Fair reckoning: a qualitative investigation of responses to an economic health resource allocation survey

Abstract

Objective  To investigate how participants in an economic resource allocation survey construct notions of fairness.

Design  Qualitative interview study guided by interpretive grounded theory methods.

Setting and participants  Qualitative interviews were conducted with volunteer university‐ (n = 39) and community‐based (n = 7) economic survey participants.

Intervention or main variables studied  We explored how participants constructed meanings to guide or explain fair survey choices, focusing on rationales, imagery and additional desired information not provided in the survey scenarios.

Main outcome measures  Data were transcribed and coded into qualitative categories. Analysis iterated with data collection iterated through three waves of interviews.

Results  Participants compared the survey dilemmas to domains outside the health system. Most compared them with other micro‐level, inter‐personal sharing tasks. Participants raised several fairness‐relevant factors beyond need or capacity to benefit. These included age, weight, poverty, access to other options and personal responsibility for illness; illness duration, curability or seriousness; life expectancy; possibilities for sharing; awareness of other’s needs; and ability to explain allocations to those affected. They also articulated a fairness principle little considered by equity theories: that everybody must get something and nobody should get nothing.

Discussion and conclusions  Lay criteria for judging fairness are myriad. Simple scenarios may be used to investigate lay commitments to abstract principles. Although principles are the focus of analysis and inference, participants may solve simplified dilemmas by imputing extraneous features to the problem or applying unanticipated principles. These possibilities should be taken into account in the design of resource allocation surveys eliciting the views of the public.

I just stuck with 50‐50, right down the middle. It’s two people.

(Participant B)

I mean, we’re not really dealing with just numbers. We’re also dealing with people. With humans.

(Participant P)

Introduction

Equity broadly interests health policy makers and researchers, yet there remains little consensus on objectives and means for pursuing health system fairness. Different allocation options (e.g. by disease severity vs. expected outcomes) rest on deeply differing ethical principles and have materially diverse consequences. Researchers use survey and experimental methods to identify legitimate and feasible equity models for public policy. ¹ , ² Surveys identify fairness beliefs held among laypersons, communities or the public. ³ , ⁴ , ⁵ , ⁶ Survey methods appeal because of their resonance with the values of public consultation and evidence‐informed policy making. Although survey methods cannot provide all information needed for policy – the popularity of a belief does not necessarily warrant the belief as ethical by normative criteria ⁷ : ‘community values themselves can be unjust’ ⁸ (p. 277) – they can provide much valuable insight. However, their usefulness as a method for obtaining views of the public on issues such as resource allocation depends on their careful design. At present, we have limited understanding of how community members interpret and respond to such surveys.

Surveys also present interpretive challenges for research users, who must generalize from how survey respondents approach hypothetical problems to how a community should approach real policy problems with real stakes. ⁹ Respondents must make analogies between survey problems and real policy problems; the resources, the recipients, allocation rules and constraints posed by survey problems ideally should correspond well to those in the policy realm. These analogies must be made both by the researchers contriving the survey scenarios and by the participants responding to them. For example, previous economic surveys aimed at revealing participants’ equity beliefs have used resources such as avocados and grapefruits, ¹⁰ or fruit plants and fruit for people on a desert island. ¹¹ Policy users of such research would need to consider if avocados are enough ‘like’ health care to make the expressed values relevant.

Ubel ⁹ notes that all quantitative methods for surveying community values suffer from limitations, so users should seek confirmatory patterns of values across multiple studies employing different methods. Some recommend qualitative investigation to aid interpretation. For example, when their survey findings contradicted expectations, Roberts et al. ⁵ suggest that qualitative methods be used to understand how respondents interpret resource allocation scenarios and the meanings of survey responses. Despite the increasing popularity of mixed method designs in health research, ¹² qualitative methods are little used in economic studies. Coast ¹³ pioneered the use of qualitative methods to understand agency preferences for health‐care rationing. Ryan et al. ¹⁴ used qualitative ‘think aloud’ and debriefing interviews to investigate how participants interpreted and responded to a discrete choice experiment. San Miguel et al. ¹⁵ qualitatively analysed economically irrational responses to a discrete choice survey concerning preferences for primary care nursing. The two latter studies found that survey participants sometimes embellish survey questions with new, choice‐relevant details drawn from imagination or personal experience; such details can even uncover beliefs not anticipated by the researchers. ¹⁴ There is sometimes incoherence between participants’ choices and their articulated reasons for choosing them. On qualitative inspection, apparently economically rational responses can in fact be based on economically irrational ideas, ¹⁴ and vice versa. ¹⁵

