Should You Compensate Research Study Participants?

Recently I got a comment from Reviewer Two that said (paraphrasing), “I’d like the authors to explain why they didn’t compensate interview subjects, because it’s ethically required to compensate people fairly.” This made me laugh, because of course the opposite is closer to true: It’s almost always acceptable to NOT compensate research subjects, and any offered compensation introduces a risk that consent is not freely given.

Money is Coercive

A central tenet of human-subjects research is that consent must be freely given. Consider the hypothetical example of two subjects who are participating in a trial of an experimental vaccine to prevent Purple Toe Disease:

John is worried about possible side effects of an untested medicine, but has decided to participate because his sister suffered from Purple Toe, and he would gladly do anything he can do to help others not suffer the same fate.

James is worried about possible side effects of an untested medicine, but has decided to participate because he is really short of money right now, and the $500 offered compensation for transportation and his time on three trips to the trial site would help his budget.

John’s participation is freely given because he is sincerely willing to participate. James’ participation is in part coerced by the compensation offered. Regrettably, a lot of medical research is problematic for this reason — all too often, we test our experimental drugs and treatments on the poor.

The Belmont Report is a key guiding document for the ethical conduct of research, and articulates three principles: respect for persons (treat people as autonomous), beneficence (maximize possible benefits and minimize possible harms), and justice (distribute the benefits and risks of research evenly across society). Historically, the research community has been pretty good at living up to the principles of respect for persons and beneficence, and terrible at achieving justice. Poor people take on too much of the risk of our experimental treatments.

Compensation Changes Your Sample

Participating in a 90-minute video interview (as in our study) is a lower level of risk than agreeing to take an experimental drug, but the same principles apply. There is nothing wrong with asking someone to spend an hour or two doing a task that they find interesting and not pay them. If you ask me to talk about my favorite online sites, I’d gladly do so. I conduct department-level human subjects review for my school, and lots of protocols have no compensation. Is it wrong to ask someone to play with a robot for an hour for free?

Another way to view this through theories of “motivation.” Motivation can be either intrinsic (valuing the thing in itself) or extrinsic (valuing an added reward). For example, reading a book because it looks interesting is intrinsic motivation, and reading a book to win points in a reading contest is extrinsic. Payment for study participation is an extrinsic reward. In the robot example, if you pay people you may get more volunteers who just want $25 and if you don’t you will get more people who are excited to play with a robot. Depending on the study design, it may be important to find people excited to play with a robot (intrinsically motivated), or that factor may be irrelevant (anyone who follows the specified procedures is acceptable). Whether you compensate people changes the characteristics of your sample of human volunteers, hence whether it makes sense to compensate people depends on the study objectives.

If your study is looking to enroll people who have a particular background or set of prior experiences, paying participants also can sometimes compromise your sample by introducing people who may not even be a good fit for the basic study criteria. This happened in my lab in Lily Bernstein’s undergraduate thesis. Lily studied what happens when people fight with friends and relatives on Facebook. We offered compensation to participants, and conducted a few interviews with people who hardly used Facebook at all. The solution was to implement a more rigorous pre-screening survey for potential study participants.

How Much Compensation is Appropriate?

Compensating participants is often appropriate, but the trick is to not compensate them so much that it becomes coercive. Returning to John who sincerely wants to help in the fight against Purple Toe Disease, offering him money to cover his time and travel for study participation is appropriate. Picking the right amount, however, is tricky. It should be enough to be respectful of John’s time, but not so much that James decides to participate despite deeply held reservations.

One of the most thoughtful papers to address the issue of how much to compensate subjects is Chris Le Dantec and Keith Edwards’ paper Designs on dignity: perceptions of technology among the homeless. Chris and Keith studied how unhoused people use technology like mobile phones. After much reflection, they chose to compensate unhoused participants less than one might normally do. This strikes me as extremely sad, but also ethically mandated. For an unhoused person, small amounts of money can be coercive. Lower compensation helped make sure consent was freely given.

My lab conducted one study recently where participants were not compensated, and we ultimately regretted that decision (as noted in the paper). The paper is a study of black content creators on TikTok, led by PhD student Camille Harris. A number of our participants were successful streamers making good money from endorsements. If you are interviewing an online celebrity who is thriving, you certainly don’t need to offer them $25 to talk to you about it, and it might be considered insulting to do so. However, other streamers in our sample were struggling to make a living as content creators. For all of our participants, the issue of appropriate compensation was salient — they feel that platforms don’t pay them enough and compensation is distributed unjustly based on race. We didn’t do anything wrong (our volunteers chose to participate knowing there was no compensation); however, if I could do it over again, I would certainly compensate participants in this case.

Minimum Wage?

The last issue I want to address is whether it is advisable to compensate subjects at least the legal minimum wage for where they live. This is tricky. Compensation for study participation is NOT pay. You are compensating them to cover expenses or to say thank you for their time. In French, the word for “tip” is “pourboire” — which literally means “for a drink.” Research compensation in theory should be like that — a token of appreciation, but not a wage.

The problem here is that there is a gap between theory and practice. Compensation should be just a token thank you, but people in need of money often participate in studies for the money. For this reason, I always compensate subjects at least the minimum wage; however, this is not strictly required.

And To Reviewer Two

It was my mistake to not explain all of this more clearly in our paper. I hope this blog post helps — these issues are not straightforward. And if I ever see a paper review that expresses concern that the subjects were compensated too much and that might be coercive, I will open a bottle of champagne.