Precarious Times, Professional Tensions: The Ethics of Migration Research and the Drive for Scientific Accountability

Abstract

How should migration scholars navigate tensions between our ethical responsibilities to research participants and growing “open science” calls for data transparency, replication, and accountability? We elaborate a three-step process to navigate these tensions. First, researchers must understand core principles behind open-science initiatives and the mandates of research ethics boards, especially those related to privacy, confidentiality, and protection from harm, and take them seriously. Second, migration researchers must think beyond routinized or mandated procedures to carefully consider the unique vulnerabilities of migrants in their study, which depend on socio-political context. Third, if vulnerabilities are significant, migration researchers should modify (or challenge) procedures elaborated in the name of open science or routinized research ethic board mandates, if inappropriate for their study. We, thus, encourage migration scholars to engage with open-science advocates but also to educate colleagues on migrants’ vulnerabilities and to double-down on data security, including vis-à-vis government authorities, as evolving technologies continue to change research practices.

Researchers across the lab and social sciences are increasingly turning a well-deserved spotlight on scientific accountability in the name of “open science,” emphasizing the need for data transparency, replicability, and pre-registration of analysis plans, in both quantitative and qualitative research (Christensen, Freese, and Miguel 2019; Haven and van Grootel 2019). This attention comes amid high-profile scandals involving fabricated survey data and concerns over “p-hacking,” “HARKing,” or “specification searching” of statistical data to obtain statistically significant results, whether from experiments, population surveys, or administrative records (Nosek, Spies, and Motyl 2012; Young and Janz 2015).¹ In parallel, qualitative researchers engage in persistent debates about the veracity of and selective quoting from ethnographic field notes or in-depth interview data (Jerolmack and Murphy 2019; Lubet 2017). In response, many quantitative and qualitative researchers are doubling-down on “open science,” a movement that UNESCO (2021) defines broadly as making “scientific research and data accessible to all” and which others conceptualize as “greater research transparency and reproducibility” (Christensen, Freese, and Miguel 2019, 4). Regardless of a researcher’s personal view on the open-science movement, its ideals of accessibility and transparency are increasingly embedded in funding rules, institutional procedures, and journal publication practices. Such principles have direct consequences for migration research.

While laudable, we worry that calls for scientific accountability and transparency can enter into conflict with important ethical concerns in migration research. In many countries, migrants and their families are sitting in the political crosshairs. In the United States, for example, immigration enforcement has targeted an increasingly wide range of immigrants (Brotherton and Kretsedemas 2008; Hiemstra and Conlon 2016; Menjívar, Gómez Cervantes, and Alvord 2018). Undocumented migrants, those with liminal legal status, refugees, asylum-seekers, migrants of Muslim faith, lawful permanent residents who use social benefits, and even naturalized citizens all face increasing risks of indefinite detention, deportation, scrutiny, and social exclusion (Akram and Johnson 2001; Armenta 2017; Chacón and Coutin 2017). The targeting and vilification of migrants occur elsewhere, too, from Australia to Hungary, Jordan to South Africa (Human Rights Watched 2020), raising critical questions of how to protect research subjects from harm and to ensure their confidentiality and privacy — central precepts for ethical research. In this context, the long-standing concern about government agencies’ subpoena powers is rekindled and amplified. As immigration enforcement strategies have expanded around the world, safeguards to protect migrants against breaches of confidentiality apply to all forms of data collection and analysis, including research removed from primary data collection. The risks that migrants face today as their data are collected and stored by researchers may equal or even exceed those of formally recognized vulnerable populations (e.g., prisoners or pregnant women).

How should migration scholars navigate tensions between calls for research transparency, replication, and accountability, on one hand, and our ethical responsibilities to research participants, on the other? How should we navigate these tensions as we also take into account evolving technology that is fundamentally changing research practices, whether through access to social media data, computing power that can re-identify individuals from millions of anonymized records, or cloud storage that places sensitive field notes, interviews, or other data at risk of surveillance, including by state enforcement agencies?

In this article, we recommend that migration researchers engage in a three-step process of understanding principles, evaluating vulnerability, and then developing procedures to grapple with scientific accountability and the moral imperatives of ethical migration research. Specifically, we argue that researchers must, first, understand and take seriously the core principles behind the procedures advocated by open-science proponents and those mandated by the research oversight bodies tasked with ethics, such as Institutional Review Boards (IRBs). Below, we delve into these principles and center our discussion on privacy, confidentiality, and protection from harm. Second, migration researchers must think beyond routinized or mandated procedures to carefully consider the unique vulnerabilities of migrants in their study. The particular vulnerabilities that migrants face are often not covered in research methods texts, institutional regulations, or standardized guidelines.² We consider some key vulnerabilities in this article, but depending on the group, place, and time, other vulnerabilities may apply. Third, we argue that migration researchers must modify (or challenge) routinized or mandated procedures (articulated by open-science advocates or IRBs) if they are inappropriate for their study. We provide examples of some of these choices and options in our discussion of upholding principles while modifying procedures. As we argue throughout this article, promoting transparency and scientific accountability is important, but practices in the name of open science can overreach in some circumstances by discounting the protections necessary when studying vulnerable populations.

We caution that modifications to procedures should be undertaken carefully and thoughtfully: the researcher should outline the ethical or scientific accountability principle(s) at issue, the reasons current procedures do not work, given migrant vulnerabilities, and what should be done instead. Such evaluation of research practices should be undertaken by quantitative and qualitative researchers, as well as by those reviewing publication submissions, grant applications, and student projects. Indeed, we call on journals such as International Migration Review to mandate that all authors submit an ethics and transparency paragraph, possibly as an on-line supplement, to promote this deliberative process and to provide models for other researchers.

