Data collection is often a burdensome and time-consuming activity for research participants, particularly when it involves hours-long surveys. Researchers may wonder if they should pay participants for their time, how much they should offer, and whether it should take the form of cash or in-kind provision. They may worry that a failure to pay participants risks exploitation, but also that the promise of payment may unduly influence or even coerce people to participate, particularly when prospective participants are poor. I explore these issues in this post, focusing first on the rationale for payment before turning to concerns regarding coercion and undue inducement.
Luke Gelinas et al. have developed what is arguably the most comprehensive framework for designing and evaluating offers of payment to research participants. They distinguish three ethically acceptable rationales for payment. First, researchers may reimburse participants for out-of-pocket expenses related to participation, including travel, meals, accommodation, and childcare. The purpose of reimbursement is to ensure that participants do not have to pay to contribute to socially valuable research. Rates of reimbursement should thus reflect the true out-of-pocket costs participants incur.
Researchers may also compensate participants for the time and burden of participation. Here Gelinas et al. suggest that participation may be equated with employment, with rates of compensation reflecting “what would constitute fair remuneration for analogously time-consuming and burdensome, albeit unskilled, employment.”
Finally, researchers may incentivize participants to enroll and remain in a study. Incentives are payments above and beyond what is necessary to reimburse or compensate participants and may be necessary to ensure that studies meet recruitment and retention targets.
Gelinas et al. suggest that reimbursement and compensation should be the ethical default on fairness grounds. These payments may be waived if there is a justification for doing so, for example if a payment is expected to bias the results. Decisions to waive payment, however, should be closely scrutinized, because payment may be necessary if researchers are to avoid exploiting participants.
Researchers exploit participants when they take unfair advantage of them, for example when they take advantage of the vulnerability of participants to secure sizeable benefits for themselves—e.g. publications in prestigious journals, career advancement—while offering participants little in return. To avoid exploiting participants, researchers must ensure that participants receive a fair share of the benefits of the research. Researchers may therefore need to conduct studies that are responsive to the needs and priorities of the host community and work to ensure that the research findings are used to meaningfully impact people’s lives. At a minimum, research participants should be reimbursed and compensated for the time and burdens of participation. These payments are particularly important for participants in the control arm who will not directly benefit from the intervention.
A further question is whether payment should take the form of cash or in-kind benefits, such as training programs, food stuffs, stationary, or water purification tablets. Researchers may argue in favor of offering in-kind benefits on the grounds that cash may contribute to certain forms of bias, particularly social desirability bias. They may also worry that cash may lead to family or community conflicts or enable participants to engage in self-harming behaviors, for example if they struggle with substance-use disorders.
Since cash is most valuable to participants in most contexts, arguments in support of offering in-kind payments to study participants face a justificatory burden. Bias is a serious concern since it may undermine the validity of a study’s results, but researchers should provide evidence to support their concerns and be willing to offer in-kind benefits substantial enough to fully reimburse and compensate participants. Providing in-kind goods rather than cash to competent participants for their own good is also objectionably paternalistic. Reimbursement and compensation are owed to participants on grounds of fairness and refusing to offer cash to competent agents—whether they have a substance use disorder or not—fails to respect them as autonomous persons. It may also institute a double standard in societies where people are free to use their legally obtained gains—e.g. employment income—as they wish. If it’s wrong to pay local research staff in-kind, why would it be acceptable to pay research participants in-kind? Researchers must therefore provide a strong justification for offering payment in-kind, for example that cash payments in certain contexts are culturally inappropriate, will lead to bias, or are likely to result in severe harms to participants and bystanders.
Lowering payment does not necessarily eliminate the harm of subjection and only makes participants worse off.
Researchers working in low-income countries may agree with the above rationale for payment but worry that the use of payment as an incentive (or even compensation) may coerce poor people to participate, thus invalidating their consent. On this view of coercion, participants are coerced to enroll in a study if they have no reasonable alternative, or if they would not have chosen to enroll absent an offer of payment. Indeed, a recent survey of US-based IRB members finds that many share this concern (and this understanding of coercion).
As Joseph Millum and Michael Garnett argue, there is something to this worry, for coercion can take the form of coercion as subjection. This form of coercion occurs when someone has no choice—e.g. due to poverty—but to do what another person wants and so is subject to their will. Since self-direction is an important component of wellbeing, coercion as subjection is harmful to participants. But, where coercion as subjection involves an offer (participate and receive payment) rather than a threat (participate or else!), it does not undermine consent. Moreover, the appropriate response to this harm is to compensate participants for it by increasing payment. Lowering payment, by contrast, does not necessarily eliminate the harm of subjection and only makes participants worse off.