We report here a qualitative investigation into how participants interpreted the scenarios and judged options presented by a quantitative resource allocation survey (hereafter, the ‘RA Survey’) that measured the effects of information types, goods and community sample on peoples’ beliefs about fair resource allocation. ¹⁶ We were interested in how participants interpret information given on the survey and make analogies between key elements of the scenarios and their own experiences and ideas to make fairness dilemmas tractable and meaningful. We also explored how participants de‐simplify survey scenarios, tacitly adding context or meaning that researchers had had hoped to hold constant or irrelevant. This study offers one of few forays into the qualitative and interpretive understanding of economic problems, demonstrates the value of investigating why economic survey participants make the choices they do, and offers insights useful for the design of quantitative resource allocation surveys.

Methods

This qualitative study was conducted in the context of the RA Survey, a large economic study of lay judgments regarding the fairness of five equity principles for the distribution of health goods: egalitarianism (the distribution of equal shares), utilitarianism (the maximization of population benefit), allocating to achieve equal benefits, allocating according to need and maximin (giving priority to improving the status of those worst off). The stated preference survey was designed using experimental methods that allowed the investigators to identify the impact of good type (health care, non‐health care but health enhancing) and description type (quantitative, verbal, both) on participants’ equity judgments. The methodology and findings are reported in detail elsewhere. ¹⁶ The RA Survey vignettes, questions and answer choices are illustrated in Table 1 (equal benefit, utilitarian, need‐based, egalitarian and maximin principles are represented by options a, b, c, d and e, respectively).

Table 1.

Illustrative RA Survey scenario and questions

A limited supply of pain relief medication is to be distributed between Williams and Taylor, both of whom suffer from a painful disease. This pain medication can provide total relief from the pain. The supply of pain relief pills to be divided between Williams and Taylor is 18 pills per 24 h. The following information is given and is known also to both Williams and Taylor

• Doctors have determined that Williams’ metabolism is such that one pill gives him 2 h of pain relief.

• Doctors have also determined that Taylor’s metabolism is such that one pill gives him 1 h of pain relief.

• Williams and Taylor are identical in all respects except their metabolism.

• After the pills are divided between them, Williams and Taylor cannot trade the pills between themselves or transfer the pills to a third person.

The pills are to be divided between Williams and Taylor. As a third party, you are asked to decide how to divide the pills between Williams and Taylor. Below, we list some possible ways to divide the pills

(a) Williams and Taylor each receive the same amount of pain relief.

• Williams: six pills (12 h of pain relief per day)

• Taylor: 12 pills (12 h of pain relief per day)

(b) Total amount of pain relief obtained by both Williams and Taylor together is as large as possible.

• Williams: 12 pills (24 h of pain relief per day)

• Taylor: six pills (6 h of pain relief per day)

(c) Williams’ and Taylor’s needs for pain relief are met equally.

• Williams: six pills (12 h of pain relief per day)

• Taylor: 12 pills (12 h of pain relief per day)

(d) Williams and Taylor each receive the same number of pills.

• Williams: nine pills (18 h of pain relief per day)

• Taylor: nine pills (9 h of pain relief per day)

(e) Williams whose body is less able to obtain pain relief from the pain medication receives at least as much pain relief as Taylor.