This article is written in the spirit of advancing public conversation and speaks especially to those for whom migration research ethics is a new or under-explored topic — whether students or early-career scholars, colleagues working in fields dominated by secondary data analysis, or scholars at institutions unfamiliar with IRB norms. The IRB and ethics terrain is evolving rapidly, but as recently as 2016, an examination of research ethics in Europe found extensive gaps and outright ignorance of ethical (or legal) codes among social scientists (Piccio 2016).³ In the United States, even a decade ago, only a single research university of 32 surveyed included immigration status as a possible source of vulnerability in human subjects research oversight (Perry 2011).

For seasoned researchers, we underscore the role that they can play in sharing insights and experiences with students and emerging scholars and, when they review manuscripts or grant applications, in reminding their review board colleagues and supporters of open science of migrants’ vulnerabilities. To this end, we also urge journal editors and authors to better detail the ethical challenges of doing migration research in published work and public presentations, both unanticipated problems and best practices. Such conversations, we argue, should not be relegated only to the methodological appendices of qualitative researchers’ books or to private conversations over coffee at a conference.

To elaborate this process of considering principles, vulnerabilities, and procedures, our article unfolds as follows. First, we provide a quick overview of “open science” precepts and the reasons for them and touch on evolving challenges associated with technology and computing power. Next, we turn to the heart of our discussion: ethics in principle and practice. We take as a starting point core principles of ethical research usually promulgated by North American Institutional Review Boards (IRBs), since they administer standardized rules and regulations that will resonate with researchers across disciplines, methods, and types of data collected. These principles include securing participants’ informed consent, minimizing harms, and ensuring that a study’s overall benefits outweigh the risks for those involved. We consider specific migrant vulnerabilities and offer strategies to tackle them. We aim to model a cross-disciplinary and cross-methods engagement with these issues — one that takes seriously our ethical responsibilities to research participants and to the scientific community. Our goal is to underscore commonalities that migration scholars face and to propose generative conversations, rather than siloed discussions about one method or type of data.

Throughout our discussion, we underscore that IRB rules and procedures offer a lower, not upper, bound to protect migrants. Research ethics must not be viewed as just completing a web-based course or satisfying IRB requirements but, instead, as embedded in practice, teaching, and publishing on migration. Migration scholars in gatekeeping roles as reviewers for funding agencies, members of IRB boards, or book or journal editors must be attentive to ethics in their assessments of research endeavors and educate peers on the possible limits of open science when studying migrants. In evaluating a research project, IRBs are instructed to consider both the probability and magnitude of harm. For migration scholars, such evaluations depend on the research subject and study and on the broader socio-political situation and time period. The risk of deportation, separation, indefinite detention, intense surveillance, and criminalization will likely prolong migrants’ vulnerability as study subjects for the foreseeable future.

The Current Challenge: Open Science, Technological Change, and Migrant Vulnerability

Just as political and social contexts shape migrants’ vulnerability, context informs contemporary discussions over scientific accountability and open science, spanning worries over scholarly fraud, scientific accuracy, and inequities in accessing research. Concerns over scientific accountability are longstanding, but they gained momentum over the past decade through scandal and political attacks against the veracity of scientific data and methods, in both quantitative and qualitative work (Young and Janz 2015; Lubet 2017; Christensen, Freese, and Miguel 2019). For example, in 2015, Broockman, Kalla, and Aronow (2015) made headlines when they charged that a widely reported experimental field study about political persuasion and gay marriage was based on fabricated results. Reflecting on the lessons from this scandal, political scientists Joseph Young and Nicole Janz (2015) advocated that “Each current journal in all of social sciences should establish policies that require data, tools, and processes to be completely open-source upon publication.” Open-science ideas and procedures are also elaborated to address the problem of publication bias, which rewards the publication of original, (statistically) significant, research over replication studies or reporting null results, thereby inflating the prominence of unusual or ungeneralizable findings (Nosek, Spies, and Motyl 2012; Christensen, Freese, and Miguel 2019; Haven and van Grootel 2019). To guard against “p-hacking,” “HARKing,” and similar practices, a rapidly increasing number of psychology, political science, economics, and other journals extend the option, or require, that researchers submit their study rationale, hypotheses, design, and analytic strategy before the study is even begun (Gonzales and Cunningham 2015), while others advocate machine-learning techniques to deal with model overfitting or replication failures (Yarkoni and Westfall 2017).⁴ For some people and institutions, promoting open science is also a way to mitigate knowledge inequalities within academia and across more and less developed countries by, for example, breaking down “pay-walls” to access findings and research data (UNESCO 2021). For all these reasons, some researchers argue that “being an ethical social scientist goes beyond our responsibilities toward study participants” and that practicing open science should itself be an ethical obligation (Christensen, Freese, and Miguel 2019, 11). Migration researchers must take seriously core ideas in the open-science movement.

Open Science: Pre-registration, Transparency, and Replication

We highlight three key open-science ideas. Pre-registration requires researchers to specify their research questions, hypotheses, and data analysis plan publicly prior to observing research outcomes and ideally before collecting data. Under this model, researchers should not deviate from these research plans so as to test their hypotheses or intuitions fairly.⁵ Transparency, for our purposes, refers to providing sufficient information and detail about one’s research design and analytical procedures, such as the statistical modeling or qualitative coding, so that peers can scrutinize the methods and evaluate the soundness of the data, analysis, and conclusions. Standards for replicability go further, expecting researchers to make original data available and to provide sufficient detail about the analytical strategy such that others can replicate the study or apply the same methods in another context or to another dataset to bolster the generalizability of findings. Adhering to these ideas will, from an open-science perspective, help hold researchers accountable to the broader scientific community, to funders, and to the general public.

