Manipulation in the Enrollment of Research Participants

1. Introduction

Dr Khan is trying to enroll patients at his hospital into a clinical trial of a new hypertension medication, but recruitment is slow. A colleague advises him: “Stop having the staff ask people if they want to be in the study. You need to use the ‘white coat effect.’ Have the nurse show the patient into your office. Make them wait, like it's a privilege to see you. Put on your coat and stethoscope and sit behind your desk. These people don't like to say no to doctors—they'll agree if you ask them.”

Recruitment is a challenge for many biomedical research studies with human participants. Recruiting sufficient eligible participants for a study is essential if the study is to be adequately powered to yield socially valuable information, but recruitment can be a time consuming and expensive process. Strategies to increase the speed and ease of recruitment are therefore valuable.¹

Various factors can influence whether someone enrolls in a research study. She may be hoping for a cure for her illness, trying to help other people, or simply in it for the money.² More subtly, the chances of her agreeing to participate may be increased by factors such as her respect for the authority of the recruiting physician, trust in the research institution, or peer pressure. When researchers are aware of the factors that affect their potential participants’ decisions, they may be able to design recruitment strategies that make use of them in order to encourage people to enroll. In the case above, for example, by adopting his colleague's suggestion Dr Khan would be leveraging patients’ respect for the medical profession in order to motivate them to take part in his study. Most discussions of inappropriate ways in which people could be influenced to give consent to research participation have focused on coercion or undue inducement.³ But Dr Khan would be engaged in neither. Instead, if his recruitment strategy is problematic, it is because it would be manipulative.

In this paper we analyze the non-coercive ways in which researchers can use knowledge about the decision-making tendencies of potential participants in order to motivate them to consent to research enrollment. We identify which modes of influence preserve respect for participants’ autonomy and which disrespect autonomy, and apply the umbrella term of manipulation to the latter. We then apply our analysis to a series of cases adapted from the experiences of clinical researchers in order to develop a framework for thinking through the ethics of manipulating people into research participation. All manipulation disrespects autonomy and is therefore pro tanto wrong. However, only deceptive manipulation invalidates the consent that results from it. Use of the other forms of manipulation can be permissible, but only if the outcome of using manipulation is sufficiently good and if the research cannot be carried out using ethically preferable means to obtain consent.

2. A taxonomy of motivation

We begin by parsing out the ways in which one agent can influence another in order to motivate a specific action which the second agent has an autonomy right to decide whether to perform. (An autonomy right is a right whose possession depends upon the bearer possessing the capacity for autonomous action such that she can exercise it.) Although we are ultimately interested in the action of giving consent to research participation, our analysis uses examples of actions from outside the domain of research over which competent adults also have autonomy rights. In each case, we determine whether one agent's influence of another respects or disrespects his autonomy.

2.1. Altering beliefs or preferences

Some forms of influence involve changing a person's perception of the options available to him in order to motivate him towards a specific action.

Persuasion

Elizabeth wants Miguel to join her on a ten mile run. To motivate him to do so, she points out that the marathon he signed up for is only a month away, and that going with her on a long run would be great training. Elizabeth presents facts about the ten mile run and makes a logical link between Miguel's existing interest in running a marathon and the benefits of the ten mile run. In doing so, Elizabeth persuades Miguel to go running with her.

One agent persuades another to pursue a specific action when she motivates him by showing rational links between his existing set of reasons to act and that action. She can do this by showing logical connections between his existing reasons and the act she wants him to perform, or by honestly presenting facts that are relevant to his reasons to act. Because she does not illegitimately interfere with his decision-making process, persuasion is respectful of autonomy.⁴

Deceptive manipulation

Bill wants Anne to buy his bike. She asks him if it is in good repair. Though he knows the bike is in poor condition, Bill tells Anne that it is in perfect shape. Anne pays Bill $100 for the bike. When she first rides it, the brakes don't work and the cogset falls off. Bill has deceived Anne into buying a broken bike.

When one agent deceives another about facts that are relevant to his reasons for or against pursuing a specific action, thereby changing his perception of the options available to him in a way that disposes him to perform that action, she uses deceptive manipulation.⁵ Deceiving him is an illegitimate way to interfere with his decision-making. It therefore disrespects his autonomy.

