Science_-_6_March_2020

sciencemag.org SCIENCE

INSIGHTS | POLICY FORUM

lawyers, who like physicians are often in a
position of power compared with prospec-
tive subjects; the provision of desired legal
services only to study participants (or select
participants); or the constraints on freedom
present in criminal settings.
Concerns about professional power dy-
namics can be addressed in part by em-
phasizing, where true, that participation
in research is voluntary and refusal will
involve “no penalty or loss of benefits to
which the subject is otherwise entitled” ( 10 ).
Nonetheless, potential subjects may feel
pressured when research is the only way to
access a desired intervention, a concern that
has been a persistent point of confusion in
research ethics. This pressure is not coercive
absent entitlement to the desired interven-
tion ( 11 ), and it is not unduly influential if
the risks of participation are reasonable and
not so great as to outweigh the benefits of
the desired intervention ( 12 ). In fact, where
research risks are reasonable and there is
the prospect of direct benefit to subjects, one
might worry less about constraints on volun-
tary consent because research participation
will be objectively reasonable.
This is important because one distinctive
feature of legal practice RCTs is the inher-
ent involuntariness of some aspects of the
legal system, particularly in the context of
criminal justice. Unlike medicine, where
there is an expectation of informed consent
to justify interventions on a patient’s body,
individuals who have been arrested, are
facing prosecution, or have been convicted
have little control over what happens to
them. At least partly in recognition of these
circumstances, U.S. research regulations
demand special protections for individuals
involuntarily confined or detained in penal
institutions ( 10 ), which can also guide re-
search in other legal settings.
To the extent voluntariness concerns per-
sist, further analysis is warranted to evalu-
ate whether securing traditional research
consent is a reasonable expectation for
all legal practice research interventions.
Current norms permit consent waiver only
for minimal risk research that would be im-
practical to conduct if consent were required
( 10 ). This would likely cover RCTs assess-
ing mandatory mediation or online dispute
resolution , for example, but “minimal risk”
is challenging to define and the subject of
intense debate in research ethics, especially
when evaluating practices that may be part
of the existing standard of care, despite re-
maining unproven ( 13 ). And at least some
legal practice interventions will pose greater
than minimal risk, such as a heightened pos-
sibility of incarceration (e.g., for arrest, bail,
and sentencing interventions) or inability to
access government benefits (e.g., for compar-

isons of self-help models versus assistance of

counsel). The response to concerns about

voluntary consent must not be to inhibit le-

gal practice RCTs, however, because without

them, potentially harmful legal interven-

tions may be adopted and potentially benefi-

cial interventions may be dismissed without

either evidence or consent ( 8 ).

Similar questions arise in circumstances

in which it may be infeasible to secure in-

dividual consent, a challenge faced across

medical, policy, and legal practice interven-

tion research. For example, it may not be

possible to study an intervention without

mandatory application to some group, such

as use of a sentencing algorithm to substi-

tute for judicial discretion. Similarly, use of

randomized cluster design can sometimes

preclude individual consent, such as when

different legal services clinics are each ran-

domized to offer only a single approach,

removing choice from each clinic’s clients

unless they seek services elsewhere. It is

important to recognize that these interven-

tions could be implemented without con-

sent outside a research context (and some-

times already have been), which may render

the inability to secure research consent less

ethically concerning ( 2 , 8 ). Nevertheless,

alternative protections may be appropriate.

For example, the data collection required

for research purposes should not unduly

infringe participants’ autonomy and pri-

vacy rights ( 6 ), and research should avoid

disproportionately testing interventions

designed to justify reducing resources allo-

cated to vulnerable populations ( 14 ).