Calls for open science are often associated with experimental methods or statistical significance testing in quantitative research, but we see examples of the push to increase scientific accountability affecting a range of data types and analytical approaches, including in migration research. For instance, anyone can download the data and analysis code that Hainmueller and Hopkins (2015) used in their survey experiments on “The Hidden American Immigration Consensus” to replicate their analysis.⁶ Quantitative researchers are more likely to publicly archive their data and analytical coding, but migration scholars who use qualitative techniques also do so at times. Sociologist Tomás Jiménez (2017), for example, archived the in-depth interview transcripts informing his analysis of how immigration changes life for nonmigrants.⁷ Beyond calls to archive qualitative in-depth interview transcripts (Lamont and White 2008) and to engage in qualitative pre-registration procedures (Haven and van Grootel 2019), some qualitative researchers — including migration scholars — are also making a push to archive respondents’ attributes and the first-stage qualitative “index” coding of interviews for secondary users to examine (Deterding and Waters 2021). Since qualitative data coding is often poorly explicated in published research, greater transparency around qualitative coding practices could not only improve replicability and transparency but also serve an important teaching purpose (Reyes, Bogumil, and Welch 2021).

Among qualitative field researchers, debates over transparency have included calls to make field notes and in-depth interview transcripts public or available to peers, to share coding or analytic procedures, and to provide significant details about the ethnographic site, including real names and locations (Murphy and Jerolmack 2016; Reyes 2018).⁸ These calls arise in part from a concern about inaccuracies or falsified data (e.g., Lubet 2017). Academic peers who advocate for “unmasking” — that is, revealing the actual name of places, organizations, and (sometimes) people — also believe that the interpretation of qualitative data becomes difficult without sufficient context in terms of place, time, and people involved (Jerolmack and Murphy 2019). Others, such as race and borders scholar Victoria Reyes (2018), note that these decisions should be made on a case-by-case basis. Recognizing the risks of full unmasking, Contreras (2019) proposes semi-biographical disclosure (strategically omitting certain biographical information), partial spatial disclosure (revealing the general geographical area of research), or invitational disclosure (inviting outsiders to meet participants in the field). Such debates about masking and unmasking are particularly relevant for migration researchers because they illuminate how a universal “transparency” edict is unadvisable but also how masking decisions should be evaluated against legitimate transparency concerns.

Indeed, calls for pre-registration, transparency, and replication are increasingly not just suggestions but are becoming embedded in institutions that serve as gate-keepers to publication or grant dollars (Gonzales and Cunningham 2015). Academic researchers at universities and policy-oriented researchers at international institutions such as the World Bank and UNESCO are working to make pre-registration and transparency scientific norms, providing training and small grants for the preparation of datasets, codebooks, analysis code, and other materials for public release.⁹ The open access journal Sociological Science provides detailed submission guidelines for presenting quantitative and qualitative evidence, including asking that “The field site and timing of the fieldwork should be described in sufficient detail that the reader can understand the context” and “Direct quotes should be used only when a respondent’s statement has been recorded verbatim.”¹⁰ Various funders require the public release of data as a condition of grantmaking to ensure transparency and to stretch resources by providing data for other scholars. The Russell Sage Foundation, a US grant maker with a history of supporting migration research, requires a plan for the public release of funded data and documentation.¹¹ In the United States, federal funding agencies such as the National Science Foundation and National Institutes of Health also have open data policies and require data sharing plans.¹² These US open-science practices may well spread to other countries, where scholars will face open data requirements of their own, if they do not encounter them already.

New Frontiers: Technological Change, Computing Power, and Other Challenges

As a public good, archiving data is certainly helpful, especially given limited resources for large-scale data collection projects. However, public data allow anyone access to the information, a growing concern given evolving technologies and computer power, and especially in the context of expanded surveillance technologies for immigration enforcement. Indeed, technological change adds further pressure on migration scholars’ assessment of their ethical obligations. We lack the space to go deeply into this issue, but at a minimum, all researchers — quantitative or qualitative — must be cognizant of the risks associated with electronic data storage and the possibility that others, including governments, can access data.

Discussions of ethics and open science should, therefore, also include researchers who conduct secondary data analysis and who may not consider issues of harm, confidentiality, consent, and privacy in depth because they do not collect their own original data. The political and scientific controversy over the 2020 US Census, first around inclusion of a citizenship question and then over then-President Trump’s call to estimate and remove counts of unauthorized residents from the census tally, is a case in point. Demographers and statisticians were thrust into the limelight — and courtrooms — as researchers assessed the implications of these directives on data quality and census participation (Van Hook 2018). Some immigrants worry about how confidential large-scale data collection efforts really are, especially given the political context in which data are collected (Held 2020).¹³ Even if privacy protocols are respected, advances in computing and statistics make it possible to re-identify the vast majority of people in supposedly de-identified population surveys and administrative datasets based on relatively few variables (Rocher, Hendrickx, and de Montjoye 2019). Thus, there are harms regarding not only data collection but also data analysis techniques. If migration researchers devise improved techniques to impute, post-hoc, respondents’ probable legal status into secondary data sets with the goal of assessing the hardships of undocumented status, should they release their code and methods in the name of open science? What if these techniques for imputing legal status are used for other purposes?

Migration researchers also must think carefully about their engagement with new technologies or data sources that can improve research and assist in humanitarian responses to migration yet also can be deployed to surveil and target migrants. For example, advances in satellite imaging, GIS analysis, and big data can improve predictions of forced migration, as is being tried in United Nations projects (Molnar 2019). Web-scraping, machine learning, and computational methods can use social media data, from email logins and Twitter to Facebook advertisement data, to pinpoint migrant populations (e.g., Zagheni, Weber, and Gummadi 2017). Migration researchers involved in such efforts have laudable scientific or humanitarian goals. As Petra Molnar (2019, 7) underscores, however, “technology is not inherently democratic”: the ability to collect and analyze data involves deep asymmetries in power, between researcher and study participant, as well as between migrants and the governments that might use data against them. Balancing power asymmetries might also inform whether and how researchers use unconventional digital archives, such as WikiLeaks, as a data source, whether in analyses of refugee deflection (FitzGerald 2019) or private foundations’ funding of immigrant advocacy (Nicholls 2019). The need to weigh new data sources and technological advances against human impact is not unique to migration studies, but we need conscious, repeated discussions of ethical challenges, especially in spaces less familiar with such conversations.