Persuasion and deceptive manipulation both involve changing a person's perception of the options available to him. Though persuasion, as we have defined it, respects a person's autonomy and deceptive manipulation does not, both involve some reliance on that person's rationality. There is a third form of motivation which also changes a person's perception of the options available to him but does not appeal to his rational faculties.

Motivational manipulation

A Hare Krishna member cheerfully hands a flower to a passerby and tells her it is a gift for her on this beautiful day. When she accepts the flower he asks if she will give a donation to the Hare Krishna Society. With the flower already in her hand, the woman feels bad about just walking on. Despite having no interest in supporting the Hare Krishna Society, she reaches into her wallet and hands him some cash.⁶

The woman's considered preference would be not to donate, but the Hare Krishna has manipulated her motivations so that her immediate desire to reciprocate his gift is stronger than her desire to avoid giving money to the Hare Krishna Society. The Hare Krishna takes advantage of the woman's predictable susceptibility to social norms of reciprocity so that she makes her decision about whether or not to donate on the basis of this immediate desire instead of on the basis of the values and preferences she would otherwise deem most relevant.

When one agent causes another to act on desires that, upon reflection, he would reject as reasons to pursue a specific action, she uses motivational manipulation.⁷ She can do this by stimulating a novel desire in her victim or by amplifying an existing desire. Either way, her interference in her victim's decision-making causes him to act based on immediate desires instead of the values and preferences he would, absent her influence, choose to involve in his decision-making. By interfering with his process of weighing his considered preferences and values, she makes his decision-making process less rational (without his consent). She thereby disrespects his autonomy.

2.2. Altering options

The forms of motivation discussed so far involve changing a person's perception of his options. It is also possible to motivate someone by changing the options themselves.

Coercion

A robber holds a gun to Rebecca's head, and says, “Your money or your life!” Faced with the robber's credible threat, and with no reasonable alternative but to comply with his demand, Rebecca hands the robber her wallet.

When one agent proposes to make another agent worse of if he does not comply with her demands, she coerces him.⁸ In a coercive situation, the coercee's action may be completely rational. Nonetheless, if the coercer does not have the right to control the coercee's decision by changing his options in this way (constraining the coercee's options and presenting new risks of harms to him), then the coercion disrespects his autonomy.⁹

Offers

Mudit offers Frank $20 to wash his car. Frank wouldn't have washed Mudit's car prior to the offer, but the prospect of acquiring $20 gives him a reason to do so.

While coercion involves a proposal to make a person worse off if he does not pursue a particular action, an offer is a proposal to make a person better off if he pursues that action. Unlike threats, offers do not constrain a person's options or present new risks of harms to him. Making an offer is a way that you can change someone's options in order to motivate him towards a particular action while still respecting his autonomy.¹⁰ Aside from coercion and offers, there is a third form of influence that involves changing the options available to someone in order to motivate him to perform a specific action.

Circumstantial Manipulation

Peter calls his friend John and asks if he wants to go out to dinner. They meet at an upscale restaurant. When the time comes to pay the bill, Peter tells John that he “forgot” his wallet and that he lives too far away from the restaurant to return home for it. As Peter knows, John places a high value on standing by his friends. Unlike Peter, John would not countenance leaving a restaurant bill unpaid. Not wanting to make a fuss or appear rude, John pays for dinner. If John had known that Peter would pull such a stunt, he would not have agreed to go to dinner with him in the first place.

By deceiving John about his intentions to leave his wallet at home, Peter has caused John to go out to a dinner at which the option of paying for the whole meal ends up becoming the most reasonable option for him. Without Peter's influence over his circumstances, John would not have chosen to put himself in such a situation.

When one agent motivates another using circumstantial manipulation, she illegitimately changes the options open to him in order to cause him to pursue a specific action. Because she uses illegitimate means, her control over him is illegitimate, and therefore disrespects his autonomy.

