Conducting qualitative research during COVID-19: reflections on methods and challenges when interviewing marginalized refugee women [version 1; peer review: awaiting peer review]

This reflective opinion article arises from our experience conducting interviews with refugee women attending ante-natal and post-natal services in an urban setting in Kenya in the context of COVID‐19. First, we explain the research context in light of the study objectives. We reflect on the methodological challenges we faced, including researcher’s positionality, and argue that conducting research within the refugee context during the pandemic is unique, therefore research design must reconsider inclusive methodologies tailored to the uniqueness of refugees’ experience in order to obtain useful data. Second, we discuss these challenges in light of our experiences and the implications for addressing the gaps we identified, with particular emphasis on the ways in which professionals working with refugee communities in the context of a pandemic can build trust and obtain meaningful and uncompromised data. Lastly, we provide recommendations for researchers conducting similar research.


Our research context
Although there is a growing body of research among refugee women's access to maternal and child health services in non-pandemic times 1 , there are limited studies that have focused on informants' willingness and ability to participate in research 2 . Over 80% of refugees globally are hosted in low-and middle-income countries, with over half hosted in sub-Saharan Africa region 3 . However, there are rarely any studies that provide a reflection on the methodological approaches and challenges of working with refugee women during a pandemic in sub-Saharan Africa. Refugee women's vulnerability is exacerbated in the context of the COVID-19 pandemic, as they face a triple crisis characterized by the impacts of COVID-19, displacement, and gender-based violence 4 . Pregnant refugee women face barriers in accessing health services with higher rates of adverse pregnancy outcomes 5 . While it is important to conduct research aimed at understanding how COVID-19 has affected refugee women, to develop evidence-informed pandemic responses and gender responsive maternal and child health interventions, it is plausible that refugee women also experience barriers in participating in research. These barriers may include, but are not limited to, endemic poverty, lack of fluency in the interviewers' language, fear of authority, mental health issues, and lack of legal status-all of which may affect informants' ability and willingness to supply the kind of information requested [6][7][8][9][10] . A gendered reflection of refugee women's participation in research during the COVID-19 pandemic is therefore important.
In October 2020, a team of researchers from the department of population health (DPH) at the Medical College at the Aga Khan University conducted research using In-Depth Interviews among refugee women, to investigate the impact of COVID-19 on their access to and utilization of ante-natal and post-natal care services. Our study was approved by Aga Khan University Ethics Review Committee (ERC), National Commission for Science, Technology and Innovation (NACOSTI) and the Nairobi Metropolitan Health Services. The refugee women we interviewed were diverse, with the majority originating from Somalia and a few from Uganda, Tanzania, Ethiopia, South Sudan and Eritrea. Unlike other refugees who live in protected settlements such as Dadaab and Kakuma refugee camps, which are staffed by the United Nations, the refugees we interviewed lived among members of the host communities and did not have legal documentation. As such, they were not able to access free maternity services in public health facilities provided for under the Free Maternity Services Policy 5 . The refugee women we interviewed were directly linked to the International Organization for Migration (IOM) Wellness Clinic in Eastleigh, where they accessed free health care services, situated in one of Kenya's urban centres in the capital city. Voluntary informed consent was sought from all the study participants.

Methodological challenges and innate bias
Recognizing the importance of a gendered reflection of underrepresented refugee women's participation in research during the COVID-19 pandemic, we present and discuss herein our unique experiences during the study to investigate the impact of COVID-19 on refugee women's access to and utilization of ante-natal and post-natal care services. These experiences fall into six main themes: (i) language and trust; (ii) loss in translation; (iii) limited understanding of social research; (iv) childcare challenges; (v) fear, misunderstanding, and insecurity; and (vi) researcher positionality.

Language and trust
The women were directly recruited through the clinic staff and most of the refugees had passed a screening for their ability to speak the local language, Kiswahili, in which the interviews were to be conducted. However, when we began our interviews, however, we observed that getting the conversations going was very difficult, with many of them indicating that they did not understand Kiswahili. The majority of them struggled to express themselves, and provided only brief and guarded statements. We surmised that the refugees, who were unfamiliar with the researchers, needed to establish trust before delving into their experiences about COVID-19 and how it had impacted their access to health services. This finding is consistent with other authors who have indicated that an extensive time commitment to build trust with marginalized refugee women 2 .

