Bioethics ISSN 0269-9702
Volume 11 Number 5 1997
INDUCEMENT IN RESEARCH1
MARTIN WILKINSON AND ANDREW MOORE
ABSTRACT
Opposition to inducement payments for research subjects is an international
orthodoxy amongst writers of ethics committee guidelines. We offer an argument
in favour of these payments. We also critically evaluate the best arguments we can
find or devise against such payments, and except in one very limited range of
circumstances, we find these unconvincing.
The guidelines for many research ethics committees state that
payment to subjects2 should not be so high as to amount to an
inducement. Here are some samples:
[T]he participants' consent must be voluntary and not influenced
by financial reward.3
Volunteers may be paid for inconvenience and time spent, but such
payment should not be so large as to be an inducement to
participate.4
The IRB [Institutional Review Board] should review both the
amount of payment and the proposed method and timing of
1
We are grateful to Jo Asscher, Andrew Brien, Alison Douglass, Tim Mulgan,
Nicola Peart, Andrew Williams, the members of the Southern Regional Health
Authority Ethics Committee (Otago), two referees, and participants in the
University of Otago's Philosophy Department Seminar, Bioethics Research Centre
Journals Club, and Bioethics Summer School 1996, for helpful comments.
2
We prefer the term `subjects' to `participants'. The latter term is ambiguous
between researchers and research subjects.
3
Health Research Council of New Zealand (HRC), Guidelines for Ethics
Committees, 6.3.2.1 (iii). The HRC Ethics Committee is the accrediting body for
many ethics committees in New Zealand.
4
National Health and Medical Research Council of Australia, NHMRC
Statement on human experimentation, 1992, 13.
ß Blackwell Publishers Ltd. 1997, 108 Cowley Road, Oxford OX4 1JF, UK
and 350 Main Street, Malden, MA 02148, USA.
374 MARTIN WILKINSON AND ANDREW MOORE
disbursement to assure that neither are coercive or present undue
influence.5
Subjects may be paid for inconvenience and time spent, and should
be reimbursed for expenses incurred, in connection with their
participation in research; they may also receive free medical
services. However, the payments should not be so large or the
medical services so extensive as to induce prospective subjects to
consent to participate in the research against their better
judgement (`undue inducement').6
The quotations just given suggest that there is an international
orthodoxy among writers of research ethics committee guidelines
against inducement payments to subjects. We critically examine this
orthodoxy in this paper, and our conclusions are generally sceptical
and unorthodox.7
After setting out the scope of our inquiry, we raise in Section 1 some
difficulties for the orthodoxy, and we present a prima facie argument
that inducement is a good thing. Section 2 begins our critical
examination of the orthodoxy, starting with the guideline writers'
claim that inducement undermines voluntary decision-making and
consent. Additionally, phrases in the guidelines such as `against their
better judgement' suggest that inducement is considered bad for
subjects. In Section 3, we examine arguments that inducement leads
subjects to act contrary to their own welfare. We group other
objections to subject payments under the claim that they are
exploitative, and the claim that they introduce bias into subject
selection. We discuss arguments for these claims in Sections 4 and 5,
respectively. Our paper limits itself to arguments for the orthodoxy
which appeal to features of the individual research studies in which
5
Food and Drug Administration, FDA Information Sheet: Payment to Research
Subjects, Rockville, United States, 1988.
6
Council for International Organizations of Medical Sciences, International
Ethical Guidelines for Biomedical Research Involving Human Subjects, Guideline 4. The
orthodoxy is very similar in Britain. See Manual for Research Ethics Committees,
compiled by C. Foster, Centre of Medical Law and Ethics, Kings College, London,
December 1994. Section II.49 of this document quotes numerous statements against
inducement payments to research subjects from the guidelines of the Department of
Health, the Royal College of Physicians, the General Medical Council, the
Association of British Pharmaceutical Industries, the British Psychological Society,
and the National Union of Students.
7
It is hard to reconcile rejection of inducements with acceptance of payment for
inconvenience, time spent, and out-of-pocket expenses, since these all increase the
probability of participation. Perhaps the guideline writers rely on a distinction
between due and undue inducement. If so, the rest of our discussion may be treated
as an inquiry into whether there can be any undue inducement in research.
ß Blackwell Publishers Ltd. 1997
INDUCEMENT IN RESEARCH 375
inducement payments are offered. We hope to address elsewhere
arguments which make claims about the broader social and research
consequences of these payments.
Research sponsors sometimes offer researchers cash inducements to
recruit subjects, or to complete studies to specified standards.