In the case of Peter's dinner, Peter's interference is illegitimate because he deceives John. His deception puts John into a situation in which the most reasonable thing for him to do is to act as Peter wants him to. Note that this is quite different from deceptive manipulation. In deceptive manipulation, the manipulee is deceived about facts relevant to the decision about whether to pursue a specific action. In this case, however, the deception is used only to get John into the situation where he has to choose whether to pursue a specific action (pay for dinner). Once there he may have a clear understanding of the facts that are relevant to his decision about whether to pay for dinner and the decision to pay may be perfectly reasonable—it is just that he would not have put himself into such a situation were it not for Peter's earlier wrongdoing.

There are several ways in which someone may control another's circumstances in order to influence his decision; not all of these constitute circumstantial manipulation. For example, suppose I want Sebastian to donate to my favorite charity. I know he has no interest in donating, so there is no point in asking him privately in his office. However, if I ask him in the break room, where his colleagues are present, he will donate something because he wants to preserve his reputation for generosity.

Being perceived as generous is something that, upon reflection, Sebastian values. He would therefore consider it reasonable to take it into account in his decision-making process (so my influence of him does not constitute motivational manipulation). However, Sebastian would still rather not be put in a position where he donates. Is my asking him to donate in front of his colleagues in the break room manipulative? If I simply make sure to ask him there when he is there anyway, our account would judge that it is not manipulative, since there is nothing wrongful in the way that I control the situation in which he chooses. On the other hand, it would be manipulative to lure him there under the false pretext that there was free cake and then ask him. I would then wrong him through deception and thereby disrespect his autonomy by illegitimately controlling his decision. These fine distinctions are ones that a normative theory of manipulation can help us make.

3. The Wrong of Manipulation

As our analysis indicates, the forms of motivation that may be described as manipulation are diverse.¹¹ Even so, our definition is not intended to try to capture all everyday uses of the term. Instead, it is a moralized definition that captures a set of forms of motivation with common normative features. There are two reasons for preferring such a stipulative definition. First, we are skeptical that the term manipulation, as used in everyday English, refers to one unified concept that could be analyzed in order to draw normative conclusions about people's behavior. Second, we are primarily interested in the ethics of different research enrollment strategies, rather than in the concept of manipulation per se. A definition that allows us to capture all the non-coercive but autonomy disrespecting ways to motivate people is therefore most helpful for our purposes.

The most well-known definition of manipulation in the bioethics literature can be found in Ruth Faden and Tom Beauchamp's A History and Theory of Informed Consent. They define manipulation as “a catch all category for any intentional and successful influence of a person by non-coercively altering the actual choices available to the person or by non-persuasively altering the person's perception of those choices.”¹² Faden and Beauchamp's definition locates manipulation as non-persuasive and non-coercive. However, it would include in the category of manipulation forms of influence that respect autonomy, such as offers. The forms of influence that we define as manipulation therefore constitute a subset of those that would fall under Faden and Beauchamp's definition. Since our interest is in analyzing the wrong of different forms of motivation, our moralized account is better suited to our purposes.

Deceptive manipulation, motivational manipulation, and circumstantial manipulation all disrespect autonomy.¹³ They are therefore pro tanto wrong. However, identifying manipulation as a pro tanto wrong is not enough to tell us how to evaluate the use of manipulation to encourage people to enroll into research. First, disrespecting someone's autonomy does not necessarily invalidate her consent. Under what conditions does manipulation into research participation invalidate consent? Second, we are interested in finding out if the use of manipulation to increase participation in socially valuable research is ever ethically permissible. Are there conditions under which the pro tanto wrong of using manipulation to recruit participants can be outweighed by countervailing reasons?

4. Manipulation and Consent

Five conditions must be met for informed consent to be valid: the person giving consent must have the capacity for autonomous action; certain information must be disclosed to her; she must understand certain facts about the act consented to; her decision must be voluntary; and she must indicate (token) consent.¹⁴ We are interested in situations in which manipulation may invalidate a token of consent from an autonomous person, that is, someone with the capacity to consent. We therefore focus on whether and when manipulation affects disclosure, understanding, or voluntariness to such a degree that consent is invalidated.

