Social science research often engages human subjects, about and from whom we collect data.
Moreover, by its nature, experimental research is interventionist.
Field experiments seek to generate real-life impacts in society, political processes, and economic outcomes. That is, experimenters are humans changing the lives of other humans — hopefully for the better, but usually without a direct request from those whose lives are being changed.
We have ethical responsibilities as researchers and human beings.
We must take great care to weigh the potential benefits of the knowledge to be gained and short and long-term improvements in the lives of individuals and communities participating in the research against the potential risks of harm to those individuals and communities.
This is not so easy.
Whether an outcome is good or bad may depend on one’s perspective, making the risk-benefit balance sometimes difficult to assess.
We are prone to significantly overestimating the benefits of knowledge — so we must be cautious and have outside checks.
A better life for some people may imply a worse life for others.
An Institutional Review Board (IRB) or other Research Ethics Committee may review your plans for direct interactions with human subjects for data collection.
Governments may have their own rules and procedures. For example, the US federal government has the Federal Policy for the Protection of Human Subjects. Groups like the OES or USAID apply this rule to their own research.
EGAP has developing COVID-related guidelines for research it supports to protect the safety of subjects, communities, enumerators, and other research staff.
There is no central authority that determines whether your interventions are ethical in a broader sense.
It is always worth putting yourself in the shoes of your research subjects — both those whom you think will gain and those whom you think will lose from any given intervention.
Beyond basic human subjects protections reviewed by research ethics committees like IRBs, it is up to the larger community of researchers to develop and enforce standards.
Standards for the protection of research subjects were developed around concerns about their direct interactions with researchers in the course of data collection.
Some infamous research that abused their subjects through the mid-20th century:
Researchers must respect subjects’
Privacy
Autonomy
The default is that researchers should first obtain the informed consent of subjects. This requires that subjects have the:
Capacity to consent to participation in the study
Freedom from coercion in deciding whether to participate
Comprehension of risks and benefits of the research
Freedom to withdraw from the study at any time
A good general rule: Think about the consent process from the subject’s perspective.
Certain people — children, prisoners, others in vulnerable positions without power — may not be, or feel, able to understand the risks and benefits or to refuse participation.
If the study can be meaningfully conducted with a less vulnerable population, it should be.
We require higher standards of potential benefits to the vulnerable populations being studied.
The default is anonymity (with all identifying information destroyed after the study), as it provides the most protection to subjects.
Sometimes anonymity is not possible. But note that researchers’ promises of confidentiality to subjects can be undone by court orders.
For example: Boston College and oral history tapes of Northern Ireland’s Troubles (https://www.bbc.com/news/uk-northern-ireland-27238797).
In the US and many other countries, there are formal regulations requiring researchers to minimize risks to subjects and to balance generalized knowledge to be gained against risk to subjects.
There is a formal (standard) application to an review board or committee for approval of research with human subjects.
This is not optional. You cannot interact with human subjects before approval.
The board may deem your research to be exempt, but it is the board that decides this (not you!).
You are required to report “adverse events” to the board.
Description of the subject population, how they will be recruited, how they will be compensated (if at all), research context, and what you will do.
Consent form and standard required elements to explain the research to subjects.
The data collection instrument (survey questionnaire, etc.).
Statement of anticipated benefits and risks, including how you will disseminate results.
How subjects will be able to ask questions and report any issues to the review board and/or to you in real time.
How data will be securely stored and subjects’ privacy protected; how identifiers will be destroyed; what data will be preserved.
Universities and research institutes or organizations (like IPA) often have their own ethics review boards.
In some countries, the ethics review board will be housed in a ministry of technology/science and be more used to dealing with medical research than social science research.
Be prepared to explain yourself a bit more. Perhaps plan a coffee meeting to contextualize social science research to a member of a medical review board.
Leave plenty of time to go back and forth with the ethics review board in case there are questions about your application and you need to make changes to your study.
Some funders will require that researchers get ethics review board approval at the home institutions of all researchers on the project, not just one.
You will not be able to publish in academic journals with human subjects data collected without ethics review board approval or exemption.
Hong Kong – Cantoni et al. (2019) provided information to university students about others’ plans to participate in protests to see how it affects their own protest participation.
Kenya – Coville et al. (2020) with the Nairobi City Water and Sewerage Company tested two approaches to increase payment rates. One arm included official disconnection notices, followed through in case of non-payment.
Those who may be affected by the study may not be the same as the set of people who interact directly with the researcher. There are more stakeholders than just the people you interact with.
Interventions in social, political, and economic processes could change who has power, which has impacts beyond your research.
Humility: Researchers themselves may be unaware of potential harms. How can we become aware of them?
Do not use interventions that we expect would harm subjects relative to what would happen without the intervention (“standard of care”).
Do not involve more people than necessary for the research.
Do not make the intervention stronger than necessary for the research.
Do not provide false information. Avoid deception.
Partnerships: You should report your involvement in the design of the intervention.
Provide a way for people to report concerns and potential harms during the study.
Collect data on potential harms. End a study early if there is an accumulation of evidence of harm that would not be outweighed by the potential benefit.
Report important outcomes that can inform programming as soon as possible.
Collect and store data in a way that protects the identity of your subjects.
Informed consent requires explaining the goals of the study. However, the research may be undermined by informing subjects that they’re being studied or making one’s treatment status apparent.
Can research be designed in such a way as to make informed consent possible?
Informed consent. How? Whose? Who are the relevant stakeholders?
EGAP Research Principles, 2011. (Link)
Cantoni, D. et al. 2019. “Protests as Strategic Games: Experimental Evidence from Hong Kong’s Antiauthoritarian Movement.” Quarterly Journal of Economics 134(2): 1021-1077.
Coville, A. et al. 2020. “Enforcing Payment for Water and Sanitation Services in Nairobi’s Slums.” NBER Working Paper 27569. (Link)