Important ethical issues arise about these practices, but they are not
our topic here. Our question is whether researchers should be allowed
by research review bodies, and by ethics committees in particular, to
offer cash inducements to competent people to act as research subjects.
From now on, we shall just call these payments `inducements'. The
focus on inducements sets aside difficult issues concerning invitations
to participate in therapeutic research, which might in some cases also
be inducements to subjects. We leave aside in-kind inducements too,
such as offers of free treatment or reduced waiting time. We also leave
aside payments to incompetent research subjects, because these raise
substantial additional issues.8 Most of our examples are medical in
nature, but we do not intend our conclusions to be limited to medical
research. Rather, they apply to all types of research which ethics
committees are asked to approve. This includes research in medicine,
nursing, psychiatry, psychology, education, social science,
management studies, and so forth.
It might help if we indicate briefly at the outset why we find the
anti-inducement orthodoxy peculiar. Given that people are allowed
to volunteer to be research subjects for no pay, it seems odd to deny
them the opportunity to volunteer for pay. Asymmetrical treatment
of acting as a subject for no pay and acting as a subject for payment
needs to be justified. This point is worth emphasising. A common
picture of inducements is one in which people who are desperately
short of money sign up for dangerous and dubious experiments. It is
no part of our aim to justify inducement in these cases. We consider
only research which ethics committees would allow to proceed, were
the subjects not paid. Our question is this: if people are allowed to
volunteer for no pay, why not for some pay?
1. A CASE FOR INDUCEMENT
Consider the following argument for allowing inducements. Some
researchers would find it worthwhile to pay inducements in order to
attract enough subjects. Those who would accept this reward would
not do so unless it were worth it to them. As a result of offering the
reward, the researchers get the subjects they want. As a result of
8
We also leave aside the question of what should be done, in the context of
inducement, when subjects wish to withdraw during a study.
ß Blackwell Publishers Ltd. 1997
376 MARTIN WILKINSON AND ANDREW MOORE
participating, the subjects get the reward they want. Both are better
off. No one is worse off. Inducement is thus a good thing.
This seems to us to be a good argument, which at least makes a
prima facie case for inducement. It has the same structure as an
argument justifying wages for work, or any other market transactions.
Many people would not work if they were not paid; in that sense
wages are inducements. Few people think that, as a result, it is wrong
to offer wages. Those who do have concerns about the existing wage
system usually object that wages are too low, not that they are too
high, or that they are offered at all.
2. INDUCEMENT AND CONSENT
Those who oppose inducements usually concede something like the
prima facie case we have just made, but then claim that for some reason
it does not constitute a persuasive case for inducements. The most
prominent objection in ethics committee guidelines is that
inducement is likely to invalidate informed consent, particularly
when subjects are poor.
Ruth Faden and Tom Beauchamp imagine a case in which a
woman in a low-paid job with five children desperately needs funds,
which she can get in exchange for undergoing various risky and
uncomfortable medical experiments.9 They think that in this case
the woman's autonomy is infringed upon. Paul McNeill tells us that
his daughter was contemplating volunteering for drug trials, since
she had run out of money. He sent her what she needed, apparently
in part because he thought her consent invalidated by inducement.10
At first sight, the idea that inducement undermines consent is
surprising. People receive inducements all the time to do things they
otherwise would not do, such as parting with their goods or working
under particular conditions for particular employers. There is no
suggestion in the vast majority of these cases that their being paid
undermines the voluntary nature of their actions. We shall argue that
first appearances are correct. There is no good reason to think that
inducements cause subjects to act non-voluntarily, or that they
otherwise invalidate subject consent.
Freedom, autonomy, consent
There is some evidence that inducements cause an increase in the
9
R. Faden and T. Beauchamp, A History and Theory of Informed Consent, New York:
Oxford University Press, 1986, p. 358.
10
P. McNeill, The Ethics and Politics of Human Experimentation, Cambridge:
Cambridge University Press, 1993, p. 176.
ß Blackwell Publishers Ltd. 1997
INDUCEMENT IN RESEARCH 377
proportion of poor subjects.11 We think this evidence is inconclusive,
but we concede it for the sake of argument. We argue, however, that
this does not show that inducements would undermine the consent of
even badly off subjects. Those who believe inducement does
undermine consent must explain why this happens for research
volunteers when it is not plausible that it happens in other cases of
market transactions, such as when one desperately needs to sell some
of one's property.