In the case of consent to research enrollment, appropriate disclosure requires at least that the researcher must tell the prospective participant information about the research that she has good reason to think may be dispositive of his decision regarding participation. The understanding requirement requires at least that the person who tokens consent must know what act or acts she is consenting to. Voluntariness in this context requires more detailed analysis.

Faden and Beauchamp analyze voluntariness in terms of control. They present a spectrum of control with coercion at one end and persuasion at the other. In addition to these two extremes, Faden and Beauchamp suggest that there are thresholds for “substantial control” and “substantial non-control.” They argue that in order for a decision to qualify as autonomous, it must be “substantially non-controlled” by any party other than the person giving consent. According to Faden and Beauchamp, manipulation is a form of influence that extends across the entire spectrum of control—some forms of manipulation are substantially controlling, and therefore invalidate consent, and some are substantially non-controlling, and therefore do not.¹⁵ They provide no specific criteria with which to determine which forms of manipulation are substantially controlling and which are not.

One way to work out the criteria for substantial control is to draw an analogy from the case in which we are confident that the degree of control is sufficient to invalidate consent, that is, the case of coercion. If we can determine what it is about coercion that invalidates consent—independent of features that distinguish coercion from manipulation by definition, such as threats—then we can investigate whether there are cases of manipulation that also have this feature or features. This will tell us whether and when manipulation invalidates consent by rendering the consenter's action substantially non-controlled.¹⁶

Not all threats are powerful enough to be controlling. For example, a threat to splash someone's shoes with water is so weak that we would expect it to be laughed off. When is a threat controlling such that it would invalidate a token of consent? We suggest that it is the point at which we would no longer consider the person threatened responsible for her actions, morally speaking. In cases of duress, we do not blame the person acting under duress but (if anyone) the agent who put her there. Likewise, in cases of coerced consent we do not attribute the act of consenting to the coercee. In the case of duress, the point at which someone is no longer responsible for her actions is, roughly, where she has no reasonable alternative but to comply.¹⁷ For example, a cashier who hands over the contents of the register to an armed robber who credibly threatens to shoot him has done nothing wrong. The same criterion can be applied to tokens of consent. Thus, someone who agrees to sex at the point of a knife is the victim of rape.

We now return to the three types of influence that fall under the umbrella of manipulation, and apply the conditions of disclosure, understanding, and substantial non-control in order to evaluate the validity of consent in each case.

Deceptive manipulation

Deceptive manipulation typically violates both the both the disclosure and understanding requirements for informed consent because the recipient of consent does not disclose relevant facts, and the giver of consent does not, therefore, understand what he is consenting to.¹⁸ For example, in the case of the broken bicycle, Bill fails to disclose something that he has good reason to think is relevant to Anne's decision about buying the bike. Both Anne and Bill then know that Anne takes herself to be consenting to buying a working bike, but Bill knows that the bike he is giving her is not working. Deceptive manipulation therefore renders consent invalid. In research, it is well accepted that deceiving prospective participants about aspects of a study that are likely to be relevant to a decision about enrollment is ethically problematic for exactly that reason. It would be impermissible, for example, for a researcher to lie about the expected side effects of an experimental treatment in order to increase enrollment. Whether deceiving participants about the methods or purpose of a study is ever permissible and under what conditions remains a matter of debate.¹⁹

Motivational manipulation

In motivational manipulation, one agent causes another agent to act on desires that, upon reflection, he would reject as reasons to pursue a specific action. However, while the manipulator changes the manipulee's preferences in ways that he would not or does not endorse, she may still disclose everything that she ought to about the act and he may still understand everything that he needs to about the act. As the example of the Hare Krishna who solicits a passerby for money shows, it is perfectly possible to know that one is being motivationally manipulated and yet for the manipulation to be successful. Moreover, motivational manipulation leaves the manipulee with the same alternatives to the act as he had prior to the manipulator's interference. His consent to the act can therefore be the result of motivational manipulation but still meet the disclosure, understanding, and non-control requirements, and so still be valid.

In the example of the Hare Krishna, the woman passing by knows that she is being asked for a donation, and knows what her money would be going towards. She also has reasonable alternatives that are plainly and clearly available to her, such as simply walking away without donating. Her consent to give money to the Hare Krishna is therefore valid—we would not say that she could legitimately take it back after walking away and thinking about the donation more carefully.