BYSTANDERS

A final concern that may arise for legal prac-

tice RCTs, as well as other types of studies,

centers on their potential impact on both

incidental bystanders and intended third-

party beneficiaries, an understudied area in

research ethics generally ( 15 ). For example,

research interventions regarding bail deci-

sions might be intended to achieve public

safety goals, to advance the interests of ar-

restees, or both. Where third parties are

affected by research but not directly inter-

vened upon, there are further questions of

consent. They may not be individually iden-

tifiable in advance (e.g., the future victim of

an arrestee released on bail due to a study

intervention), making their consent impos-

sible. Even if bystanders are identifiable,

however, it is unclear that they should be

given the authority to withhold consent and

preclude research that offers the prospect

of direct benefit to research subjects and

future individuals like them (e.g., arrestees

that may benefit from a bail decision). Yet,

to the extent bystanders face research risks,

those risks seem worthy of consideration

when assessing a study’s ethical permissi-

bility. Conversely, although we traditionally

consider risks to research subjects, where

research interventions are intended to pro-

duce social rather than individual benefit,

and the intervention does not entail any

breach of a subject’s rights or entitlements,

it is not clear that subjects should be al-

lowed to withhold their consent. Although

IRBs are directed to consider benefits that

extend beyond study participants when as-

sessing a study’s risk-benefit profile, they

have not typically considered risks to by-

standers. This area deserves further atten-

tion in both law and medicine.

CONCLUSION

Although objections to legal practice RCTs

often have been framed in ethical terms re-

lated to randomization and consent, the big-

gest ethical challenge lies in failing to seek

reliable answers to guide the legal profes-

sion. The profession—and IRBs—should

embrace RCTs, while being thoughtful about

their methodological limits and devoting

further attention to unresolved issues, in-

cluding which legal interventions should

be viewed as entitlements; when research

consent should be required; how to address

bystander interests; which RCTs should be

prioritized; when it may be appropriate to

proceed with a promising intervention in the

absence of confirmatory evidence; and how

to best engage with communities that would

be affected by this research. j

REFERENCES AND NOTES

1. D. J. Greiner, A. Matthews, Annu. Rev. Law Soc. Sci. 12 , 295
   (2016).


2. M. Abramowicz, I. Ayres, Y. Listokin, U. Penn. Law Rev. 159 ,
   929 (2011).


3. A. S. Venkataramani, K. Underhill, K. G. Volpp, JAMA 323 , 21
   (2020).


4. World Health Organization, Ethical considerations
   for health policy and systems research (2019); http://www.
   who.int/alliance-hpsr/resources/publications/
   ethical-considerations-hpsr/en/.


5. D. MacKay, Bioethics 32 , 59 (2018).


6. D. MacKay, A. Chakrabarti, Public Health Ethics 12 , 188
   (2019).


7. S. Joffe, F. G. Miller, Nat. Rev. Clin. Oncol. 9 , 230 (2012).


8. M. N. Meyer et al., Proc. Natl. Acad. Sci. U.S.A. 116 , 10723
   (2019).


9. R. A. Crouch, J. D. Arras, Hastings Cent. Rep. 28 , 26 (1998).


10. 45 C.F.R. 46.116 (2018).


11. A. Wertheimer, Coercion (Princeton Univ. Press, 2016).


12. E. J. Emanuel, Am. J. Bioeth. 5 , 9, discussion W8-11, W17
    (2005).


13. S. Joffe, A. Wertheimer, Ethics Hum. Res. 36 , 16 (2014).


14. J. Blustein, J. Policy Anal. Manage. 24 , 824 (2005).


15. S. K. Shah et al., Science 360 , 158 (2018).

ACKNOWLEDGMENTS

Some of Access to Justice Lab’s work is supported by Arnold

Ventures, The Chan Zuckerberg Initiative, The Charles Koch

Foundation, The NOMIS Foundation, and The Robert Wood

Johnson Foundation. Support for elements of this article was

provided in part by The Pew Charitable Trusts. The opinions,

findings, and conclusions or recommendations expressed here

do not necessarily reflect the views of any of the entities listed.

H.F.L. is funded by the Greenwall Foundation as a Faculty Scholar.

10.1126/science.aay3005

1080 6 MARCH 2020 • VOL 367 ISSUE 6482

Published by AAAS