Desperate need is not sufficient to undermine consent. This point
can be overlooked, if consent is thought to rest on freedom rather than
autonomy. Skating over many complications, a person is autonomous
when she decides what to do for herself. This requires her to have a
sense of what she wants, which she cannot if, say, she is comatose.
Furthermore, her autonomy would be undermined if others imposed
their will on her. This could be done by manipulation or by coercion.
Freedom requires the presence of more than one acceptable
option.12 If one will starve unless one takes the only job available
(pleasant or not) then one is unfree to refuse. We shall presently show
that this does not rule out the morally important consent valued by
ethics committees. Even if people are capable of consent, however, is
it not morally important that they might be unfree to refuse
inducements? It is, but on any view of freedom, denying people the
option of taking inducements reduces their freedom, since it removes
an option that they prefer to the alternatives.
Let us now make good our claim that desperate need ö unfreedom
ö does not, in and of itself, undermine consent. Our argument
depends on the doctrine of informed consent. If the sole alternative
to death is some lifesaving treatment, then one is unfree to turn it
down, but this does not rule out autonomous choice of the treatment.
All the features of autonomous choice might be present: careful
deliberation, correct understanding of the options, no manipulation,
and so on. If informed consent is possible, despite the dire choice one
faces, it cannot be because one is free to refuse the treatment. It must
be because one can nonetheless act autonomously. We conclude that
doctrines of informed consent aim to protect autonomy, not
freedom.13
11
See the evidence cited by McNeill, p. 176.
12
The claim that freedom requires a plurality of acceptable options is
controversial. For its denial, see H. Steiner, An Essay on Rights, Oxford: Blackwell,
1994, pp. 10^21. We adopt this controversial claim because it generates more
difficulties for our argument than does its rejection.
13
See, for example, Faden and Beauchamp, pp. 344^5 and, for exhaustive
discussion of the subtleties, J. Feinberg, Harm to Self, New York: Oxford University
Press, 1986.
ß Blackwell Publishers Ltd. 1997
378 MARTIN WILKINSON AND ANDREW MOORE
If consent protects autonomy rather than freedom, then it is a
mistake to infer an absence of informed consent in those who cannot
afford to turn down payments. The mere fact that they are badly off
and have no acceptable alternative does not establish the absence of
either autonomy or proper informed consent.
Coercive offers
If badly off people were in some way coerced into participating as
subjects, then their autonomy would be infringed upon and their
consent invalidated. Coercion is paradigmatically a case of the denial
of autonomy, since it consists in the deliberate imposition of one
person's will on another. However, coercion usually takes the form of
threats, which restrict people's options. Inducements are offers, not
threats, and they expand people's options.
Can there be coercive offers?14 In our view, it is a necessary
condition of an offer's being coercive that the offerer is also responsible
for the bad circumstances of the offeree. For example, if we poison you
and then offer to provide the only available antidote in exchange for
your stamp collection, that is coercive. If you are poisoned in a way for
which we are not responsible, and we make the same offer, that is not
coercive.15 Our argument here depends on the idea that coercion
consists in the deliberate imposition of one person's will on another.
Let us re-examine our earlier case of the medical patient who gave
uncoerced informed consent. With the exception of the lifesaving
operation, this person's options were all unacceptable, but that did
not make her consent coerced. Nor did the fact that her doctor was
resonsible for her one acceptable option do so. If coercion requires
the deliberate imposition of one person's will on another, and if
making available the only acceptable option(s) is not sufficient for
this, then coercive offers must involve something else. In our view, a
person who makes a coercive offer must not only make available the
victim's only acceptable option(s), but must also be responsible for at
least one of the unacceptable options. Only then is the coercer
responsible for the circumstances in which the victim has no choice
but to surrender her or his will.
The argument we have just made has clear implications for consent
in the context of inducement. As long as those making an offer are not
responsible for the circumstances of the potential subjects, their offer is
not coercive. A final empirical claim: researchers are very rarely
14
For a summary of the large and rather baroque literature on coercive offers,
see Feinberg, ch. 24.
15
It might still be an exploitative offer. We discuss exploitation in Section 4.
ß Blackwell Publishers Ltd. 1997
INDUCEMENT IN RESEARCH 379
responsible for the financial situation of subjects.16 We conclude that
inducement need not be coercive and that consent in this context to
act as a research subject cannot be invalidated by the claim that it is
the acceptance of a coercive offer.
We have argued in this section that it is implausible to think that
inducement invalidates consent. One argument for the contrary view
relies on the claim that desperate need invalidates consent. We argued
that this is false. Another argument depends on the notion of a
coercive offer. We argued that that notion is largely irrelevant to
inducement in research. We have found no persuasive consent-based
case against inducements.