In short, migration scholars must pay attention to open-science debates and engage in a genuine evaluation of whether procedures to increase transparency and accessibility make sense or endanger the groups they study. There are good reasons to increase transparency and replication protocols in migration research, and funders have a legitimate interest in making data available. At the same time, we need more public conversations — with other migration researchers and with non-experts — about balancing open science with ethical obligations toward those we study. Migrants face particular harms and vulnerability, distinct from the citizen population. Those without citizenship can face detention and removal from family, community, and their economic livelihood because they lack status or because an infraction or violation of their visa status was revealed during data collection. In countries criminalizing migrants, state agencies’ subpoena power can be brought to bear on researchers, forcing them to make decisions about providing data or risking contempt of court and the possibility of being jailed. Even if data are not made public but are partially accessible to those in educational institutions (e.g., through a university cloud-based drive or other storage medium), other institutional agents could provide the data to outsiders, irrespective of researchers’ wishes (Jaschik 2010). Given amplified interior immigration enforcement, detention, deportation, and the risks of indefinite family separation in countries around the world, what are best practices for scholars? How do we balance the aspiration to conduct research that can be held to the highest scientific scrutiny and our moral responsibility and personal convictions to protect those we study? We now turn to a consideration of core ethical principles and their particular application to migration research.

Institutional Review Boards and Beyond: Key Ethical Principles and Practice

Historically and today, ethical considerations in research are addressed in part through disciplinary codes of conduct and researchers’ disciplinary training (Haggerty 2004). For instance, members of the American Sociological Association are expected to ensure research participants’ confidentiality and obtain their informed consent.¹⁴ In contrast, the American Political Science Association’s guide to professional ethics includes only a very short section on “Principles Governing Research on Human Subjects” that states, “Possible risk to human subjects is something that political scientist should take into account.”¹⁵ Other professional social science associations have surprisingly little to say on research ethics.¹⁶ Even when they are in place, ethical codes and disciplinary training have proven insufficient at times. Infamous social science studies from the 1960s and 1970s have become incorporated into teaching research ethics, such as psychologists’ Stanley Milgram’s obedience studies and Philip Zimbardo’s Stanford Prison Experiment or sociologist Laud Humphrey’s study of male-with-male sexual encounters. These studies raised serious questions of deception and consent, physical and mental harm, and the violation of research subjects’ privacy. More recently, staff of the International Rescue Committee have reported ethical violations in working with forced migrant populations in Asia and Africa (Behnam and Crabtree 2019).

Such ethical violations have led, in various places, to the development of oversight infrastructures that go beyond disciplinary codes of ethics. In the United States, when scandals broke out over abuses of participants in medical studies, the federal government became involved in establishing an ethical framework for biomedical and behavioral research, crystallized in the 1979 Belmont Report (National Commission 1979; Schrag 2010). In the 1980s, federal agencies formalized rules and regulations and spurred the bureaucratic oversight of research through Institutional Review Boards (IRBs), which undertake formal evaluation of research plans to assess ethics before research is conducted. Most academic researchers will deal with university-based IRBs, but human subject review boards can also be found in hospitals, funding agencies, and school districts; they can even be stand-alone businesses, usually catering to industry research. In 1991, the “Common Rule” incorporated human subject protections into the regulations of, at the time, 16 US federal agencies, including units that provide research funding to universities and hospitals (Williams 2005). By 2005, at least 5,000 IRBs were estimated to exist in the United States (Williams 2005, 36). Today, in many US universities, government regulation, the desire for research dollars, and fears of litigation or reputational damage have generated a highly legalistic field of human subjects oversight, including mandatory IRB review of most research projects involving human subjects, whether conducted by faculty or students, and standardized ethics training, such as the on-line Collaborative Institutional Training Initiative.

A longstanding complaint of social-science researchers is that a medical model dominates the IRB process, imposing at times unnecessary obstacles to research (e.g., heightened scrutiny even when pregnant women only participate in an oral interview) (Schrag 2010). Yet, at other times, IRBs ignore real vulnerabilities not traditionally considered in medical or therapeutic research, such as a person’s legal status. A study of IRBs at 32 US research institutions found that while all categorized children and prisoners as vulnerable, only 11 universities (about 34 percent) identified people who do not speak English as potentially vulnerable research participants, and only one university included immigration status as part of a list of attributes to consider (Perry 2011). Thus, IRB or Research Ethic Board guidelines and requirements, however strict, are necessary but insufficient protections for migration research.

Framing the discrepancy between ethics regulations and on-the-ground ethics (or personal morality) as similar to the “gap between law on the books and law in practice,” Heimer (2013) highlights the uneasy relationship between official ethical guidelines, with their universal regulations and solutions, and practices on the ground, especially among vulnerable populations. She argues that on-the-ground practices often are “off the radar screen both because they are evolving, situationally specific adjustments and because they have an uneasy relationship with official ethics” (2013, 375), particularly in qualitative research (see also Iphofen and Tolich 2018; Roth and von Unger 2018; Zapata-Barrero and Yalaz 2020). In short, formal ethical procedures play an important, necessary role but are not enough to prepare researchers to manage on-the-ground ethical considerations, especially with at-risk-populations (Ellis 2016). We turn next to these challenges.

Informed Consent

Informed consent comes from the idea that autonomous individuals should agree to participate in research voluntarily (National Commission 1979). To do so, they need adequate information about the research so that they can weigh the risks and benefits to make an informed, and real, choice in participation. This ethical principle carries numerous implications, and challenges, for migration scholars.

First, researchers must consider carefully what it means to be “informed” for their population of study. Those asked to be part of research must be able to understand the project, procedures, risks, and benefits, if any. This understanding requires that information is comprehensible, tailored to participants’ linguistic needs, and delivered in a culturally congruent manner. Unfortunately, institutional consent forms often read like dense, complicated legal contracts, with language that may dissuade participation when respondents do not understand it or are suspicious of the formality. In migration research, consent documents, interviews, and surveys may have to be translated and shared in non-jargon-laden language. In research on migration in Mozambique, Agadjanian and his team (Agadjanian, Arnaldo, and Cau 2011) composed documents in the local language first and then translated them into English for IRB approval, thus ensuring that respondents understood their rights in nontechnical language.

