Interviews

While participant observation afforded a broad understanding of conditions in the field and facilitated initial contact with participants, interviews were essential in understanding Roma participants’ distinct experiences of using UK public services and their impressions of life in the UK. Furthermore, interviews with health professionals and community advocacy workers offered alternative perspectives on the operations of health and public service institutions, thus providing insight into dimensions of the broader social context of health in Roma communities.

I considered a range of frameworks for structuring interviews, considering the relative advantages of structured, semi-structured and unstructured interviews. Structured interviews adhere strictly to a pre-determined set of questions and do not deviate from these topics, even in cases where novel impressions arise from the conversation (Green & Thorogood, 2009). While this strategy may have ben productive in studies that seek to validate or invalidate a pre-determined set of concepts, I intended my research to explore dimensions of Roma health beliefs and experiences that I might not have considered in the development stages of my project and thus deemed structured interviews to be unsuitable for

meeting my research objectives. Unstructured interviews, by contrast, lack the rigid framework and predetermined questions of structured interviews, and instead allow the interview to progress as an open conversation between interviewer and interviewee (Brinkman, 2018). The interviewer refrains from actively guiding the discussion and encourages the exploration of topics identified by the interviewee. Semi-structured interviews strike a balance between structured and unstructured interviews, adhering to a set of pre-determined questions while also allowing for exploration of unanticipated topics that arise over the course of the interview (Green & Thorogood, 2009).

My initial intention was to use semi-structured interviews, which would allow me to select a set of topics that were relevant to my research questions. While asking each participant the same general set of questions, however, I remained open to the development of conversation along previously unanticipated lines, which, as discussed earlier in this chapter, could lead to substantial divergences from the topic guide. Early in the process of conducting interviews, I observed that health-related communication often took the form of storytelling and that, in many cases, I could gather the richest data if I could stimulate participants to narrate a specific interaction with health or social services.

Adhering too strictly to a discrete set of interview questions could put strain on interactions with interviewees, as highly structured questioning could recall experiences of interactions with officialdom in participants’ countries of origin (Brown et al., 2017). As such, I found it effective in many cases to ask broad questions at the beginning of each interview to build the respondent’s familiarity with the topic and then to allow each individual’s particular experiences and impressions to guide the direction of the interview.

Once my health advocacy volunteer work put me in direct contact with participants, I engaged in participant observation as my sole method of data collection for approximately one month before inviting community members to participate in interviews. This enabled me to formulate a set of initial impressions to further explore and verify. Considering these impressions in conjunction with information gathered through my literature review, I developed an interview topic guide (Appendix 4) that addressed health-related beliefs and behaviours in the Roma community, perceptions of health services and interactions with health professionals. In order to ensure that non-directive questions ‘stimulate[d]

the interviewee into talking about a particular broad area’ (Hammersley & Atkinson, 1983, p. 117), I employed a topic guide from a previous study conducted with Eastern European Roma communities as a

template for developing my own interview questions, adjusting this as my knowledge of the field increased (Tobi, Sheridan & Lais, 2010).

My questioning strategy evolved over time, as participants’ prior responses suggested new topics to explore. This practice of making continuous adjustments to the topic guide aligns with the grounded theory methodology, which calls for adjustment of questions to reflect new concepts arising from previous interviews or observations in the field (Strauss & Corbin, 1998). When, for example, I noticed that benefits for people with disabilities and chronic illnesses appeared to be significant to many community members, I incorporated a set of related questions into my topic guide.

Language barriers between participants and myself constituted a major practical consideration in conducting interviews. While a number of interviewees were confident in participating in an interview in English, many required the assistance of an interpreter. Bilingual RSG and LRT staff members provided invaluable support to my research in serving as interpreters and moreover as cultural mediators. They were able to inform me when my interview questions might need clarification, and helped to explain the goals of my study in a manner that participants found meaningful (Hennink, 2008). In the event that participants preferred that the interview proceed without the presence of an RSG/LRT advocacy worker, I included the option of allowing family members to serve as interpreters.

Although use of family interpreters carried certain risks associated with interpreters’ objectivity and the accuracy of the information provided, there is precedent for use of family members as interpreters in health research settings (Hadziabdic et al., 2009; Hadziabdic et al., 2014; Karliner et al., 2007). While acknowledging that interpreters’ English language skills, their understanding of the purpose of my study and participants’ possible reluctance to answer questions in the presence of family members could impact the quality of my data, excluding non-English speakers would have imposed greater limitations on my study. Furthermore, as my interview questions did not ask participants to disclose sensitive details of health conditions, there was limited risk that they would find the interview too invasive in the presence of a family member. If, however, the presence of a family member appeared to make a participant uncomfortable, I would check to make sure that the participant was still happy to continue with the interview and would discontinue the interview if necessary.

To ensure that interviews were conducted in settings that were convenient for all involved parties, they took place either in community centres or nearby cafes. All participants were provided with an information sheet prior to interview, which was read to them in the event that they were illiterate or unable to read English. After ensuring that participants understood the purpose of the study and the nature of their involvement, all interviewees signed written consent forms. Interviews were audio recorded with participants’ consent and were transcribed verbatim. Two participants declined to be audio recorded, and in these cases I took detailed notes during the interviews, noting direct quotes as much as possible.

In total, I conducted 28 interviews with Roma community members and an additional ten Interviews with health professionals and advocacy workers. I began to interview professionals at approximately the midpoint of my interviews with Roma community members, after I was sufficiently confident in my understanding of Roma health experiences to explore them from a different perspective. My topic guide for health and advocacy professionals focused on access to services and challenges in cross-cultural communication, seeking to reveal other dimensions of Roma community members’ responses.

Interviews lasted between 30 minutes and one hour. Although topic guides aimed to present questions as straightforwardly as possible and were reviewed with Roma community members for clarity, some Roma interviewees encountered difficulties in formulating the abstract responses required by certain interview questions. Narrating a sequence of events presented no challenge, yet when asked questions such as ‘how would you describe communication with your GP?’ participants would often respond with either ‘good’ or ‘bad’, and appear confused when asked to elaborate. When I discussed this observation with one of my community organisation partners (who is from the Roma community himself), he explained that many respondents had never been educated beyond the primary school level and therefore might have been unsure of how to answer questions that sought to capture more abstract feelings and impressions. To address this, I constructed questions that stimulated respondents to tell stories about their interactions with health services. Where direct questions about health experiences could be intimidating to some participants, encouraging them to construct narratives effectively removed many of the communication barriers imposed by more formal interview frameworks.