3. INDUCEMENT AND WELFARE
Can inducements be rejected on the grounds that they damage the
welfare of potential subjects? The mutual gain argument offered
above suggests that they cannot. Altruistic motives aside, why should
people accept inducements unless they judge themselves better off for
receiving them? In the absence of a good answer, we should expect
inducements to enhance the welfare of subjects, not damage it. One
might oppose this conclusion by claiming, paternalistically, that
subjects choose against their own best interests. We discuss this claim
below. But first we examine a non-paternalistic argument which holds
that people benefit from not being in a position to accept
inducements.
The non-paternalistic argument
The non-paternalistic argument claims that poor people are better off
not having the freedom to participate in research for reward, because
they could otherwise be forced to use that freedom against their
interests. An analogy here could be the refusal to allow people the
option to sell themselves into slavery. If people had that freedom then
perhaps their creditors or others with power over them could force
them to use it.
This argument is not paternalistic. It does not suppose that people
need protection from themselves, or that we should restrict their
autonomy to protect their autonomy. It holds instead that a denial of
freedom is a way of helping people get what is better for them, because
16
Faden and Beauchamp bizarrely overlook this point, and their own freedom/
autonomy distinction, in their confusing discussion of welcome and unwelcome
offers. That discussion is an attempt to salvage their view that inducements
undermine the autonomy of the poverty-stricken woman mentioned earlier. See
their pp. 358^9.
ß Blackwell Publishers Ltd. 1997
380 MARTIN WILKINSON AND ANDREW MOORE
the loss of freedom gives them a bargaining advantage over those who
want them to act in a way that is worse for them. But it is difficult to
evaluate this argument. It points out a truth, namely that sometimes
it is prudent to want not to be free. But it is also sometimes perfectly
prudent to want to be free. The argument merely points out a general
disadvantage of freedom. If it succeeds in ruling out voluntary
slavery, that may be because slavery is so bad. But in general, being
a subject in research is not so bad, whether or not one is paid for it. It is
at least as good as engaging in risky work, for instance, which few seek
to prohibit. In short, the non-paternalistic argument against
inducement fails.
Paternalism
Consider the following paternalistic argument against inducement. If
offered inducement, there is a risk that some subjects would volunteer
against their best interests. If there is this risk, then inducement should
not be allowed. So it should not be allowed.
The first premise of the paternalistic argument is that if offered
inducement, there is a risk that some subjects would volunteer against
their best interests. They might do so if they underestimate the risks of
volunteering, overestimate the benefits of being paid, or volunteer
despite their accurate assessment that doing so is against their
interests. For purposes of the paternalistic argument, it does not
matter whether researchers or subjects are responsible for these
failures.17
It is a difficult empirical question whether or not subjects tend to
act imprudently. Ethics committees should certainly not simply infer
from the claim that it is imprudent for people in their own generally
comfortable middle-class circumstances to volunteer, to the
conclusion that it is imprudent for research subjects to volunteer. This
inference is particularly unreliable for poor people, for whom
volunteering might be a perfectly prudent use of their time, when
compared to their alternatives. Still, it is hard to deny that if offered
inducements, some subjects might volunteer against their own best
interests.
Consider now the second premise of the paternalistic argument, the
claim that if there is an inducement-generated risk of some subjects
volunteering against their best interests, then inducement should not
be allowed. A major initial problem for this claim is that there is also a
17
In their discussion of the infamous Tuskegee syphilis experiment, Faden and
Beauchamp seem to suggest that the researchers used money as a blind to persuade
would-be subjects not to consider the matter properly. See their pp. 165, 356.
ß Blackwell Publishers Ltd. 1997
INDUCEMENT IN RESEARCH 381
risk of some subjects volunteering against their best interests even if
they are not paid. If it is not to yield a prohibition of all research
involving human subjects, the paternalistic argument must
consequently claim that, compared with a no-inducement regime,
an inducement regime is one in which more people volunteer against
their best interests, and/or people act more against their best interests,
and these differences are enough to justify paternalistic intervention.
Why should one believe these claims? The fact that some people who
would not otherwise volunteer would do so if offered inducement does
not establish these claims, for the introduction of pay might make
volunteering worthwhile for those people. Without a much better
defence of the claim we have outlined, the paternalistic argument
against inducement is a non-starter.