Second, researchers must consider the idea of “consent,” which raises questions of power, vulnerability, and cultural appropriateness. In the United States, research with prisoners is held to higher scrutiny by IRBs because incarceration may make an invitation to participate feel like a requisite or a means of getting privileges, not a choice. The traditional heightened bar for informed consent applied to prisoners carries parallels for research in refugee camps or immigrant detention facilities (Clark-Kazak et al. 2017). In addition, for most research purposes, children cannot consent to their participation alone. Rather, they “assent” while parents or legal guardians provide consent. This arrangement raises challenges for research on unaccompanied migrant children, a growing population that requires special ethical attention since they are not only minors, but often also out of status themselves, and arrive without legal guardians (see Galli 2019; Quas and Lyon [2019] for models for this research).

Less extreme, but important, researchers who offer incentives (be it money or something else of value) to encourage participation in an experiment, survey, or semi-structured in-depth interview must balance inducement and coercion. If the incentive is too high, economically disadvantaged migrants may feel pushed to participate even if they are uneasy about a study (Waters 1999). Also, as the practice of monetary incentives becomes more common, researchers must assess whether payment is even appropriate in a particular context and not assume that immigrants, even those who are poor, are automatically grateful for monetary compensation (see Rendón 2020).

More broadly, as Josephson and Smale (2020) argue, the individualism inherent in Western notions of informed consent may not travel to places or populations that “have conceptions of the self that extend beyond an individual making the typical wet [hand-written] signature” (see also Corrigan 2003; McLaughlin and Alfaro-Velcamp 2015). Thus, survey and oral history interviewers should be trained to be culturally competent or hired from the contexts where research is carried out, and ethnographers must be attuned to local or community customs when they enter a site and explain their presence.

In short, to adequately adhere to the principle of informed consent, migration researchers must be attentive to power differentials, especially when interacting with vulnerable populations such as forced migrants, those with uncertain legal statuses, those with limited formal education, or those from cultural backgrounds that make it hard for them to feel as if they can affirmatively and independently consent to research (see the IASFM Code of Ethics 2018).

Migration scholars must also be attentive to the process and procedures of ensuring informed consent. Many IRBs’ routinized practice is to ask a potential study participant to sign an informed consent form. For those offering inducements, administrative bodies often want a record of participants’ name, address, phone number, and/or social security number. These bureaucratic procedures are meant to ensure that consent was, in fact, procured and that research funds are not used fraudulently. Both practices are understandable, but highly problematic when working with migrant populations in legally precarious positions, especially if the consent form is the only document that links identifying information to participants’ data. Increasingly, even noncitizens with secure legal status are reluctant to provide their name or identifying information if study questions might touch on their past migration experiences or carry possible repercussions for legally precarious family members. Indeed, one scholar conducting research with return migrants reports that fear of the US government led Mexican nationals now back in Mexico to express concern over participating in her research.¹⁷

Because we see IRB regulations as a lower-end starting point for protections for migrant research subjects, we recommend that researchers challenge IRB informed consent procedures if and where they seem wrong, while upholding the principles behind informed consent. Ensuring that migrants understand a study’s risks and voluntarily provide consent does not require a paper trail. In some cases, migration scholars will need to educate their review boards regarding migrants’ vulnerabilities and offer alternative procedures. Our sense, based on our experience and speaking to other scholars, is that a decade ago, people serving on IRBs often had little knowledge about the possible problems of traditional informed consent procedures for migrant populations. Today, the 2017 revisions to the Common Rule in the United States note that there may be subjects for whom signing forms is not appropriate. In these cases, researchers can, and should, request approval for alternative mechanisms to document informed consent.

Alternative procedures are possible. Options for not having study participants sign consent forms are, in fact, included in IRB applications in some research universities. A researcher may obtain oral consent through audio recording. The researcher, first, discusses the project — its goals, procedures, compensation (if any), benefits, and possible harms — and then asks for consent, which is recorded (see Gómez Cervantes 2019). Once the participant agrees, the interview, survey, or experiment begins. More rarely, drawing on medical or community health research practices, a third person (or “consent monitor”) can serve as a witness, attesting to proper consent processes.¹⁸ A researcher explains the study to the migrant with someone else present — a family member, religious faith leader, or community member, for example — and if the migrant consents, the witness signs and dates a form attesting that informed consent was received.

Finally, we encourage migration scholars to consider informed consent procedures not just as necessary bureaucratic and ethical procedures but also as a possible opportunity to improve research. Dina Birman (2006) recounts, in her research with Vietnamese and Soviet refugees in the United States, that fear of Communist spies and prior negative experiences with state institutions led to a long consent procedure. While more time-consuming, the research team’s engagement with potential participants also led to helpful modifications in procedures and even turned some suspicious community members into partners willing to recruit additional participants.

Risks of Harm: Distress, Privacy, and Confidentiality

Traditionally, “harm” was thought of as physical and, subsequently, mental health, especially since early ethical codes were responding to barbarous or immoral medical experiments (Schrag 2010). These legacies remain in US regulations that define particular sub-populations, such as pregnant women, fetuses, children, and prisoners, as especially vulnerable and, thus, as requiring greater IRB scrutiny. For social scientists, evaluation of harm usually involves respondents’ possible emotional or psychological distress, stigmatization, or possible damage from the inadvertent disclosure of private, identifiable information. The harm from breaching confidentiality is usually understood as the negative impact from disclosure on participants’ psychological, social, or economic status.

For migration scholars, an obvious danger is public disclosure of migration status or clandestine entry into a country. Yet IRBs may not formally classify immigration status or border-crossing as a source of harm that generates significant risks. The inadvertent disclosure of migration status can, nonetheless, produce direct harms, such as deportation and detention (of the individual or their family members), as well as distress, embarrassment, or stigma, if disclosed to others.