• Williams: six pills (12 h of pain relief per day)

• Taylor: 12 pills (12 h of pain relief per day)

What do you judge to be the fairest division of the pills? Please choose one of the above allocations: (a) (b) (c) (d) (e)

What do you judge to be the least fair division of the pills? Please choose one of the above allocations: (a) (b) (c) (d) (e)

Note: This is an example of a ‘full information’ scenario with both verbal and quantitative descriptions. Scenarios with the quantitative descriptions only list only the bulleted information in (a) through (e) on the actual division of the pills. Scenarios with the verbal descriptions only omit the bullets and present only the verbal descriptions in (a) through (e).

The RA Survey study was conducted 2006–07 and sampled from university‐based population (N = 307) and a Hamilton, Ontario community population (N = 253) recruited from local shopping malls and public libraries. The study protocol was approved by the McMaster University Research Ethics Board and followed standard procedures for voluntary, informed consent from participants. Participants received a modest fee in compensation for participating in the study.

The objective of the qualitative interview component was to explore individuals’ rationales for choosing allocations as fair (or not) and why they believe them to be fair (or not). We also wanted to understand how people interpret stylized resource allocation dilemmas and how they relate such hypothetical scenarios to their own ideas about fairness, and to their own ‘real‐world’ images and experiences.

Qualitative interviews were conducted in two waves with volunteers from the university‐based sample (n = 39; 13% of RA Survey participants) and a single wave with the community participants (n = 7; 3%). As participants completed the computer‐based RA Survey, they were invited to participate in an interview when an interviewer (DD, MG) was available. This pattern of selection was not associated with individuals’ responses to the RA Survey (confidentiality prohibited our access to their survey responses). We conducted semi‐structured debriefing interviews with 46 volunteers in three waves. The first two waves were with university subjects and occurred over six pilot RA Survey administration sessions (n = 19) and over 10 final‐version RA Survey sessions (n = 20). The third wave was with community subjects (n = 7) over one final‐version RA Survey session. A typical version of the interview guide appears in Table 2. Additional probes, and variations on these questions, were used to explore answers and emerging themes. Interviews lasted approximately 15–30 min and were audio‐taped and transcribed verbatim. Transcripts were coded, and the coding scheme was developed through open and selective coding stages ¹⁷ by two investigators (MG, DD). Coding categories were developed between each wave of interviews and used to guide further data collection. We pursued saturation of emerging categories through revised interview foci and by comparing categories between interview waves, between participants and between groups based on self‐reported survey choices. Data and coding were managed with word‐processing software. Analysis was guided by principles of interpretive grounded theory, ¹⁸ focusing on ‘how– and sometimes why– participants construct meanings and actions in specific situations’ ¹⁹ (p. 130). The analysis pursued two aims: (i) to characterize the range of fairness constructs and rationales that respondents used to make sense of the problems presented in the RA Survey and to select their answers and (ii) to explore how participants experienced the survey context itself and understand how these experiences might inform the interpretation of RA Survey responses.

Table 2.

Interview schedule for the qualitative interviews

General questions:
How did you feel doing this exercise?
Is there anything you wish you could have written or added to your answers?
Did any of the problems remind you of any real‐life problems you have heard of, or have encountered yourself?
Review two vignettes, one at a time, asking the following question of each:
Did you find this vignette or the questions difficult to understand in any way?
What is the essence of the problem, to you?
Do you remember which answer you chose?
Why is it the best choice, in your opinion? Why?
How sure do you feel about your choice? Do you have any doubts about it?
What might have been your second choice? Why?
Did you wish you had any additional information about the problem? What?
What would have been the worst, or most unfair choice, to you? Why?
Do you think the other participants in the study probably made the same choice as you did, or not? Why?
General questions:
Did you think differently about those two vignettes? What made a difference for you?
Any further comments about the exercise or the study? Any questions for me?

Note: This is a general guide of the topics covered; interviews were open‐ended.

Results

Findings address several aspects of fair decision making: (i) how participants make analogies between the survey’s allocation dilemmas and their personal resource allocation experiences, (ii) how participants characterize goods in ways relevant to their fair distribution, (iii) desire for, or imputation of, additional information and criteria not provided by the survey scenarios, (iv) the articulation of a fairness theory not anticipated by the RA survey researchers and (v) approaches to making survey choices unrelated to fairness beliefs.