There are, of course, also general ethical worries about the
legitimacy of paternalism. In our view, even the strongest arguments
for paternalism justify that approach to inducements only if the effects
of subjects volunteering against their best interests would be harmful.18
Being interviewed for a dull and badly thought out management
project, for example, might not be worth the sum offered, contrary to
what the subjects originally thought, but their mistake would not
harm them. Another difficulty is that disallowing inducement to
protect some people harms others who do not need protection and
would like to be induced. To make paternalism acceptable, it would
need to be shown that the benefits to those who might be harmed in
the absence of paternalistic intervention have greater ethical
importance than the benefits to those who will not.
The difficulties we have just outlined would face even a successful
paternalistic argument. They suffice to show that not all inducement
could be prohibited on paternalistic grounds. Many readers might
want to go further, and hold that any paternalistic approach is
entirely unjustified. But we think there probably are some defensible
cases of paternalism, and that it is important to show in a more fine-
grained way that these fail to establish a case against inducement.
One defence of the paternalistic argument proceeds by way of
analogy from certain claims about the duties of doctors. It could be
argued that doctors have a duty not to offer treatment if they think it
is clearly against the interests of their patients, whatever the patients
themselves think. Suppose, for example, that a patient mistakenly
thinks only a major and dangerous operation can cure her condition,
although a simple and entirely harmless pill would in fact do so. Some
would argue that her doctor has a duty not to operate on her, whether
or not there is a duty to tell her that it is possible to do so. By analogy,
18
See Feinberg, pp. 118^121.
ß Blackwell Publishers Ltd. 1997
382 MARTIN WILKINSON AND ANDREW MOORE
one might think ethics committees have a duty not to allow
researchers to offer subjects the opportunity to make money out of
research if they think it is clearly against their interests to do so, no
matter what the subjects think.
The argument we have just sketched for the paternalistic duties of
doctors might in itself be defensible. One reason is that patients
sometimes want dangerous operations, and doctors have expertise on
these matters which patients generally lack. Another reason is that in
such cases, the doctor is likely to know a lot about the patient's
relevant circumstances. Neither analogous claim seems to be true of
ethics committees. Firstly, it is not easy to argue that there is danger
to paid subjects, if there is no danger when subjects in those same
studies are not paid.19 Secondly, any justifiable medical paternalism
is undertaken in the light of highly individualised information about
the patient, but in general, ethics committees can have no such
knowledge of individual subjects. We conclude that even if
paternalism can be justified in the medical case, it does not follow that
ethics committee paternalism is justified concerning inducement.
Friends of ethics committee paternalism about inducement might
now move to a different analogy. There are cases in which the state
paternalistically requires health insurance or retirement
contributions from citizens. It does this not on the basis of knowledge
about individuals, but on the basis of general knowledge that people
tend imprudently to discount the future. Might ethics committees
require researchers not to pay subjects for analogous reasons? We
think not. It is uncontroversial that ill health and poverty in old age
are bad, but it is controversial that paid participation in research is in
itself bad. Secondly, the evidence that people imprudently discount
the future is relevant to state paternalism about health and retirement
funds but irrelevant to research, except in the very rare cases in which
bad consequences of study participation are likely to occur long
afterward. We know of no other evidence of imprudent general
human dispositions which would make the case of inducement
analogous to the case of state compulsion concerning health or
retirement funding.
In sum, we have found no clear welfare-based case against
inducement. The non-paternalistic argument is at best inconclusive,
and the paternalistic argument faces major problems. We have found
no convincing argument that research subjects are not the best judges
of their interests in this setting, or that they will not act on those
interests.
19
For one special case in which this argument can perhaps be made, see Section
5.
ß Blackwell Publishers Ltd. 1997
INDUCEMENT IN RESEARCH 383
4. INDUCEMENT AND EXPLOITATION
It is sometimes thought that paying poor people, and perhaps anyone,
to act as research subjects is exploitative. The exact nature of this
concern is unclear, mainly because the idea of exploitation is unclear.
On one account, people are exploited when they do not receive the
full value of their labour product. We doubt that exploitation in this
sense need be wrong.20 But even if it is wrong, the solution in the case
of research is to require bigger inducements, not to prohibit them.
This variant of the argument from exploitation backfires on
opponents of inducement.