A. Leach and colleagues studied the informed consent process in an influenza vaccine trial in the Gambia. Some of the women they spoke to explained that they consented to enroll their children in the trial because “they were influenced by the group pressure of seeing other mothers joining.”²⁰ Suppose a research team were to use this information about the likely effects of peer pressure in order to maximize enrollment in a new study with women in the Gambia. They might design their information sessions so that potential participants are asked to talk to the group about how they feel about enrollment, and they might select women who they know are enthusiastic about the research to speak first.

The researchers involved in this study would know that their potential participants tend to overweigh the visible desire of their peers to participate when making their own decisions about participation. Outside of the context of the group dynamic that the researchers have created in the information session, a potential participant would not desire so strongly to emulate the choices of other women in the community. By constructing the recruitment process in this way the researchers would therefore be engaged in motivational manipulation. However, their tactics would not inhibit their disclosure of relevant information or the women's ability to understand it, and they would not remove the reasonable alternative of refusing to participate.²¹ The women could therefore still give valid consent, although the motivational manipulation would be disrespectful of their autonomy.

Circumstantial manipulation

In circumstantial manipulation, one agent illegitimately interferes with another in order to alter the options available to him such that one specific action becomes the most reasonable action to pursue. The wrongful means that the manipulator uses to alter the manipulee's situation makes her control over his decision disrespectful of his autonomy. However, when faced with the decision about whether or not to pursue the specific action, the manipulee may well have been told and have understood everything he needs in order to make an informed decision. (Indeed, since that act is now the reasonable thing to do, it may be in the manipulator's interests that the manipulee have a good understanding of the choice he is presented with.) Thus, circumstantial manipulation does not preclude the fulfillment of the disclosure and understanding requirements.

In the case of Peter's dinner, John may not initially realize that Peter intends to leave his wallet at home. However, once the bill comes, John understands fully what has happened and that what is being proposed is that he pay for the meal. John also has reasonable alternatives, such as paying his half and leaving. These alternatives are reasonable, even if John might strongly prefer to not choose them. His consent to pay for dinner is valid—Peter has not robbed him.²²

Duncan Ngare describes a situation in which medical students helped to conduct the consent process for a malaria treatment trial. In the community in which they were recruiting:

The culture also dictates that it is good practice to welcome visitors into the community and discourages disappointing them. Therefore, visitors are received warmly and almost anything they request will be provided without resistance...Whenever [the students] arrived at a household they would introduce themselves and seek informed consent...The research assistants were surprised that, even before completing their explanations of the project, individuals would consent to join the study...Moreover, although some women were reluctant to be interviewed, they did not say that they were unwilling because their culture does not allow them to show disrespect to visitors.²³

How would we judge them if the students had continued to ask people for consent in their homes after discovering these cultural norms? It might be that the norm of hospitality is not only deeply engrained in the members of this community but that it is also endorsed by them—that they would consider, on reflection, that the custom of being agreeable to guests is a good one, even if sometimes it risks making an irrelevant factor (hospitality) relevant to a decision about an activity like research. If the medical students were to have continued to go house to house recruiting, and, perhaps, delayed explaining the purpose of their visit until they had been invited inside, they would be engaged in circumstantial manipulation and therefore disrespecting the autonomy of their potential participants.

As in the case of using peer pressure, however, the potential participant's consent could still be valid: the research team could disclose everything that they ought to, the participants could understand what was being proposed, and refusal would still be a reasonable option, albeit one which the potential participants’ value system would strongly dispose them against. Thus, although the students would have acted wrongly in manipulating the potential participants, they would not have invalidated the participants’ consent.

5. Outweighing the wrong of manipulation

Having analyzed the circumstances under which manipulation invalidates consent, we can now address our second question: when manipulation does not invalidate consent, is it ever permissible? Motivational manipulation and circumstantial manipulation disrespect autonomy even though they do not invalidate consent and so are pro tanto wrong. However, if the pro tanto wrong of manipulation were outweighed by some other good consequence of the manipulation, then it could be ethically permissible, all things considered.