Thus, we argue, migration researchers should proceed as if migrants are already classified as vulnerable populations, while recognizing that not all migrants are the same, nor do they all face the same potential harms. McLaughlin and Alfaro-Velcamp (2015) offer one schematic for determining when and whether migrants should be considered a vulnerable population, based on three risks: lack of access to or knowledge of everyday contract and informed consent expectations; dependence on agencies of settlement or protection; and heightened risks associated with precarious legal statuses. One might add possible trauma from the migratory experience. As vulnerabilities multiply, researchers should heighten their scrutiny. Migration researchers engaged in team-based projects, especially with research assistants who collect data, must provide ethical training to their staff and, where appropriate, may want to consider asking team members to sign a confidentiality agreement. For instance, Menjívar (Menjívar, Agadjanian, and Oh 2020) had each community member who assisted in data collection for a survey of immigrants with Temporary Protected Status obtain the Collaborative Institutional Training Initiative certificate and trained them in the university’s IRB protocols.

In evaluating a research project, IRBs are instructed to consider both the probability and magnitude of harm.¹⁹ We underscore, in migration research, that such evaluations depend not just on the research subject and study but also on the broader socio-political situation and the time. Both the probability and magnitude of harm have increased for migrants in many countries in the last two decades. Thus, even research protocol or procedures that carried little risk of harm in the past might now require re-evaluation.

Mitigating harms: Distress.

The possible harm due to emotional and psychological distress is a real concern for migration researchers. When Bloemraad (2006, 2012) interviewed Vietnamese refugees in metropolitan Boston and Toronto, some respondents cried when they recounted the harrowing experience of fleeing their country by small boats on dangerous seas or trekking for days through the jungle, vulnerable to sickness, pirates, and smugglers. More than one lost a loved one during their flight, a trauma layered on prior experiences of war, deprivation and, for some, surviving re-education camps. Similarly, when Menjívar (2000, 2006) interviewed Central American immigrants, many of whom had fled violence, they narrated the extremely dangerous conditions of exit, harrowing journeys, and continued struggles to adapt to life in their new home, sometimes disclosing painful personal experiences for the first time during an interview. When research plans involve asking migrants questions about difficult experiences, researchers must consider the possibility of re-traumatization.

Social scientists are generally not equipped to provide direct services to study participants experiencing abuse or re-trauma, but researchers can compile lists of available resources potentially relevant to study participants and hand them out. In Phoenix, Menjívar (2006) and her research team distributed information to immigrants who participated in her study about a range of social services, including domestic violence services, housing, legal aid, and counseling services. As the International Association for the Study of Forced Migration underscores in its recent Code of Ethics, “doing no harm” “means proactively prioritizing the dignity, safety and well-being of participants … Particular attention should be paid to the ways research — directly or indirectly — can (re)traumatize, as well as contribute to racism, xenophobia and the criminalization of migration.” This mandate includes not just the data collection process but also dissemination in writing and oral communication to media and policy-makers.²⁰

At the same time, migration researchers should not over-estimate the possible harms from distress compared to the benefits that participants may receive from having their experiences acknowledged and shared with a wider audience. Having an empathetic listener who takes you seriously can be empowering. Menjívar (2000) and Fukui and Menjívar (2015) found that older immigrants appreciate a listener who was genuinely interested in their stories. Adult and young immigrants appreciated the opportunity to talk about the challenges in their lives, as evidenced in Leisy Abrego’s (2014) research on Salvadoran migration. Indeed, in some cases, a researcher is a preferred interlocutor; participants can at times confide to a stranger issues that they would rather not discuss with those close to them (Bloemraad 2012). As Cromer and Newman argue in the context of research on violence against women, benefits to those who participate “include feeling valued and listened to, being treated with respect and dignity, gaining personal insight, and finding the experience to be meaningful” (2011, 1542). To the extent that a researcher communicates findings to other researchers, students, policy-makers, or the general public, the research enterprise can also generate benefits for the larger immigrant community. Thus, even for those who have experienced trauma, participating in research can be empowering. What is important is building appropriate protections into the study design, such as ensuring that interviewers are trained to ask questions in a supportive, respectful manner and to respond to subjects’ reactions appropriately (Abrego 2019).

Confidentiality and privacy.

Breach of confidentiality, invasion of privacy, and the anonymization of study participants’ names and identities are often key ethical concerns in social science research. For ethnographers or for those doing field-based experiments, invasion of privacy can occur if personal information is accessed or collected without subjects’ knowledge or consent. For those who conduct interviews, whether survey-based or oral histories, confidentiality can be breached if a subject’s participation in a study and their identifiable information are revealed to others. Scholars working with certain migrant and refugee populations must be especially sensitive to safeguarding confidentiality, given elevated risks if identifiable data are revealed.

IRBs have developed standard procedures to mitigate these harms. Researchers are often enjoined to collect as few direct individual identifiers as practical (e.g., avoid full names, addresses, or phone numbers) and, before sharing data with others or reporting findings, to de-identify data. De-identification includes removing individual identifiers from notes, transcripts, and electronic data (e.g., audio recordings, survey responses) and, if necessary, creating alternative identifiers, such as pseudonyms or ID numbers. If a link between identifiers and de-identified data is retained, researchers must maintain the linking document separate from data files. For quantitative datasets, researchers also regularly collapse coding categories for rare responses of biographical information (e.g., assigning birthplace as “Southeast Asia,” not “Laos”) to ensure larger aggregate tallies for a particular variable. Researchers may also mask context or place details out of concern that if location, institutional, or organizational details are revealed, those who have knowledge of the place or context could identify study participants. These twin concerns — over individual identifiers and place or context details — lead some migration researchers to advocate for targeted, minimalist data collection (about individuals) and heavy masking of information (for context). De-identification and masking are common in both quantitative and qualitative research.