Allen Wood has recently argued for the following alternative
account of exploitation, which is well suited to our discussion:
When people, or something about them, are its objects,
exploitation consists in the exploiter's using something about the
person for the exploiter's ends by playing on some weakness or
vulnerability in that person.21
On Wood's account, an exploitative offer actually increases the
exploitee's options, and also benefits that person, usually by more
than it benefits the exploiter. Wood also argues that exploitation is
not necessarily coercive, and that exploitees might well give their fully
voluntary consent to it, and have acceptable alternatives.22 Despite
all this, he argues that exploitation is bad:
Proper respect for others is violated when we treat their
vulnerabilities as opportunities to advance our own interests or
projects. It is degrading to have your weaknesses taken advantage
of, and dishonorable to use the weaknesses of others for your ends.23
Let us accept Wood's account of the nature and moral badness of
exploitation. Who might be exploited if inducement is allowed in
research? The only plausible candidates are those who would not have
volunteered but for the money and, within that group, those who need
the money. Generalising somewhat, we shall call these people `the
poor'.24
20
For valuable and sceptical discussion of the idea of exploitation as theft of
labour product, see P. van Parijs, Real Freedom For All (Oxford, 1995), ch. 5; and G.
Cohen, Self-Ownership, Freedom, and Equality (Cambridge, 1995), ch. 6.
21
A. Wood, `Exploitation', Social Philosophy and Policy, 1995, p. 147.
22
Wood, pp. 147^9.
23
Wood, pp. 150^1.
24
Note that exploitation is avoided if the non-poor are the only subjects paid.
This can be done by making poverty an exclusion criterion for research subjects.
We assume, however, that this is objectionable on other grounds.
ß Blackwell Publishers Ltd. 1997
384 MARTIN WILKINSON AND ANDREW MOORE
An ethics committee might have agent-neutral concerns about the
exploitation of subjects, and in particular, reasons which hold
regardless of whether or not it participates itself in exploitation. The
committee might in addition or alternatively have agent-relative
concerns, and in particular, reasons concerning its own participation
in exploitation. For example, committee approval of research
protocols which involve exploitative inducements might be regarded
as ethics committee participation in the exploitation of subjects.
The existence, nature, and strength of agent-relative and agent-
neutral reasons are matters of controversy.25 We shall sidestep these
general issues, simply by assuming that reasons of both kinds exist,
and by considering exploitation-based arguments of both sorts against
inducement.
Consider the agent-neutral reasons an ethics committee might
have to prevent exploitation of subjects, whoever the exploiter might
be. Let us simply assume straight away that inducement of the poor to
be research subjects is bad exploitation, and that ethics committees
have agent-neutral reason to disallow it. Even given these
assumptions, we still do not have a persuasive agent-neutral case
against inducement. One reason arises from the fact that on Wood's
account, exploitation is a pervasive feature of the social life of human
beings. As he notes, it `. . . is endemic to all . . . transactions whose
terms are significantly affected by the fact that the participants are in
unequal bargaining positions'.26 In virtually all their exchanges with
people more powerful than themselves, including their alternatives to
accepting inducement and their acceptance of invitations to be
unpaid research subjects, the poor are exploited. Those who wish to
make an agent-neutral argument from exploitation must
consequently argue that exploitation comes in different amounts,
and that there is more of it in inducement than in the alternatives
open to the poor. As far as we know, they have yet to make any such
case.
The following argument suggests that inducement is in fact less
exploitative than are the alternatives open to the poor. Suppose the
25
For a sample of the large literature on these topics, see T. Nagel, The View From
Nowhere, New York and Oxford: Oxford University Press, 1986, especially ch. IX;
S. Scheffler, The Rejection of Consequentialism, Oxford: Clarendon Press, 1982; D.
Parfit, Reasons and Persons, Oxford: Clarendon Press, 1984, section 57; S. Kagan,
The Limits of Morality, Oxford: Clarendon Press, 1989; J. Dancy, `Agent-relativity
ö the very idea', in R. Frey and C. Morris (eds.), Value, Welfare, and Morality, New
York and Cambridge: Cambridge University Press, 1993; and C. Korsgaard, `The
reasons we can share', in E. Paul and F. Miller (eds.) Altruism, New York and
London: Cambridge University Press, 1993.
26
Wood, p. 156.
ß Blackwell Publishers Ltd. 1997
INDUCEMENT IN RESEARCH 385
amount of exploitation is a function of the extent to which the
preferences of the exploitees are fulfilled. By hypothesis, paid subjects
prefer this to any available alternative, otherwise they would not
volunteer. If denied that option, they would have to do something
they prefer less. On this measure of exploitation, they would therefore
be even more exploited, and denying them the option of inducement
would increase their exploitation. It follows that an ethics committee
with agent-neutral reason to prevent exploitation of the poor has all-
things-considered reason of this sort to allow inducement.