Analogies to own experiences of resource allocation

When asked whether the RA Survey vignettes and dilemmas reminded them of real‐life experiences, interviewees seldom offered health‐related examples. Only one interviewee recalled personal experience with health care, concerning a family member on life support. A community interviewee compared the trade‐offs to ‘everyday’ tasks such as splitting a pack of cigarettes or cutting a cake. Nine interviewees offered the image of adults dividing goodies among children (putting themselves in either the adult’s or child’s role), for example:

‘There’s a cake, and how much do you want to give everybody [laughing] – I don’t know… [laughing] giving candies out to kids or something. But, um, like, should the older kids get more? I don’t know, should they get the same amount?’

(Participant T)

Many ‘real‐life’ examples offered by the interviewees were not from personal experiences, but from stories (e.g. from television shows or coursework) that involved rationing under scarcity, for example:

‘I don’t know, people suffering from diseases in, like, Africa and not being able to, like, distribute, like, medication properly and give them the pain relief they need, I guess. That kind of stuff’

(Participant N).

Interviewees infrequently made comparisons to policy. One respondent compared the dilemmas to questions of entitlement to welfare and child support benefits. One recognized general relevance to the Canadian health system is as follows:

‘I remember one question was about the pain reliever, and it reminds me of the, the medical service in Canada or something… How you distribute the time, doctors, nurses and the resources. The treating patients who need most, who need less, like that’

(Participant M).

Characterizing the goods to be allocated

We were interested in how qualitative differences between the different images of distributing apples for nutrition vs. distributing pills for pain might affect thinking about fairness. Because particular resources and health needs were linked in the vignettes (pills to pain; apples to nutrition deficiency), it is difficult to attribute different patterns of choice to differences in the resource vs. differences in the health condition. Even so, the many participants who saw the vignettes as about essentially different problems felt ‘…pain becomes more real than a nutritional deficiency’ (Participant R), whereas ‘[Vitamin F] is not really a life and death thing so it’s not such a hard decision to make’ (Participant B). The urgency, ‘realness’ or relative seriousness of pain was noted. For example:

‘I don’t think the apples were quite as serious as the medication. I think the medication is a more serious vignette because I mean there’s pain on the line there’s something serious on the line. So yeah they’re a little bit different. One’s more true to life’

(Participant A).

For some, pain entails a stronger imperative to distribute fairly:

‘It’s like, the same as the apples one, except that this one is, like, you feel more, like, it’s more important to share equally between the two, because this is just, like, pain and that’s more like just, like, food or a little bit, ah, like essential nutrients – but this is, like, pain, so you want to make sure it’s equal’

(Participant L).

Another interviewee noted that ‘taking away their pain medication is more of a big deal’ compared with the problem of trying to improve nutrition (Participant O). Even so, one interviewee found neither scenario serious enough to tap into real values:

‘The problems aren’t all that severe. You mentioned about vitamin F and about pain relief. Now, if you were to ask me questions of a more serious content…’

(Participant F).

Desire for additional information and considerations

The survey vignettes offered short, simple descriptions of the resource, recipients and potential claims on the resource, stating that in all other respects the recipients were identical. When asked what else they would like to have known about the situation to determine what is fair, interviewees identified several characteristics not included in the vignettes. Two interviewees said they would like to have known the ages of the recipients; one wanted to know recipients’ weights. One interviewee was interested in knowing whether either of the participants was poor, particularly as it related to their ability to get the resource. Another queried whether the desired benefit might be achieved by some other means. Three interviewees noted additional possible benefits from the resource (besides nutrition or pain relief), such as enjoyment, satisfaction or health benefits not specified by the scenario.

With regard to health in particular, at least 15 interviewees wished to know the downstream consequences of the health benefit described in the vignettes, for example, whether pain relief improves quality of life or whether vitamin F improves health. Some were interested in clinical endpoints (the ability of apples to increase vitamin levels):‘I was expecting a statement here that mentions the least each one can absorb’ (Participant P). Others focused on symptomatic benefits (the ability of pills to give pain relief), or more general health benefits (well‐being, functionality, etc.).

