Experimental Design Details
FULL-SCALE STUDY; CONDUCTED ON AMAZON MTURK; 3,000 SUBJECTS
Our study is based on a survey experiment conducted on Amazon Mechanical Turk. We restrict the sample to US residents with at least 80 percent approval rate.
Participants were directed to a survey website created with Qualtrics. The first screen includes information on the current situation of kidney procurement and transplantation in the United States. The second screen describes three different systems of procurement and allocation of kidneys from live donors:
In System A, unpaid donors can donate a kidney to a specific recipient or to an anonymous candidate on the waiting list. Kidneys donated to anonymous candidates are allocated by an agency (coordinated by the U.S. Department of Health and Human Services) according to priority rules based on patients' medical urgency, blood and tissue match with the donor, time on the waiting list, age and distance to the donor.
In System B, kidney donors can receive $20,000 from a public agency. Kidneys from paid donors are allocated by an agency (coordinated by the U.S. Department of Health and Human Services) according to priority rules based on patients' medical urgency, blood and tissue match with the donor, time on the waiting list, age and distance to the donor. Donors can decide not to be paid, and can also make directed unpaid donations.
In System C, kidney donors can receive $20,000 directly from the kidney recipient (e.g., out of pocket or through privately purchased health insurance). A registry of patients in need of a kidney and of potential donors is kept by an agency (coordinated by the U.S. Department of Health and Human Services); transactions occur directly between the paying recipient and the donor or through a private agency. Donors can decide not to be paid, and can also make directed unpaid donations.
The third phase of the survey gauges the participants’ assessment of the “morality” of each of the three systems. We focus on five key ethical concerns (exploitation of the poor, coercion of choice, fairness to the patients, fairness to the donors, conflict with human dignity, and conflict with one’s values), and ask the respondents to rate each system on a scale from 0 (minimum concern) to 100 (maximum concern). Respondents may choose any integer number in the 0-100 interval on a slider; we also provide, above the slider, verbal expressions to guide the interpretation of different values. We write, for example “No coercion” above the zero mark, “Minor coercion” above the 25 mark, “Moderate coercion” above the 50 mark, “Considerable coercion” above the 75 mark, and “Maximum coercion” above the 100 mark. We also have a sixth question to include any other moral considerations that the respondents may have. The questions asked whether a given system was against the respondent’s values, and the scale is the same as for the other answers.
Fourth, the survey asks the subjects to choose their preferred system. We offer three different choice sets, and for each choice set, the subjects have to indicate the systems that they think should be implemented. The systems among which to choose are the same in the three choice sets; however, in addition to the description of the systems given at the beginning of the survey, we ask the respondents to assume that each given system X would be expected to procure Y kidneys annually (with Y expressed both in absolute value and as a percentage of the annual demand). The value of Y is randomly assigned between systems, participants and choice set. In particular, for the “unpaid donor” system we randomize between the values of 17,000 and 21,500 kidneys (corresponding to 49% and 61% of the estimated 35,000 annual demand for kidneys); for the two systems including payments, we randomize a wider range of values: 17,000, 21,500, 26,000, 30,500 and 35,000 (or 100% of annual demand). Moreover, with 50% chance, we randomly assign the participant to receive, in addition to the hypothetical information about the number of kidneys procured and the corresponding share of annual demand covered by each system, also the corresponding financial savings or benefits for taxpayers resulting from the additional transplants. We include this arm to the experiment to test whether individuals perceived tradeoffs between efficiency and morality differently according to the type of information about efficiency that was provided. For example, the additional information on monetary savings may increase the weight that individual give to efficiency considerations in their choices; conversely, individuals might find information about monetary savings less strong to justify bending their moral beliefs about a transaction, as opposed to focusing exclusively on the number of transplants without putting a dollar figure next to them.
In addition to asking the participants to select a system as the one that should be implemented according to the respondents, we also asked them to express an overall rating of each system (in each choice set) on a 0-100 scale.
We then added an “attention” test, and a question to gauge additional information on the nature of preferences for the different systems. First, we asked the respondents to remember their choices of systems. The non-exclusive options that could be selected were three: “I chose system X in at least one of the choice opportunities”. Second, we offered three, again non-exclusive (although not all true for a subject), sentences. The sentences read “I would have never chosen system X, regardless of how many kidneys it would procure” (we also added a fourth possibility, “None of the above”). In addition to function as a further consistency check, the answer to these last questions will provide information about individuals with “extreme” preferences, i.e. not willing to select a given system even if, in principle, much more efficient than the other two.
The fifth stage of the survey consists, first, of a series of socio-demographic questions, including age, sex, state of residence, education, occupation, marital status, annual income, political preferences, religious beliefs, etc. Next, we include four questions aimed at gauging the respondents’ beliefs regarding the innate vs. cultural nature of morality, and whether they have utilitarian views. We conclude by asking the participants if they would be willing to be recontacted in the future, and we give them the opportunity to leave a comment if they so wish.
The full survey is attached to this trial registration.
PRE-TEST 1
In the full study, we ask the subjects to express their morality ratings on each of the systems before making assumptions about the number of kidneys procured under each system. For this to be a valid exercise, it has to be the case that respondents’ moral valuations (e.g., how much they consider a system to exploit the poor or lead to coerced choices) are not a function of the number of kidneys procured. We performed a pre-test to assess if indeed individual’s repugnance assessments were affected by the efficiency of a system, and the answer is negative. In our pre-test, we asked respondents to rate the five morality features of a system after informing the subjects also of the expected efficiency (number of kidneys procured, corresponding % of annual demand satisfied; and, in a second version, also the expected savings for taxpayers). We conducted this pre-test with 300 respondents on Amazon Mechanical Turk. The results indicate that, within each system, its level of efficiency did not affect morality considerations. This finding validates our choice to ask, in the full study, the subjects to rate the morality features of each system before showing efficiency levels.
PRE-TEST 2
In our full study, we present three possible systems to the subjects. However, there are various other possible systems that we could also have presented. For example, we could have considered in-kind compensation rather than cash compensation. In a second pre-test, we present 300 subjects on Amazon Mechanical Turk with 10 possible systems (listed below), and ask each of them to express their morality ratings on five of these, randomly selected. The results indicate that the morality ratings are roughly the same for systems with donor compensation and organ allocation performed by a government agency, irrespective of whether the compensation was in cash or in kind (e.g., a deposit in a college or retirement fund); the results were seminal for systems with payments through private transactions, again with similar morality ratings irrespective of whether the compensation was in cash or in kind; and the same held true for systems with no payments (purely altruistic system, “hero” medal system, and priority system). Because of these results, we chose to focus, in the full study, on the three systems described above.