information and allocation decisions are in separate screens.

Figure 6. Dictators’ information-choice screen, Hidden Information—Separate Screens
Figure 6 shows the information-choice screen. After this screen, the dictator is directed to the allocation decision screen (Figure 7). If the dictator chose information, the payoffs are displayed. Otherwise, the unknown payoffs are represented by a `?’.

Figure 7. Dictators’ allocation-choice screen, Hidden Information—Separate Screens, with payoff state not revealed. If payoff state is instead revealed, the both instances of `?’ are instead replaced with the appropriate number.

2.7 Hypothetical Informer—Separate Screens
The Hypothetical Informer—Separate Screens treatment uses th

Not available

Role (dictator/recipient or informer) randomized by computer.
Treatment randomized by session timing, unknown to participants.

Individuals

No

Baseline treatment – 100* (individuals)
Hidden info – 200* (individuals)
Informer - Split Screen - 100 (pairs)
Hidden Info-Split Screen – 100* (individuals)
Hyp info - Split Screen - 100 (pairs)
Hyp Info - 200 (pairs)
Informer - Keep collecting data until we have at least 100 in each group (partly determined by subjects' choices - see project plan for details).
Informer-Bonus - Keep collecting data until we have at least 100 in each group (partly determined by subjects' choices - see project plan for details).

Single player treatments denoted with asterisk. Apart from the Info-Bonus people this yields 400 single players and 500 pairs so 1400 subjects. Not straightforward to guess how many pairs we need for Info-Bonus.

See Planned Number of Clusters above.

At least 100 clusters per treatment. See Planned Number of Clusters above.

More details available in the analysis plan. The main tests of the hypotheses are described in Table 2 above. Stata code to undertake the tests can be found in the appendix. We will also test the hypotheses using regression analysis allowing us to control for confounding variables collected through the questionnaire. Consistent with Dana et al. (2007), we design experiments to attempt to be able to discern differences of about 0.2. Considering the worst-case scenario of comparing proportions of 0.4 and 0.6, we obtain the following power relationship: To obtain a power of 80 % we should expect to need a sample of 97 subjects per treatment. To simplify, we aim for 100 subjects in each group in each comparison. In some of the treatments, however, we focus on the 50% of subjects with non-aligned interests, requiring twice the sample. Moreover, the treatments with informers also require an informer, doubling the number of participants.

Research plan: Strategic Ignorance with Third-Party Information Provision

[March 2024] Research plan: Strategic Ignorance with Third-Party Information Provision