Three participants raised questions about the recipients’ respective health status, particularly the duration, curability or seriousness of illness, and life expectancy. Four wished to know whether either recipient was personally responsible for their condition, while one raised personal responsibility as a factor he/she would deliberately not want to know about. In the case of pills for pain, 10 participants wished to know more details about the pain experience: recipients’ relative pain tolerance, pain while asleep (mitigating need for 24‐h relief), causes of the pain and potential for addiction to pain pills.

A common concern was the nature of relationship between recipients, or between the recipients and the allocator. This theme arose both from our query about desired additional information and as interviewees mused about the nature of a fair allocation. Interviewees wondered whether there was any potential for sharing apples or pills between the recipients (although this was prohibited in the scenario). Three interviewees wondered whether the recipients would be aware of each others’ respective problems or be able to respond to the others’ needs by re‐allocating directly between themselves (also prohibited in the scenario). For two interviewees, the relationship between the allocator and the recipients mattered: for example, would the allocator communicate to the recipients the reason for unequal portions? One interviewee noted that unequal divisions of apples would require an explanation to be fair, whereas an unequal division of pills would be more acceptable given the medical context.

Attitudes towards specific, qualitative features of the vignettes sometimes coloured their allocation approach. The survey researchers had intended these features as neutral, background information and not salient determinants of fairness. An extreme example concerns the variety of apple:

‘… they’re just telling you plain apples, maybe what type of apple? Was it green, Granny Smith or a Macintosh apple or whatever. It sounds silly but I think it would have helped a little bit more …’

(Participant C).

A few interviewees noted that they found the scenarios unrealistic, for example: ‘It doesn’t usually happen that one person would start dividing up apples’ (Participant G) or ‘…I don’t even know of a vitamin F’ (Participant D). One interviewee described approaching hypothetical choices differently:

‘I guess one of the biggest issues for me was thinking about it in practical terms as opposed to the ideal … like the survey was, like, okay this is the ideal situation what would you do. But I was kind of debating, like, I was thinking about like what… like, in life, would we actually know those circumstances and stuff? So I was trying not to let that kind of thing affect my decision’

(Participant J).

An unexpected principle: everybody gets something, nobody gets nothing

The survey was designed to distinguish preferences across five fairness principles conventionally recognized in resource allocation ethics (egalitarian, utilitarian, equal benefits, according to need, and maximin). Choices self‐reported by interviewed participants (Table 3) favoured allocations according to need as ‘most fair’ and allocations according to utilitarian maximization of population benefit as ‘least fair’. These choices were consistent with patterns in the larger RA survey samples (Table 3). In the first wave of the interviews, we asked interviewees to describe their reasons for their allocation choices. Unexpectedly, seven of the 39 interviewees articulated a principle we had not included in our framework and uncommon in bioethics and health economics literatures. This principle may be simply stated: ‘Everyone must get at least something, regardless of what the rest of the distribution may look like’; in the words of one respondent: ‘…like, what’s the least fair? And it’s, like, well, obviously the one where one of them gets zero and one of them gets a lot’ (Participant K). These interviewees rejected any allocation that left any recipient with nothing, even if the rationale for the allocation was otherwise appealing, for example:

Table 3.

Survey choices made by qualitative interview participants (self‐reported), compared to those of the quantitative survey participants (actual)

	Chosen as ‘most fair’		Chosen as ‘least fair’
	Qualitative interview sample	Quantitative survey sample1	Qualitative interview sample	Quantitative survey sample1
Apples for ‘vitamin ‘F’ N in sample2	46	560	41	560
Fairness principles (%):
According to need	47.9	40.9	0.0	3.0
Equal benefit	26.1	24.6	4.9	3.3
Egalitarian (equal division)	17.4	15.4	7.3	9.3
Maximin	8.7	14.8	2.4	5.9
Utilitarian	0.0	4.3	85.3	78.6
Total (%)	100	100	100	100
Pills for pain N in sample2	43	560	42	560
Fairness principles (%):
According to need	44.2	41.1	0.0	3.0
Equal benefit	39.5	28.0	2.4	5.5
Egalitarian (equal division)	9.3	7.1	11.9	17.3
Maximin	7.0	19.5	0.0	8.8
Utilitarian	0.0	4.3	85.7	65.3
Total (%)	100	100	100	100

¹These represent choices made by all of the quantitative survey participants when allocations were described to them in both verbal terms and quantified allocations (16).