Furthermore, if ethics committees have agent-neutral reasons for
action, these include not just reasons to prevent exploitation of
subjects, but also reasons to promote subject freedom and welfare. As
we argued above, both sorts of reasons support inducement. In short,
considerations of exploitation do not generate any agent-neutral case
against inducement, and considerations of subject freedom and
welfare generate a case to allow it. Overall, then, there is an agent-
neutral case in favour of inducement.
Turn now to the agent-relative argument from exploitation.
Again, let us simply assume that ethics committees which allow
inducement thereby participate in exploitation, and have agent-
relative reason not to. It is in the nature of agent-relativity that this
reason would be undefeated even if disallowing inducement were to
result in greater exploitation of would-be subjects. We have assumed
that there is such an agent-relative reason. This is something an
opponent of inducement would need to argue, but even if there is such
a reason, it makes no clear case specifically against inducement. One
reason is that on Wood's account, it will often be bad exploitation to
invite the poor to be research subjects, even without inducement. It
follows that there is no distinctive objection to those invitations which
are also inducements. Even if there is an agent-relative argument
from exploitation which singles out inducements, the case is still not
clear, for there are conflicting agent-relative reasons here. If ethics
committees have agent-relative reasons for action, it is plausible that
these include reasons to benefit subjects, and to facilitate their
voluntary decisions. Yet ethics committees which disallow
inducement thereby decide not to benefit subjects financially, and to
rule out volunteering for pay.
In short, if ethics committees have agent-relative reasons for
action, they have harm-related and freedom-related reasons of this
sort to allow inducements, and exploitation-related reasons to
disallow them. To make good their argument from exploitation,
opponents of inducement must resolve this conflict among agent-
relative reasons by showing that the exploitation-related reasons
defeat those which concern the freedom and welfare of subjects.
ß Blackwell Publishers Ltd. 1997
386 MARTIN WILKINSON AND ANDREW MOORE
In summary, an exploitation argument must show: (1) that ethics
committees which allow inducement thereby participate in
exploitation; (2) that this generates an agent-relative case specifically
against inducement in research; (3) that this exploitation argument
defeats opposing arguments from subject freedom and welfare; and
(4) that it also defeats agent-neutral reasons of exploitation, freedom,
and benefit, which largely favour inducement. An exploitation
argument could in principle meet all these conditions, but we have
yet to find any actual argument which does so, and we doubt that we
will find one.
Those who are concerned about the exploitation of research
subjects might understandably still feel troubled. It is appalling that
someone should participate in unpleasant research because, for
instance, she cannot pay her children's medical bills. The fact that
some people are in these situations is appalling, but in our view, the
fact that researchers offer these people inducements is not. The poor
might well be in these situations as a result of fundamental social
injustice, but ethics committee decisions will not make that go away.
They must decide, against this background, whether or not to allow
inducements. If they are concerned to help the poor, it is hard to see
how denying them the option to make money through research helps.
Even if they share instead Allen Wood's view that `solidarity with them
[is] a far more vital achievement than any positive contribution to
their welfare',27 it is still hard to see how denying the poor the option
to make money through research achieves even that goal.
5. SUBJECT SELECTION BIAS
According to the argument from subject selection bias, if inducements
are allowed in research, this will undermine the science and
consequently the ethics of these studies. A more formal version of the
argument is this: (1) inducements bias subject selection, (2) biased
subject selection makes for bad science, (3) bad science is always
ethically bad science, and (4) ethics committees should not allow
ethically bad science; therefore, (5) ethics committees should not
allow inducements.
In our view, the argument from subject selection bias is not
persuasive. To begin with, there is an important scope problem with
(1), the claim that inducements bias subject selection. It is plausible
that this sometimes happens, but it is also plausible that inducements
sometimes reduce or prevent bias. This happens whenever relevant
individuals or groups will not agree to be subjects unless they are paid.
27
Wood, p. 153.
ß Blackwell Publishers Ltd. 1997
INDUCEMENT IN RESEARCH 387
We conclude that since (1) is false of some inducements, it cannot
ground a general argument against them. Subject selection bias
consequently has to be looked at case by case.
Premise (2) of the argument claims that biased subject selection
makes for bad science. This claim inherits the scope problem of
premise (1), and adds a further problem of the same sort. Subject
selection bias is hard to eliminate altogether, and in general, scientists
should look for the least biased method amongst the feasible and
scientifically worthwhile alternatives. Even if inducement always
introduces selection bias, this does not show that it always or even
usually generates bad science. Roughly speaking, the addition of
inducements to a subject selection method turns good science into
bad only if there is a scientifically worthwhile and feasible alternative
study design which is free of inducements, and which generates less
bias.