We appreciate efforts to avoid identifiers entirely in some cases but recognize that the texture and accuracy of migration research, whether a nuanced ethnography or complex statistical analysis, often rely on a depth of information that requires collecting some identifiers. Collecting contact information also makes it possible to do longitudinal studies; the lack of longitudinal data is a serious obstacle to studying immigrant integration over time. Instead of not collecting any identifiers, masking may be the most appropriate strategy, with the additional ethical obligation not to release or make masked data available publicly, despite calls for more open science.

We also underscore that some practices of de-identification, previously considered sufficient, might no longer be adequate, given advances in statistical modeling and more powerful computing. Rocher, Hendrickx, and de Montjoye (2019) show that while the standard de-identification strategies we summarize above traditionally were sufficient for producing “anonymized” datasets built from large population surveys, administrative data, or social media data, today, it is possible to re-identify individuals with over 99 percent accuracy, using a dozen or fewer variables.

Further, now that electronic and cloud-based storage of data is ubiquitous, mitigating the risk of a confidentiality breach has become more complex. Historically, scholars would collect and keep data on hand-written notes, typed transcripts, or a computer disk locked in a file cabinet that sat in a locked office. Today, digital technologies are used to acquire, transmit, analyze, and store data. Research data, whether quantitative or qualitative, are almost always collected and stored electronically, often with back-up on cloud storage platforms. Data analysis tools, such as statistical programs or qualitative data analysis software, are also increasingly hosted on servers accessed through the internet.

These changes not only make study participants’ data more open to hacking but also multiply the number of copies of the data should an outside authority demand access. Most migration researchers are not experts in cyber-security and encryption. Standard procedures to mitigate risks include using (and protecting) computer passwords, putting passwords on electronic files, storing or sending data via encryption, and, in some cases, storing and accessing data only on computers without internet connections. When researchers travel, these risks can multiply. Some universities flag their concerns over forced disclosure of electronic data in foreign countries, if government officials confiscate computers and cell phones.²¹ Increasingly, such concerns extend to the United States and other liberal democracies. For instance, US border control agents have demanded access, including passwords, to unlock electronic devices (ACLU 2019). In Norway, police have asked asylum-seekers to provide Facebook log-in details to examine the veracity of claimants’ accounts (Brekke and Staver 2019). Cloud storage servers or servers that run on-line surveys are generally subject to the jurisdiction and laws of the country in which the server is located, not where the researcher or study participants live, although legal and legislative actions are continuously changing jurisdictional rules (Schwartz 2018). Out of confidentiality concerns, some Canadian-based researchers avoid using US cloud storage applications.²² Indeed, the US-based National Longitudinal Study of Adolescent to Adult Health Survey (Add Health), a rich, highly regarded population survey, explicitly told participants in its fourth wave that “your identifying information will be electronically transmitted to the Add Health Security Manager in Canada” since “[s]toring different types of information in different places helps keep your identity confidential” (Add Health 2021). Qualitative researchers might also consider such cross-border de-identification strategies. We have no easy solutions to the challenges in the 21st century’s electronic frontier, but we urge migration researchers to grapple with the possibilities and implications to safeguard research subjects.

Safeguarding confidentiality is, in general discussions of ethics, often focused on protecting participants’ privacy from disclosure to family, friends, and acquaintances or to individuals in authority such as employers or teachers. Migration researchers, perhaps even more than before, are also acutely sensitive to the dangers of information disclosure to governments. Such risks can come from authorities in migrants’ homelands, who could use data to target loved ones who have not migrated, or from immigration and enforcement agencies in the country of residence, as Chaudhary and Moss (2019) discuss for Pakistani, Syrian, Libyan, and Yemeni diasporas. Certain migrant groups, especially those of Muslim faith, could find themselves accused of terrorism or security threats, even based on seemingly innocuous data about their ties to homeland charities. More generally, revealing migration status or the process of migrant entry could lead to severe harms for participants, including detection, detention, and deportation.

Significantly, in many jurisdictions, social science researchers do not enjoy the same protection from subpoena or legally mandated information disclosure as journalists, doctors, lawyers, or religious leaders, all of whom can protect the identities and information shared by sources, clients, and congregants (Robinson 2016). Assessing social scientists’ ability to resist courts’ subpoenas in the United States, Robinson concludes, “only a few courts have recognized any sort of privilege for academic research, and many courts have explicitly rejected it” (2016, 330).²³ Further, research institutions might provide data without researchers’ consent. For example, in 2010, the University of Arizona provided some research information following a court-issued subpoena on a study of segregated “Mexican rooms” for English Language Learners to the state superintendent of education; researchers accused the university of breaking their pledge of confidentiality to research subjects in the face of a politically motivated attack on their research (Jaschik 2010; Robinson 2016). Social scientists’ lack of protection from subpoena power has been a longstanding concern. Wax and Cassell (1981) argued that the real risk to study participants comes not from what social scientists do in the field, but from what government agencies with subpoena power may do afterward with the data (see also Broadhead 1984). These scholars distinguish between doing harm and doing wrong in research along a continuum: harming increases at the biomedical and experimentation end of the research methodological spectrum, while doing wrong increases at the survey and fieldwork end (Wax and Cassell 1981, 227).

In the United States, researchers can secure a layer of protection against government or court demands for identifiable research data through a Certificate of Confidentiality (CoC). Usually provided by the National Institutes of Health, these certificates protect researchers from compelled disclosure of identifying information in civil, criminal, administrative, legislative, or other proceedings, whether at the federal, state, or local level.²⁴ Previously, researchers had to affirmatively apply for a CoC, but as of October 1, 2017, NIH-funded investigators automatically receive such certificates, which specify conditions for disclosure. A researcher who is not funded by the NIH can apply for a CoC from the NIH. Although CoCs are targeted to health research, NIH adopts a capacious definition of health and well-being that may cover many migrant-centered projects. One review of CoCs’ efficacy in the face of subpoena requests concludes that “Certificates have generally been effective as a deterrent to legal demands for research data and have also been successful when disputes end up in court,” but the legal analysts caution that “those protections have some vulnerabilities, particularly arising from changing technological and informational advances” (Wolf et al. 2015, 604).