Loss in translation
First, when we began to use interpreters fluent in some of the languages spoken by the informants, it soon became evident that a great deal of information was not being translated for us. Some interviewees spoke for 3 minutes in their language, and the interpreter then summarized the information in 1 minute or less. To get the most out of the interpreters, researchers had to watch closely for participants' facial expressions and body movements. For instance, when participant Y, speaking Somali, tearfully related her experience during delivery, the interpreter nodded her head in disbelief and said, "There is a lot of suffering among refugees and she is sad about what is happening." But only when we probed why participant Y was so distressed did the interpreter add: "She says she saw a woman giving birth by the road during the lockdown, when hospitals Z and M were closed. The police were harassing everyone, and the woman could not get any help. She heard that the child died, and it was very sad." This is just one example where valuable information about the effects of the lockdown on pregnant refugees could have been lost in translation, if the interviewer had not probed the interpreter further. This highlights the importance of working with skilled translators, who can directly interpret without overly summarizing or sifting any information.
Limited understanding of social research A number of the participants recruited at the Wellness Clinic did not seem to understand what social research is in general, or what our study aims were in particular. During the interview process, interviewees frequently diverted the discussion to how they could be helped during the COVID-19 epidemic. For instance, when participant X was asked what measures she had put in place for her family to remain COVID-19 safe, she responded, "Any money you can give me, I will use it wisely to put measures in place and feed my four kids and husband. My husband lost his job and we have no income for five months." Many interviewees took part in our research with the hope of monetary or material assistance, which they commonly inquired or requested during the interview process. It is crucial to communicate clearly about the research purposes and to provide some financial incentives for participants 10 . We also posit that while research is a first step to understanding the refugee women's experiences, and could in the long term lead to policy and programmatic changes, however, where possible during the research process, researchers should work with aid agencies that can provide material support or offer referrals for participants in dire circumstances.

Childcare challenges
The participants recruited for our study did not have child care and brought children and other family members to interviews. When they came to be interviewed, children who were eligible to wear masks did not have any. For example, participant Q showed up in the interview room with a 2-month-old baby, a 15-month-old girl, a 3-year and a 7-year-old boy. Mothers frequently required to be excused to breast-feed their babies before returning to be interviewed. At the same time the accompanying children and family members were also constantly present throughout the interviews. Study design and planning, especially in pandemic conditions, needs to take into account that refugees have large families, are too poor to buy masks, and bring their dependents to interviews because they cannot afford child care. In hindsight, none of these factors had been taken into account in the planning for this field study. Although the clinic provided masks for the children, this highlights the many risks refugee children may experience by virtue of accompanying their mothers in public without masks. To gather meaningful data and reduce the undue burden on participants, researchers need to consider providing childcare to participants 10 .
Fear, misunderstanding, and insecurity Some women selected for this study felt uncomfortable with the consenting process, and more often they associated it with being deported, as most of them asked, 'Are you from the government?' It was also clear that we needed to take time to explain to them what research is, why we were asking for their consent, and why we needed to audio-record the conversations. For instance, although we had spent time explaining the purpose of the study, the importance of voluntary informed consent and that we would record the interview, nevertheless, when participant C saw a researcher turning on an audio-recorder, she asked, 'Are you recording me on WhatsApp [she immediately stopped talking]?'. At this point, we stopped the interview and explained the process again to her as well as her right to withdraw if she felt uncomfortable with the interview. The interviewee seemed tearful and fearful and we opted not to interview her. This led us to reflect about the extent to which research among refugees could increase their emotional distress and what better ways research among refugees could be conducted. In the case of participant J, she kept looking at the audio-recorder after she had consented and we had started the interview. We noticed that the audio-recorder made her uncomfortable, so we stopped and put the recorder in our bag, but she never spoke again and we ended the interview process. Participant L instantly stopped talking when we started going through the consent form and explained that we will be audio recording. Although she signed the consent form giving us permission to audio-record, she refused to talk. Consenting processes overall took a long time, sometimes requiring health care workers familiar to them to be present to further assure them that the information they gave was confidential, for the purpose of research and not for government monitoring. There might have been a power dynamic resulting in some participants feeling obliged to participate in the study. The use of audio recorders might have compounded the situation. Similarly, other authors have found that refugees may not be comfortable with the use of tape recorders and researchers will need to be content with taking detailed notes instead 2 . In general, the suitability of the use of audio tapes among this population should be explored in our research context.
As a community health volunteer explained, the migrants did not have documentation, and most of them were careful in writing, signing or being recorded as they believed this information might be used against them. In addition, most interviewees had not heard about research, and some thought we were officials who had brought them some assistance from the government.