In a variety of everyday situations, the use of manipulation is intuitively judged to be ethically permissible. Consider, for example, government-backed anti-smoking strategies. In November of 2011, the Australian Parliament passed legislation mandating that cigarettes may only be sold in plain “olive-green” packages, following research results that deemed olive-green the most unappealing color in the eyes of young people.²⁴ The Australian government's packaging legislation is an attempt at motivational manipulation. If effective, it will prevent people from buying cigarettes by playing on the aversion they feel towards the olive green packets. However, no one would say that the color of cigarette packets is a genuine reason to smoke or not to smoke. By interfering with the potential smoker's existing alignment of desires and preferences, the Australian government makes her decision-making process less rational without her permission, thereby engaging in motivational manipulation. However, while this is manipulative, it appears to be widely considered that such strategies are acceptable ways to reduce the prevalence of unhealthy behaviors like smoking.

If we agree that manipulative strategies like the Australian government's are ethically permissible, this indicates that sometimes good outcomes justify the use of manipulation. However, the fact that manipulation is a pro tanto wrong implies that it cannot be justified only in virtue of resulting in a sufficiently good outcome; it must also be that other ways of achieving that outcome without similar or worse wrongs are not available. For example, if there were a morally innocent way to prevent adults from smoking, that would be ethically preferable to manipulating them. However, straightforward education to inform people about the risks of smoking does not significantly lower smoking rates, and more extreme measures that would achieve the desired outcome, such as criminalizing tobacco, are not currently feasible and might be more morally fraught.

Cases of justified manipulation may arise in the context of research, too. For example, in communities where it is customary to ask the permission of local chiefs or elders before conducting research, researchers may reasonably be concerned about manipulation. Even if the permission of the chief does not mean that he coerces people into participation, it may still be very influential and potentially manipulative. For example, in their qualitative study of recruitment and informed consent in the Kassena-Nankana District of northern Ghana, Paulina Tindana and colleagues describe the gate-keeping role played by traditional chiefs.²⁵ Though some informants, including several chiefs, stated that community members are free to say no even when the chief has given the research his stamp of approval, the authors note that the approval of the chief has a powerful influence on whether people agree to enroll. It might be that the approval of the chief has this influence because community members infer that he has assessed the costs and benefits of the research and decided that it is in the community's interests. Alternatively, however, community members might simply feel internal pressure to conform to social norms of respecting the requests of the chief. In the latter case, getting the chief's approval before asking people to enroll would be a form of motivational manipulation. It would therefore be pro tanto wrong.

However, Tindana and colleagues also suggest that in this region it would not be possible to conduct the research at all without going through the chief.²⁶ In contrast to the cases described in the previous section, in which it seemed likely that the research could proceed without using manipulation, the pro tanto wrong of manipulation could therefore be overridden in this case. When non-manipulative alternatives are genuinely impractical, the use of manipulation can be justified if the study is sufficiently important. For example, a study that is expected to substantially benefit participants or host communities (on whose behalf participants are asked to take on risks and burdens) might be important enough to justify manipulation. In cases where the direct benefits of the study are non-existent or uncertain—such as a phase 1 oncology trial in the United States—the social value of the knowledge expected to result would have to be extremely high in order to justify manipulating potential participants into enrolling.

6. Heuristics for decision-making

Motivational and circumstantial manipulation make use of potential participants’ decision-making tendencies in ways that undermine their autonomous decision-making. However, there are ways to make use of such tendencies and preserve respect for autonomy. One important example is the use of heuristics.

Consider trust. When trust is warranted, it is a legitimate reason to rely upon another person's opinion as a reason to pursue a particular action; it therefore provides a helpful short-cut or heuristic for decision-making. However, when trust is unwarranted, it can be illegitimately leveraged to motivate individuals to make decisions that they would not, upon reflection, endorse. Its use is then manipulative.