Finally, consider (3), the claim that bad science is always ethically
bad science. The standard justification for this is that research always
imposes costs on subjects, and these are unjustified unless the science of
the study is good. We agree that research always imposes costs on
subjects, though these are often small. We also agree that these costs
should be justified. One attempt to do so points out that subjects
voluntarily incur these costs. It is plausible, however, that some
subjects volunteer only because they think they are doing something
worthwhile. If the science is poor, and subjects are not paid, then they
are not doing anything worthwhile. If they are paid, however, this
offsets their costs, and if they are paid enough, participation will
actually be a benefit to them overall, even if the science is bad. Any
residual concern about the voluntariness of participation of subjects
who wish only to be involved in scientifically worthwhile studies can
be met by requiring researchers whose studies do not meet this
standard to state this on their subject information sheet.
To sum up: if subjects are sufficiently well paid, then bad science
imposes no net cost on them. Nor need the voluntariness of their
participation be compromised.28 Bad science can thus be ethically
acceptable, especially when subjects are adequately paid for their part
in it.
Three premises of the argument from subject selection bias are
false, at least if they are taken to have the broad scope they need to
make a general case against inducements. We conclude that the
argument as a whole fails.
28
One further argument for the inference from bad science to bad ethics appeals
to the waste of researcher time and money it involves. This issue need trouble only
researchers themselves, and their funding bodies.
ß Blackwell Publishers Ltd. 1997
388 MARTIN WILKINSON AND ANDREW MOORE
Biased subject selection can nevertheless generate serious ethical
problems in one sort of case. We call this the excluded subject case.
The worry is that pay might induce the participation of financially
hard-pressed people who are ineligible to participate under a
study's exclusion criteria. Consider, for example, a poor person with
a heart condition which would exclude her from a pharmaceutical
study, who is tempted to deny its existence so she can secure the
$100 fee for subjects, and thus feed her children. It seems that any
ethics committee which allows inducement makes excluded
participation in her sort of case more likely than it would otherwise
be. Serious injury or even death to one or more subjects might
result. Legal liability and damage to the study's science might also
result, but we shall assume that these are issues primarily for
researchers, their sponsors and their employers, rather than for
ethics committees.
Not all excluded subjects cases are ethically worrying. For
example, studies with no harm-related exclusion criteria are not a
concern. Where there is at least one harm-related exclusion criterion,
ethics committees should determine whether adequate methods are
available, independent of subject self-report, for checking subject
eligibility. If such methods are available, ethics committees should
simply require researchers to use them. No prohibition on inducement
is needed.
That leaves studies with at least one harm-related exclusion
criterion, and no adequate independent method for researchers to
check potential subjects against it. Ethics committees should
probably not allow these studies to proceed at all. If they are
allowed to proceed, on condition that they do not involve
inducement, then excluded subjects might still participate at
significant risk to themselves. They might do so because they are
ignorant of their own excluded conditions, because they overlook
or misunderstand harm-related exclusion criteria, or because they
are disposed to please researchers and to believe they always want
volunteers. Opponents of inducement must argue that ethics
committees take an acceptable risk when they approve studies
without inducement, but with at least one harm-related exclusion
criterion against which researchers have no adequate independent
method to check potential subjects. They must also argue that the
addition of inducement turns these acceptably risky studies into
unacceptably risky studies. As far as we know, opponents of
inducement have not argued for these claims. We do not insist that
the required arguments cannot be made. We conclude merely that
at most, there is a very narrow range of excluded subject cases in
which ethics committees should disallow inducements.
ß Blackwell Publishers Ltd. 1997
INDUCEMENT IN RESEARCH 389
6. CONCLUSION
We have argued that there is an international orthodoxy opposed to
inducement payments for research subjects, and we have presented a
prima facie case against this orthodoxy. In examining arguments for
the orthodoxy, we concentrated on those which appeal to features of
the individual research studies in which inducements are offered, and
we considered the strongest case we could find or devise. No argument
for the orthodoxy based on consent or welfare was persuasive. Nor
were any of the arguments from exploitation. That left only
arguments from subject selection bias, and here, at most one quite
narrow set of circumstances presented difficulties for inducement.
With this single possible exception, then, we have found the
arguments against inducement in research to be unpersuasive.
Department of Political Studies Department of Philosophy
University of Auckland University of Otago
ß Blackwell Publishers Ltd. 1997