²For a minority of interview participants, data are missing regarding these specific opinions (e.g. some answers were not interpretable in these quantifiable terms, or the interview took a different turn in focus), hence the smaller number of individuals represented in some of these columns.

‘So, well, economically the most optimal thing to do would be E so that you know the total amount of Vitamin F that they get is 1200, but then the other person wouldn’t get any. So I think that’s why I chose the next best answer, which is B’

(Participant U).

Sometimes reasoning focused on the unfairness of one person getting everything, rather than the person getting nothing:

‘… one of the answers, give as much apple as possible to them, which I don’t think is the fair at all because, like, if we satisfy everyone’s want by giving them unlimited, like, resources it’s no fair to other people on earth’

(Participant Q).

Two interviewees framed getting less as a type of ‘punishment’ and fair only when punishment is deserved. In one variation of this idea, the neediest is unfairly punished because this person not only has a health problem but also makes less efficient use of treatment: ‘…just because Taylor [an individual in the survey scenario] can’t process it or processes it better I guess in terms of how long it lasts. He shouldn’t be punished for like the supply is limited…’ (Participant R). Another variation argues that the fittest person not be punished:

‘The natural selection idea – that you shouldn’t be punished because you’re stronger in that regard’

(Participant R)

‘I don’t know, I didn’t see why one guy should be penalized for apples just because the other guy happens to not metabolize the same way as he does’

(Participant I).

Disjunctures between survey choices and fairness beliefs

It is possible that some interviewees’ survey responses reflect reasons other than those intended, or imagined, by the survey researchers. In the debriefing interviews, interviewees often wavered when we asked them to recall what they chose. Occasionally, an interviewee indicated that they had made one particular choice on the survey, but then corrected it a moment later. One participant was unsure that his/her choice reflected his/her fairness beliefs: ‘I’m sure about the principles but I’m not sure about the survey’ (Participant E). Some interviewees read unexpected things into the vignettes. For example, two interviewees imagined that the health interventions might have been given in the context of a clinical trial and thus that the allocation should be driven by the methodological goal of measuring the resource’s effects, for example:

‘… I thought of it as a study and I wanted to see how, well, I took it as how it was consumed by each one of their bodies. I thought it would be easier if they both had the same when they’re getting results’

(Participant S).

One interviewee interpreted the essential issue as medication dosing, similar to his/her personal dilemma with a prescription drug. One could not think of any similar ‘real‐life’ situations, because he/she ‘[didn’t] really know anybody that takes, that’s taking pain killers right now’ (Participant H). In an unexpected twist, one posited that apples and pills could be substituted for each other: ‘Pills are addicting – if you’re an addict or something, I think you’d want to get away from it and stick with the apples’ (Participant C). Finally, three interviewees candidly reported that their responses were unrelated to the resource allocation problem or to their notions of fairness – rather, these participants simply chose the first answer. One explained that the first answer is usually ‘right’ on multiple choice tests.

Discussion

This study has explored how participants in an economic survey interpret stylize allocation scenarios, identify ‘most fair’ allocations and rationalize their choices. Qualitative debriefing interviews revealed that the RA Survey vignettes were sometimes interpreted in unintended, yet analytically relevant and interesting, ways. These findings have implications both for understanding lay thinking about fairness in the health system and for survey methodology.

Survey researchers want participants to approach simplified allocation choices using the same principles they would endorse for complex policy dilemmas. However, the reduction of a macro‐level allocation dilemma to a two‐person scenario qualitatively alters the problem in ways that may affect the interpretation of the survey choices. The RA Survey researchers intended pills and apples to represent, by synecdoche, health‐care and non‐health‐care resources in health systems. Further, individuals (named ‘Williams’ and ‘Taylor’ in the scenarios; see Table 1) with competing claims on resources at the micro‐level were intended to represent types of claimants at the population level. In our interviews, we found that not all participants readily make these conceptual leaps, for example, from ‘pills’ to ‘health‐care goods’ or from ‘Williams’ to ‘greater ability to benefit’. Some participants leap instead in unanticipated directions.