Tindana and colleagues describe the relationship between the residents and the local research institution as follows:

to some extent the longstanding relationship between the NHRC and the community has contributed to the latter's willingness to participate in research [...] Several participants voiced a belief that the NHRC has always brought interventions that have improved the health status of the people in the district. They therefore assume that anything the NHRC proposes is in their best interest.²⁷

Suppose a new researcher were to join the NHRC, and that his new colleagues advised him: “Always emphasize your affiliation. That way, people will trust you, and will agree to join your study.” Suppose this particular researcher's intentions and plans did not mirror the beneficial work of the local research institution. Though he emphasized his affiliation in order to encourage people to enroll, his research program was designed to generate publications that would further his career, and was unlikely to lead to findings that would benefit the community. The researcher's emphasis on institutional affiliation would mean that he was playing on the tendency of potential participants to trust studies that emerged from the local research institution in order to motivate them to join a study that bears no relation to the reasons they have for trusting the institution. He would be using manipulation (either deceptive or motivational manipulation depending on exactly how we understand the way trust motivates).

However, if the researcher's intentions and plans were in line with the work of the institution, then a potential participant who relied on her trust of the institution to make a decision about participating in this particular study would be using a legitimate decision-making shortcut. Because community members’ trust of the institution is warranted in such a situation—that is, because the institution actually does what the community members trust them to do and because this particular study actually does what the institution generally does—the researcher's leveraging of that trust would not make their decision-making process worse. Potential participants’ trust in the research institution would be a valid heuristic for good decision-making, so the researcher's use of this trust would not disrespect their autonomy.

7. What should researchers do?

It is often advantageous for a researcher to design her recruitment and consent processes in ways that make potential participants more likely to enroll. She may be confident that her strategies do not involve threatening potential participants and so she is not in danger of coercing them. Nonetheless, the researcher may be concerned that her methods are manipulative and therefore unethical.

Our analysis suggests several steps that concerned researchers should take when they design their enrollment processes. First, they should examine the effects of their use of knowledge about the decision-making processes of potential participants. If their use of this knowledge is making the decision-making process better—as when potential participants are persuaded by a clear presentation of the facts or are being encouraged to use a valid heuristic—then they will not thereby disrespect people's autonomy.

If, however, the recruitment process is designed to encourage people to enroll in a way that is likely to make their decision-making process worse—by deceiving them about facts relevant to their decision, by causing them to make a decision based on immediate desires instead of considered preferences, or by illegitimately changing their choice situations— then the researchers are engaged in manipulation. If the manipulation involves deception about facts that are relevant to a potential participant's decision to join a trial, then the manipulation will render consent invalid. In general, therefore, researchers should not engage in deceptive manipulation. However, if the manipulation does not involve deception about the enrollment decision, then further analysis is required. The researcher might leverage someone's weakness of will to motivate him to join the trial, thereby using motivational manipulation. She might illegitimately change a potential participant's circumstances such that joining the study becomes the most reasonable choice, thereby using circumstantial manipulation. If the researcher uses these types of manipulation to motivate participants to join a trial, she will not prevent potential participants from giving valid consent. However, this does not mean that the manipulation is ethically unproblematic. Since manipulation is still a pro tanto wrong, its use must be the least bad way to achieve a sufficiently valuable outcome—for example, great benefits to participants or the production of research results that are extremely socially valuable.

This framework can be illustrated by applying it to the case with which we began. Dr Khan is not threatening his patients and he is not deceiving them about the study. It seems unlikely, however, that capitalizing on patients’ propensity to agree to physicians’ requests will be making their decision-making better. Rather, by playing on potential participants’ desires to please high status individuals, Dr. Khan would be engaged in motivational manipulation. Moreover, it is hard to imagine that enrolling this group of people in a trial of a hypertensive medication is both important enough to justify manipulating them into it and requires such manipulation in order to get them to consent. If, for example, trial participation would provide substantial individual benefits to participants, say because they would receive ancillary care that they could not otherwise afford, then we might expect that a less manipulative presentation of the facts would suffice for enrollment. We therefore judge that the recruitment strategy proposed by Dr. Khan's colleague is unethical.

Table.

Forms of Influence

	Respects Autonomy		Disrespects Autonomy
Motivational method	Altering perceptions	Altering options	Altering perceptions		Altering options
Form of Influence	Persuasion	Offers	Deceptive manipulation	Motivational manipulation	Circumstantial manipulation	Coercion