Researcher positionality
The issues related to the researchers' positionality had a tremendous impact on the data collected from participants. In our initial interviews, when we presented ourselves as researchers, participants were guarded about what to say. The second week of data collection, having realised that most of our participants were of Muslim origin, the research team dressed in long clothes and burkas, and spent substantial time at the clinic each day with the aim of immersing with the potential participants and gain acceptance. This approach helped the researchers to yield some useful data from all women, including their perceptions of COVID-19. For example, when participant Q was asked about how COVID-19 had impacted her access to ante-natal services, she retorted, 'As you know, I am not affected because COVID-19 is not for us (pointing to the researcher wearing a burka) but it is a punishment for Christians and all white people'. As a researcher, we felt that this type of sentiment would only be shared under a sense of trust and that how researchers position themselves as insiders has implications for the richness of the data collected 11 . For instance, following our failed first attempt at conversations in week one, we obtained rich data in week two when we dressed in attire favoured by refugees and covered our heads. In addition, consenting and audio recording process did not seem to traumatise participants. Although, it seems clear that dressing in burkas and spending substantial amount of time at the clinic finally led to useful data, our shifting positionalities, wearing burkas and being accepted, while it led to useful data, left us with ethical dilemmas as to whether disguising in the form of immersion is an appropriate method to collect data among vulnerable communities who are culturally difficult to engage. Methodologies such as in-depth interviews whereby researchers go to the field to interview vulnerable participants like refugees must be reconsidered. It may be better to engage the local refugee women where possible to collect data among their communities rather than have outsiders do so.

Conclusions
The methodological challenges we encountered raise critical issues that must be explored further while conducting research with vulnerable populations such as refugee women. Researchers need to reconsider strategies to engage refugee women and collect useful data in an ethical manner. First, the methodology used in this study calls upon researchers studying refugees or migrant populations who speak a foreign language to consider training individuals in the participants' own communities to conduct research as well as use refugee interpreters to transcribe such data to avoid meaningful data gaps. This would ensure greater value in the data being collected and could help place participants more at ease. Second, prior to recruiting and interviewing refugee women participants, community sensitization is necessary to build a shared understanding and trust on the purpose and value of research and to avoid perceptions that the research should involve any immediate direct benefit. However, it also appears that small and modest incentives should be considered for refugee women's participation in research. Third, refugees, and in this specific case refugees of Muslim origin, may have large families and child care can be challenging. Consideration for child care must be in place when interviewing refugee women of reproductive age to ensure quality data collection. Fourth, research about refugees must be conducted in a language in which they are fluent, and if interpreters are needed, they must be people with whom the respondents feel most comfortable and who are competent to translate the interviews without losing valuable information. Fifth, the provision of financial incentives by researchers can improve the participation of under-represented refugee women in research. Finally, researchers need to operate in an ethical manner to balance securing the trust and willingness to participate among respondents with the collection of quality data. To conclude, the need for research among vulnerable populations such as refugees is great, but researchers need to pay close attention to their methodologies to ensure they collect quality data in an ethical manner and in ways that promote inclusion for under-researched refugee women.

Data availability
No data is associated with this article