Risks to migrant communities: Stigmatization and targeting.

US law, and arguably US social norms, favors individual over group rights, and attention to the individual is embedded in many principles and procedures guiding human subjects review. These individualist orientations can diffuse to other countries when US practices serve as models elsewhere. Informed consent is predicated on a migrant as an individual agent who presumably can and should make an independent choice to participate in research, not on the consent of others in their social networks. Similarly, risks tend to be evaluated for the individual, not the broader community within which a person finds themselves. This individualist focus can pose a problem for immigrants’ families and communities. For instance, even though legal status is conferred on individuals, enforcement practices spillover to family members and entire communities, including neighbors, co-workers, and co-religionists (see Abrego 2019). Disclosing information about an individual can potentially affect families (including citizen members), neighborhoods, religious institutions, and workplaces. Thus, even though IRB protocols assure individuals of their rights to voluntary participation and confidentiality, we urge migration researchers to be cognizant of potential consequences of research practices beyond individual study participants.

One way to advance consideration of migrant communities in research ethics is to establish community advisory boards or engage community members in research design and implementation (see Menjívar, Agadjanian, and Oh 2020). The role of community members can take various forms, from a committee that acts as an occasional sounding board to deep community-based participatory research, which involves immigrants in developing research questions, collecting data, and collaborating on analysis and communication of findings (Wallerstein, et al. 2017). Researchers studying worker health and safety among restaurant workers in San Francisco’s Chinatown argue that community-based participatory research is especially useful when studying populations that have limited education or proficiency in the majority language, hold precarious legal status, fear retaliation from employers, or/and are afraid of authorities due to earlier life experiences (Chang et al. 2013). Community participation can provide perspective on risk and benefit trade-offs, broaden recruitment or research engagement, or signal when some subgroups risk being “over-researched.” Indeed, engaging an immigrant community in research discussions may not just be ethical but also produce better research, shifting the project from a top-down academic endeavor to a more collaborative one.

Moving Forward

As should be clear, we believe that the protocols being created to make our craft more scientific cannot be used in cookie-cutter fashion for all populations, especially migrants. Formal protocols for protecting confidentiality are necessary but not sufficient, particularly when the risks to participants can amount to removal from the country. We recommend that migration researchers engage in a three-step process of understanding open-science and ethical principles, evaluate vulnerability, and then develop appropriate procedures for their migration-related project. We have outlined a number of procedures and strategies, such as use of oral consent, tailored sensitivity, and confidentiality training for everyone engaged in data collection, securing a Certificate of Confidentiality (in the United States), storing files to re-identify data in different political jurisdictions, and masking. In terms of masking, and against the push toward wide-open transparency, we believe that migration scholars should continue to use pseudonyms for individuals’ names and locations and to obscure select details about respondents and their locations. Indeed, migration scholars have started to put extra safeguards in place, such as avoiding public identification of any geographical information, even the name of the jurisdiction where they conducted the research, especially if they describe programs or policies unique to a particular location (see Van Natta et al. 2019). Quantitative scholars, who have long stripped obvious identifiers from their datasets, might have to do more and resist calls to archive data and analysis code. A demographer who has developed techniques to estimate the undocumented population to the local level might decide not to share her code or resultant dataset. More broadly, researchers should double-down on data security, from data collection and storage to analysis and the communication of findings.

Taking such steps in the name of protecting those we study can jar against professional obligations of scientific accountability in the eyes of some. We call on migration researchers, as a community, to engage with advocates of open science, whether in academia, among funders, or in the publishing field, to share ethical concerns and develop ethical procedures that shield migrants from harm. We worry that the enthusiastic embrace of pre-registration, replicability, and transparency, including the requirement to make data and code publicly available, could lead to dire, if unintended, consequences. Early-career scholars, in particular, may feel pressure to adopt a narrowly defined version of scientific accountability. More established migration scholars — those who mentor others, review publications, and help decide on funding — must be vigilant in educating colleagues outside the migration field about our challenges. They can do so by advocating for accommodations within journals that embrace open-science procedures and reminding peers that they should avoid assuming that scientific procedures are universally applicable. Journals such as IMR can also require authors to explain their decisions around data sharing, masking, and other ethics and transparency choices. Similar discussion and negotiation are needed when researchers sit on scientific review panels for grantmaking agencies that have or want to adopt requirements for the public dissemination of data and code. At a minimum, there must always be a procedure to negotiate such rules and to recognize special circumstances.

There are multiple additional ethical challenges that we have not tackled in depth. To end, we flag one final issue: the power inequities that come with the production and dissemination of research-based knowledge, an ethical dilemma in all research, including that on migration. Migrants or those who work with migrant communities tell regular stories of researchers who swoop in, collect data, and then are never seen again (McLaughlin and Alfaro-Velcamp 2015). Various anthropologists and other allied scholars increasingly argue that ethical research, especially with forced migrants or vulnerable migrant populations, must include migrants as part of data collection, analysis, and writing (Selim et al. 2018). The IASFM’s (2018) recent Code of Ethics underscores migrants’ autonomy in being experts on their own experience, the need for equity and diversity, researchers’ obligation to exercise scientific and cultural competency, and the need for genuine partnership. It is not enough to consent to be a research object; rather, migrants should be agents in knowledge production. Similarly, ethical codes, institutional review procedures, and writing about ethical migration research tend to be produced in Western countries, especially North America. As Dina Birman notes, researchers face a challenge “balancing potential differences in the ways ethical behavior is defined by the culture of the researcher and the research community versus the culture of the research participants” (2006, 156). This challenge does not mean that the basic ethical ideas and procedures we outline cannot be applied elsewhere, but clearly, they should not be imposed mechanically. Deliberation on migration research ethics among scholars and with members of migrant communities is critical. Our hope is that this article can advance those conversations.