Unlike coercion as subjection, consent-undermining-coercion involves the use of threats, not offers. More specifically, consent-invalidating coercion involves one agent—A—threatening to harm or violate the rights of another agent—B—unless B performs or refrains from performing some action. Whereas the use of threats invalidates a person’s consent to participate in research, there is good reason for rejecting the idea that offers undermine consent. Even if B agrees to A’s offer because B has no reasonable alternative, or if B would not accept the offer if it didn’t involve payment, A does not necessarily wrong B and B’s acceptance is not necessarily involuntary. I may agree to chemotherapy for my prostate cancer because I have no reasonable alternative to doing so; I may not show up to work each day unless I am paid. Nonetheless, my consent to treatment and my consent to employment are valid.
Since consent-undermining coercion involves a threat not an offer, offers of payment do not undermine participants’ consent. A related worry is that offers of payment may unduly influence people to enroll in a study if the payment is so high that they fail to adequately consider or appreciate the risks of research. This concern may be legitimate in cases where researchers must employ payments as incentives to meet recruitment goals and prospective participants are poor.
But while undue inducement of this kind is possible, it’s not at all clear that it is likely. Of the limited existing empirical evidence on this issue, one study suggests that offers of high payments for participation may increase people’s perceptions of risk and time spent examining risk information. Two other studies find no effect of increased payments on people’s perception of risks. In addition, where studies are low-risk, concerns regarding undue influence are far less pressing, because a lack of appreciation of a study’s risks is less consequential for participants’ wellbeing.
To sum up, research sponsors should set aside money to reimburse participants for their out-of-pocket expenses and to compensate them for their time when such payments are unlikely to bias the study’s results. Doing so is often necessary to ensure that participants are not exploited. It is also permissible to incentivize participation to meet recruitment and retention goals. While researchers may worry that payments may coerce or unduly induce people to participate, thus undermining their consent, the former concern is misplaced while the latter is not supported by the available evidence.
After reading this blog from Dr. MacKay, the 3ie team offers these reflections
1. We need to consider how we value the input we (research funders, producers, and consumers) are all often dependent on: the time and personal data provided by research participants
We make strong arguments that our research produces a social value and the data produced is a global public good that can support learning and evidence generation for more effective policymaking. This data is used by researchers to advance their careers and underlies analyses in publications that are placed behind paywalls, fueling a lucrative academic publishing business. For ethical conduct of research, we should be carefully considering participant reimbursement, compensation, or incentives.
2. We need to clearly define and quantify the survey burden for research participants
3ie funds and implements surveys that can range from 30 minutes to more than 2 hours in duration and require engagement with multiple household members, resulting in a burden of several hours per household per interview round. When conducting panel surveys, this burden can be doubled or tripled for the same research participants over the research life cycle for just one study. This burden to participate in the research should lead to careful consideration of participant payment.
3. Participant payments—alongside other methods for valuing participation and balancing power—must be discussed and defined through community engagement
The research team should rely on community engagement to determine the appropriateness of participant payments, including whether, when, how much, and via what modality they should be provided. This process requires an understanding of the potential barriers to participation in the study, such as transportation, childcare, or phone access, that should inform payment levels or alternative resources provided to the participants. Careful discussions with the community and research partners can reveal the potential influence of a payment or the potential for feelings of exploitation in the absence of a payment. These efforts should complement other work by the research team to balance power between researchers and research participants, such as establishing a stronger feedback loop for research participants.
4. Limited funding is not an ethical argument for not paying participants
We need to think about research participant payment as early as possible in the research life cycle to correctly budget the study. Failing to account for appropriate and ethical consideration of research participant payment keeps the true cost of the research artificially low.
5. We need more transparency and standardization around the factors that inform decisions about whether, how much, and how to pay participants
After decades of evidence generation and data-driven efforts, there is little data available on average survey duration, burdens placed on research participants, and research participation payment standards. We know there are some contexts where participant payments are simply not feasible, such as when it goes against research partner or local regulations. But, there is a need for more transparency and standardization in how data-driven professionals value the main input for their activities: participants’ time and personal data. And as researchers, this gap presents an opportunity for us to generate more evidence around these issues.
This is the third in a series of blogs on ethics in social science research. Read the introductory post here.
This article was originally published on on 3ie’s blog: Evidence Matters.