Most of the participants we interviewed did not readily make analogies between the RA Survey problems and ‘real‐life’ problems. Of those who did, some saw the micro‐level survey dilemmas as similar to other micro‐level, non‐health‐sharing tasks in life (commonly, dividing treats or attention among children). Others evoked more macro‐level, but fictional, dramatizations from television or coursework. Because our participants were mostly young adults, it is possible that we may not have found this in groups with more professional or personal experience [although we would note that the distributions of judgments themselves did not differ meaningfully between the university and community samples ¹⁶ ]. Ryan et al., ¹⁴ for example, in their qualitative study of older adults found more participants drawing on personal experience to make quantitative economic trade‐offs.

The RA Survey vignettes aimed to focus participants on very few features of resources or recipients, corresponding to five abstract principles of fairness. However, we found that some participants drew additional features and implicit principles into their reckoning. Foremost, the type of resource matters. Health economists commonly argue that health resources should be understood primarily as instruments of health production and that policies concerned with equity should focus on the fair distribution of health produced (rather than distribution of instruments). ¹⁶ Contrary to this view, our participants did not readily distil resources into their final health effects. The two types of resource (apples, pills) also differed in equity‐relevant ways beyond the health‐care vs. non‐health‐care dichotomy intended in the survey. Participants saw the need for pain relief as qualitatively different – more ‘real’– than nutritional needs. Walzer’s ²⁰ classic study of the organizational parameters of justice suggests that goods in different spheres (education, health, etc.) entail different moral frameworks and allocation principles. It is possible that smaller ‘spheres’, with their own idiosyncratic justice imperatives, exist within the larger sphere of the health system.

We found that some people desire detailed information before making equity judgments. Other qualitative studies of quantitative economic tasks have also found this and have suggested possible economic rationality behind such desires. ¹⁴ , ¹⁵ In our study, however, interviewees did not always relate their demands for extra information to a specific analytic task (such as quantitatively calculating and comparing trade‐offs). For some, extra details would aid inductive scanning for more possibly relevant issues (recall the participant who wished to know the variety of apple, even while acknowledging that this detail may be ‘silly’). Faravelli has defined ideologies concerning equity in terms of the presumptions people make about the differences between competing recipients for a resource in the absence of information about any actual differences. ¹¹ Insights into such assumptions may help us understand the nexus of ideologies and rational principles in the creation and acceptance of fair policies. Ryan et al.’s qualitative study ¹⁴ similarly found people inferring extra information to make simplistic economic trade‐off tasks tractable. Paradoxically, their participants seldom made use of more than one or two of the attributes provided within the task descriptions – they found, importantly, that the focus on few attributes was not an attempt to simplify the problem but because of the special importance of those attributes. It may be important to identify the full range of attributes that people care about before attempting to discern their relative salience to fairness.

In our study, a number of participants would have preferred to see health resources allocated in a more fine‐grained way than the vignettes allowed, with consideration to special features of the context, the recipients or the resources. Potentially salient features of the context include relationships among the recipients and allocators, how affected individuals might communicate or understand allocation reasons and recipients’ power to reallocate resources at their discretion. Potentially salient features of the recipients include their personal responsibility for health problems, ability to tolerate the health problem, access to substitute goods and age. Certain features of resource might also affect the chosen principle for its fair allocation; these include the resource’s downstream effects on health or life, incidental risks (e.g. addiction) or available alternatives for achieving the same outcome. Such concerns are well recognized in health services and health economics. Another concern is whether the allocators can relate personally to ‘needing’ the resource themselves. This feature is potentially relevant to the political and social dynamics of fairness in health policy and may warrant future investigation.

