Sharing Responsibility for the Good

We will recruit participants through Amazon’s Mechanical Turk platform (mturk.com). They will complete an online survey that collects all independent as well as the dependent measures, i.e. participants’ decisions in two experimental conditions (between-subjects design). The full survey as well as a full description of the rationale of this study are attached to this preregistration as a PDF. They will be made public upon completion of data collection.

2018-03-12

2018-03-19

Dependent variable: We will measure participants’ willingness to pay (WTP, in US-$) for a complete measles vaccination of one child (administered through UNICEF) in two experimental conditions: ‘individual decision making’ and ‘bilateral decision making’.

Independent variables: treatment (individual vs. bilateral decisions), belief about partner's WTP, demographics, participants’ responses to the Moral Foundations Questionnaire (MFQ30) and to two trolley scenarios (‘Switch’ and ‘Footbridge’, see, e.g.: Rusch 2015, Rationality, Markets & Morals) as well as several explorative self-report items.

We will recruit participants through Amazon’s Mechanical Turk platform (mturk.com). They will complete an online survey that collects all independent as well as the dependent measures, i.e. participants’ decisions in two experimental conditions (between-subjects design). The full survey as well as a full description of the rationale of this study are attached to this preregistration as a PDF. They will be made public upon completion of data collection.

Step 1: Diligence check. In order to exclude participants who are not following our experimental instructions carefully, the survey begins with a diligence check. Participants will have to copy, letter by letter, a meaningless string of characters backwards and in all uppercase letters. Based on the results of previous studies, we expect about 25% of participants to fail this test. These participants will be excluded from further participation in the study.

Step 2: Demographics. Participants then answer a brief demographics questionnaire.

Step 3: Background information on measles. In order to make sure that all participants are informed about the prevalence and health risks of measles, participants are asked to carefully read a summary of the official fact sheet on measles provided by the WHO. Participants then need to answer three comprehension questions concerning the fact sheet summary. After this, participants are informed about how measles vaccinations are provided by the researchers (through a donation to UNICEF after data collection), about the current price of two doses of measles vaccine (required for full protection of one child), and receipts for the researchers’ previous donations to UNICEF are shown in order to maximize credibility of the instructions.

Step 4: Randomization of participants into treatments. Our experiment has two treatments: Individual decision making (shorthand: ‘IND’) and bilateral decision making (‘BIL’). In IND, one participant actively makes a decision (see below). This participant’s decision is implemented and affects both her/himself as well as a second passive participant who is matched to the active decision maker. In BIL, two participants actively make decisions and the joint result of their decisions is implemented, affecting both of them. All decisions made, i.e. in both IND and BIL, also have a consequence for a third party (each decision either results in a donation of the amount needed for a complete measles vaccination of a child in need to UNICEF by the researchers or not, see below).

Step 5: General instructions. Subjects are informed about the matching mechanism and payment procedures used in this study (post-hoc matching with delayed payment after all decisions by all participants have been collected, donations to UNICEF are also made after completion of data collection). Afterwards, they are required to correctly answer a comprehension question concerning these procedures.

Step 6A / 6B: Treatment instructions. Participants read the instructions for the respective treatment that they were randomly allocated into. Afterwards they have to correctly answer several comprehension questions in order to be able to proceed. In brief, decisions are as follows (see full survey for details).

In IND, half of the participants are randomly allocated to the passive role. These participants are informed about the decision that the active participants will make. They are then asked to make a hypothetical decision in the role of an active participant, but are informed upfront that this decision will have no monetary consequences. The active participants make the following decision. They receive an endowment of $2.50. Then, they are asked to indicate the maximum amount that they are willing to pay (WTP) out of this endowment in order for a child in need to receive full vaccination against measles. We use the Becker–DeGroot–Marschak method (Becker et al. 1964, Behav. Sci.) to elicit this WTP, i.e. if a randomly drawn ‘price’ for the vaccination is smaller than a participant’s WTP, the amount needed for the vaccination is donated to UNICEF by the researchers and the participant receives her/his endowment minus that price. Else, the participant receives her/his full endowment, but no vaccination is donated. Payoff consequences for passive participants are exactly the same as for their respective matched active decision maker.

In BIL, the two randomly matched active decision makers make exactly the same decision as the active participants in IND. However, the donation to UNICEF is only made, if both participants’ WTP are greater than or equal to the randomly drawn price for the vaccination; else, no vaccination is donated.

Step 7A / 7B: Decisions. Active participants make their decisions at this stage. Passive participants make their hypothetical decision at this stage.

Step 8: Expectations. After the decision stage, we elicit all participants’ beliefs about the WTP of their matched partners. I.e. we ask participants to estimate how much their respectively matched other participant is (or ‘would have been’ for matched passive partners) willing to pay for the measles vaccination.

Step 9: Final questionnaire. Participants answer several, non-incentivized, questions (the MFQ30, two trolley scenarios and a couple of explorative questions).

Step 10: Debriefing. Participants are thanked and informed about when they will receive their show-up fee and bonus payment (i.e. payoff resulting from their decisions), and where to find proof of the researchers’ donations to UNICEF.

Planned exclusions and robustness checks:

Apart from excluding participants who failed the diligence check, we plan to run all primary analyses on (a) all data and (b) the subset of participants who stated a WTP > $0.00 and < $2.50. Results from a pilot with N=30 participants (see attached PDF) indicate that irrespective of the interval used to elicit participants’ WTP, a fraction of all participants always states $0.00, i.e. these participants are not willing to pay any positive amount for the vaccination of a child through UNICEF, and another group of participants always states that they are willing to pay any positive amount within the range used to elicit the WTP. Based on the data from our pilot, we expect roughly 20% of participants to choose one of these two extreme ends of the WTP-range.

As implemented in the software used to collect survey data (Qualtrics).

Individual

Yes

We will use the TurkPrime service to recruit participants on MTurk, resulting in groups of 9 participants answering the survey at roughly the same time. However, since we will use within-cluster randomization of participants into treatments, we do not expect this recruitment procedure to have any effect on the results.

360 adults

120 per group (groups: active individual, active bilateral, passive individual)

Using a non-parametric two-tailed Wilcoxon-Mann-Whitney test for a difference in the WTPs between our two treatment groups (group size: 120 each), our design allows us to detect effects of d = 0.37 at a 5% significance level and a power of 0.8.

Relative to individual decision-making, decision-makers in markets have been argued to be more willing to accept negative externalities of their actions. A key mechanism potentially explaining this phenomenon is a diffusion of responsibility between actors in markets. In the preregistered experiment reported here, we test for an effect of responsibility diffusion in a new context: economic transactions with positive externalities. In particular, we test if participants’ willingness to pay for the vaccination of a child against measles differs between individual and bilateral decision-making. We find no such effect. Our study thus adds an(other) instructive null-result to the literature trying to pin down precisely which aspects of market interactions affect precisely which types of morally relevant decisions.

Greiff, Matthias, & Rusch, Hannes (2022). Sharing responsibility for the good. Journal of Behavioral and Experimental Economics, 101, 101953. https://doi.org/10.1016/j.socec.2022.101953

https://doi.org/10.1016/j.socec.2022.101953