Safeguarding personal integrity while collecting sensitive data using narrative interviews – a research note

ABSTRACT Narrative interviews with young victims of crime can provide information regarding their unique situations and how the victimization has affected their lives. However, the method can be intrusive, and not all young people are able to safeguard their personal integrity. This research note offers reflections on the use of narrative interviews with young victims of crime, and on interview situations that raised ethical quandaries about whether to discontinue the interviews to reduce the risk of harm. The note starts with a brief description of research on sensitive topics, the study, and the narrative method. After that, I reflect upon some interview situations that have left me wondering what I could have done differently, resulting in suggestions on what could be changed.


Introduction
This research note discusses the tensions arising from the needs of a research method that is founded on participants sharing sensitive stories about their lives and the ethical need to protect participants' personal integrity, especially when focusing on vulnerable groups who might not be able to safeguard their personal integrity (cf. Cater & Øverlien, 2014). The definition of a sensitive research subject varies, but one definition is that the topic may be potentially harmful to those involved, for example, when the data collection or the dissemination of the results may have problematic consequences for them (Lee & Renzetti, 1993). Researching sensitive topics raises several ethical concerns due to the potential risks for the participant. However, research shows that most participants are not harmed, with some even stating that their participation was helpful to them (e.g. Alexander et al., 2018;Decker et al., 2011;Kavanaugh & Ayres, 1998). Still, precautions are needed to be able to identify risks of potential harm (Draucker et al., 2009). For example, some participants may not verbally express that they are distressed during the interview, making it important to identify non-verbal cues of distress. The present research note focuses on how to identify distress and potential harm to participants during ongoing narrative interviews on criminal victimization, and how researchers might be able to help participants safeguard their personal integrity.
between long and short narratives. Long narratives focus on the participant's life from his or her first memory to the present day, while short narratives focus on a specific moment in life, such as a traumatic event, and build the interview around the participant's story from that event up until the time of the interview. Regardless of the type of narrative used, the stories are co-created in the social setting of the interview (Plummer, 2001). This is also emphasized in 'teller-focused' interviews, where the researcher uses open-ended questions to let the participants tell their stories in their own words (Hydén, 2000(Hydén, , 2008. This can be understood as a refinement of the narrative interview to focus more closely on the 'teller' and on the interaction with the researcher when life events are presented and discussed, shifting the focus from the story itself to the participant and the communicative process (Hydén, 2000). This is also apparent in the different forms of narrative analysis, with some focusing on the teller or the communication, and some focusing on the story (Riessman, 2008). The common ground between these traditions is that they all aim to capture how one event leads to another (McAllum et al., 2019).
In my research project, I chose to take a middle path between these different emphases within narrative methodology, using a short narrative focusing on a victimizing event. The reflections in this research note are based on narrative interviews with 19 participants aged 18 to 22 years in 2016-2017, who were asked about experiences of victimization when they were 15 to 19 years old (see Thunberg, 2020). The participants were recruited using public court verdicts, and they chose the time and place for the interview. The participants had been victims of assault, threats, robbery, sexual harassment, harassment, theft, and/or youth intimate partner violence. Each interview began by informing the participants about the interview method and their right to decline to answer any question and to withdraw their consent at any time. Any questions the participants had were answered before they gave their written consent to participate. Before the audio-recording, I also informed them that if they wished to take a break, they just had to let me know and I would pause the recording.

Reflections on narrative interviews on sensitive topics
Narrative interviews depend on participants being willing to share their stories with the researcher. With sensitive topics, it is especially important that the social setting makes participants feel that they can tell as much or as little as they feel comfortable with, and that they are in control of the conversation. This is easier said than done, due to the potential power imbalance in the situation, which may cause young people to feel they must answer all the questions to please the researcher, even if they do not want to. The researcher is dependent on the participants, but the participants might not understand that this gives them control over the situation. This creates a tension between the needs of the research method, which builds on participants sharing sensitive stories, and the ethical need to safeguard the participants' personal integrity by making sure they do not share more than they want to. To reduce the risk of the young people feeling forced to talk about something they would rather keep to themselves, I tried to pay particular attention to non-verbal cues such as tone of voice, emotional expressions, and body language. Before the interviews started, the participants were reminded that it was up to them to decide what and how much they want to tell me (cf. Draucker et al., 2009). Two situations stood out as problematic in this regard: when participants seemed to share too much, and when others clearly stated that they did not want talk about something. As interviews on victimization can evoke strong emotions, I continue to reflect on these situations to learn what could have been done differently.

Sharing too much
The first situation concerns young people seeming to share sensitive information that they had not planned to share originally. In most cases the information concerned other instances of victimization than what was the topic of the interview. This sharing came after the question 'Have you been victimized before or after the victimization in focus for this interview?' This invited them to talk about other experiences but could also be answered with a 'yes' or 'no'. The question was asked to capture occurrences of poly-victimization, but not to go further into other potential experiences. I was thus mentally prepared for other victimizing experiences, but not for the types of severe experiences they described. One participant, for example, first went silent, and then curtly stated that she had been sexually victimized by a relative. Her initial silence can be interpreted as hesitation, which may be a non-verbal cue of distress. I did not ask any follow-up questions, but instead continued the interview with a new theme to move away from the subject. In a later interview, a similar situation occurred with the participant describing an attempted rape that occurred a few months prior to the interview but after the victimization in focus. Here I was more observant and asked whether she had received any support, which she had. These were two of the first interviews I conducted, and I did not reflect upon the situation more thoroughly until afterwards when I had gained some distance (cf. Mauthner & Doucet, 2003), and I can see in the material that I am more observant in the later interviews, suggesting that as my experience grew I became better at listening for signs of distress.
I handled both situations by continuing the interviews, as I did not want to discourage them from continuing. Although, in the second interview (attempted rape), I reminded her that we could pause or end the interview at any time. Also, it did not seem at the time that they were distressed or that they wanted to end the interview. Still, I keep wondering if that was enough, or if I had opened a Pandora's box, risking potential emotional harm. I tried to focus on non-verbal signals, as they might not align with verbal signals, meaning that a person can send out contradictory signals regarding distress that are hard to read. For me it came down to trying to make sure that the participants' personal integrity was protected, and if it seemed that they did not want to talk about something, I tried to steer the conversation in a different direction (e.g. by continuing the interview with a new theme). However, that goes against the narrative method, according to which the researcher should not steer the conversation, but instead should let the participants tell their stories in their own words, at their own pace, and with as little interference as possible. This raises questions concerning where the researcher's 'loyalty' should lie -with the method or the participant? Of course the researcher should not switch roles from researcher to support provider, but I would still say that one's loyalty needs to be with the participants, to minimize the risk of harm (cf. Cater & Øverlien, 2014). If a specific interview situation diverges too far from the method, then that is something that can be discussed as a limitation or, in the worst case, the interview might need to be excluded, as it cannot be analyzed as intended.