The survey setting is a social setting, and survey participants respond to the social cues and symbolic systems of this setting. Participants can be distracted or influenced by details that researchers might consider incidental or artefactual. Some participants read imagined circumstances (e.g. a clinical trial context, a multiple choice ‘test’) into the scenarios, which in turn affected their reasoning. It is worth considering how the survey setting resembles other social settings familiar to the participants to avoid, for example, implications that there is a ‘right’ answer, and to discourage strategically motivated responses.

In designing the survey, the RA Survey researchers focused primarily on the relative quantities of a good to be allocated to the two individuals and related these carefully to underlying distributive principles (relative needs, ability to benefit and so forth). However, some participants paid unexpected attention to absolute quantities of resources received, in addition to their relative distribution. Foremost, several participants considered unfair any allocation in which an individual gets none of a good. We had not foreseen the ethical salience of allocations entailing ‘zero’ quantities. In the survey, a ‘zero’ allocation was always conflated with the utilitarian rationale, which was consistently the least‐favoured choice for most fair (Table 3). Later administrations of the quantitative RA survey tested whether the utilitarian allocation was judged ‘most fair’ less frequently when it allocated zero to one potential recipient than when it allocated positive amounts to both individuals. A statistically significant, but small, effect was found. ¹⁶

We note some limitations of this study. Given its role in a mixed methods project, this qualitative investigation was shaped by the context and aims of the quantitative survey. If we had interpretively studied fairness beliefs in a broader context, we would ask more exploratory questions and might have re‐interviewed individuals as theory develops (this was not possible in our case, with an ethics protocol that required anonymity for the survey). Second and importantly, the legitimacy of our respondents’ specific values for policy making has not been established. Some values expressed by our participants (e.g. personal responsibility for health problems as a condition for fair resource allocation) are clearly ethically problematic. Finally, this investigation is not a test of the validity of the RA Survey questions, although it provides insight into how people can interpret such questions. The qualitative findings inform theory building, particularly for understanding how concepts of fairness are constructed.

Conclusions

This study demonstrates the value of investigating the ideas behind preferred resource allocation scenarios. Qualitative inquiry sheds light on processes, imagery and principled reasoning behind lay judgments of fair allocation.

The RA Survey researchers distil allocation problems to an essential triad of concerns: resources (health care or non‐health care), recipients (by health status or ability to benefit) and principle‐based claims to resources. Some participants related these to other micro‐level situations, rather than to macro‐level policy. Further work is needed to understand both the ethical and conceptual legitimacy of extrapolating inter‐personal sharing to population‐wide distributions in the health sector. Some participants either wished to, or did, complicate the simple scenarios with additional features (of the context, recipients or resources) to make the allocation problem more tractable. Some researchers may interpret this as a strategy to evade tragic choices. However, it is worth investigating other possibilities, for example, the mediating role of other fairness‐relevant criteria (e.g. social relationships), the appeal of more casuistic approaches to fair reckoning and inductive interpretations of what is at stake and what matters in any given allocation dilemma.

Our participants made analogies between the survey scenarios and spheres other than health care, raisings doubts about the relevance of such surveyed beliefs to health policy decisions. Future research might explore how personal experiences, narratives of rationing or scarcity in the popular media, and other images of ‘real‐world’ rationing affect fairness reasoning applicable to health policy. This also warrants investigation into how people characterize the ‘spheres of justice’ ²⁰ implied by various allocation problems and how their fairness reckoning translates (or not) from one to another.

Finally, when tasked with making fair allocations, people attribute meaning to absolute, as well as relative, quantities of a resource. Giving a lot, a little –or, especially, nothing – may seem unfair regardless of the principle justifying the amounts. This idea has little traction in formal ethical or economic theories of fairness, yet often arises in politics where, for example, a popular maxim advises that ‘nobody gets everything, nobody gets nothing, everybody gets something’. ²¹ (p. 101) The RA Survey ¹⁶ quantitatively tested support for this principle in a larger survey sample and found little effect. These paradoxical findings support future inquiry into relations among popular, political and ethical constructs of fairness, and their material consequences for fair resource allocation policies.

Conflicts of interest

None.