Safeguarding personal integrity
The second situation was when participants did not contribute much information at all, which occurred more often with the young men. One case stands out, where the participant also had committed crimes. He talked about how he usually hangs out with his friends and does various things, and that because of this he did not need support. I asked what kinds of things they usually do, as I was curious if there was anything they did that made him feel supported. He immediately responded that I must not ask about that; he could not tell me what they do together. He exerted his agency and power over the storytelling and gave me a clear verbal signal that I had overstepped some boundary concerning what he was willing to talk about. Although he safeguarded his personal integrity by doing so, it caused me to wonder if I had missed any non-verbal cues regarding his boundaries. Perhaps I should not have asked a follow-up question, but at the time it seemed relevant to the aim of the study. Still, as he was clear about what he was willing to discuss, it does not seem that he was emotionally affected by the interview, when compared to the participants described above.

Discussion
The examples presented above illustrate how narrative interviews can affect the personal integrity of young victims, as some young victims are more aware of their personal boundaries than others. In my experience, it is easy to be drawn into the storytelling and become an active part of it. By this, I do not mean taking over the interview situation and asking a lot of questions, but rather that my body language, such as nodding or leaning towards the teller, and interjections like 'hmm' or 'yeah', can signal to the teller that, as a researcher, I find something particularly interesting. All of this is natural in a conversation, but in a research situation it can have consequences for both the individual participant and the research project, as verbal and non-verbal cues from the researcher may lead participants to share information that they had not planned to disclose. For this reason, researchers constantly need to be aware of both their own and the participants' verbal and nonverbal signals. The non-verbal cues are the hardest to identify and, as described above, can take the form of changes in tone of voice or a participant suddenly going silent. It is also important for researchers to maintain professional distance and not get too involved (Cater & Øverlien, 2014;Dickson-Swift et al., 2006).
In hindsight, I would have changed some things. I would have liked to be more prepared for these kinds of situations, specifically to handle potential over-disclosure and thereby help participants to safeguard their personal integrity. As shown above, not everyone is able to say that they do not want to answer a question. Had I been better prepared, the situations that arose might not have come as such a surprise and been so difficult to handle on the spot. It would also probably have been better to use an even more open-ended question, asking the participants about their lives before the victimization, rather than specifically asking about other experiences of victimization. Follow-up questions could then have concerned family life, school life, other experiences of victimization, friends, spare time activities, and so on. This way, the participants would have been freer to choose what aspects of their lives prior to the victimization they wanted to talk about. Also, reminding the participants regularly that they do not have to share information unless they want to, could have been a good strategy. On the positive side, the interview guides were individually tailored to each participant based on the public court verdict, and information about support organizations was handed out before the interviews began, to make sure that the participants knew where to turn for support if the interviews triggered painful emotions related to the victimization. Still, you can never fully prepare for all possible scenarios.
I would not suggest that the methodology as such needs to be changed, but practical training with the technique is important, especially when a project focuses on sensitive topics, to reduce the risk of harm (Kavanaugh & Ayres, 1998). Without proper training, the risk of harming the participants increases. It is also important that researchers adopt a reflexive stance (cf. Cater & Øverlien, 2014), and share their positive and negative experiences of different methods with each other, to enable others to learn from their experiences and reduce the risk of mistakes being repeated. This reflexivity is also of importance throughout the research process, as a number of factors can affect the results (Mauthner & Doucet, 2003). These reflections, therefore, need to be made a part of the research process and the method to acknowledge what has happened in specific situations, and how this might have affected the process and the people involved. As far as I know, none of the participants availed themselves of support from any of the support organizations they received information about. Several of them also said that although it was difficult to talk about their victimization, they felt better afterwards, and that during the interview they felt like they were talking with a friend (cf. Alexander et al., 2018;Decker et al., 2011;Dickson-Swift et al., 2006). This means that the concerns I am raising here may not have been a real-life problem for the participants, but even so, it is important to reflect upon what happened, and on what could have happened, to learn from any mistakes and thereby reduce the risk of harm to participants.

Funding
The data collection was supported by the Swedish Crime Victim Compensation and Support Authority under grant [08563/2016]. The arguments made in this paper are the author's; the funding agency was not involved in the writing.

Notes on contributor
Sara Thunberg is a PhD and Senior lecturer in Social Work at Örebro University, Sweden. Currently, her research interests focus on young victims of crime and their need of support. This also includes how young people construct their own victimhood and how this affects their perceived need of support, and what kind of support they perceive as available to them. She uses qualitative methods, as well as mixed methods in her research.