Safeguarding and Agency: Methodological Tensions in Conducting Research with Survivors of Sexual Violence in Universities

: This paper examines the tension between safeguarding measures and participant agency in conducting feminist interviews with survivors of sexual violence in universities. There is a core contradiction inherent in feminist research of gender-related violence, including sexual violence, because participants have been traumatized: Research with survivors of violence must enact appropriate safeguarding measures to ensure their emotional wellbeing, yet in designing these safeguarding measures, researchers must also ensure that survivor participants can exert agency within the research process. These phenomena are often at odds as safeguarding—the work of protecting participants through limiting their exposure to upsetting stimuli—appears to circumscribe participant agency, or a participant’s ability to make informed choices for themselves that respond to and change the structures in which they are situated. Using part of my doctoral thesis research’s methodology, I detail the safeguarding measures I implemented for participants as well as highlight how and where I attempted to build in agential engagement for survivor participants, and whether, or how often, survivors took up these options. The article concludes by suggesting ways gender-related violence research more broadly can reﬂect on and continue to interrogate how researchers balance safeguarding requirements while enabling survivors to assert their agency in the research process.


Introduction
This paper examines the tension between safeguarding measures and participant agency in conducting interview research with survivors of sexual violence. There is a core contradiction inherent in feminist research of gender-related violence (Alldred and Biglia 2015;Gorelick 1991), including sexual violence, because participants have been traumatized: Research with survivors of violence must enact appropriate safeguarding measures to ensure their emotional wellbeing, yet in designing these safeguarding measures, researchers must also ensure that survivor participants can exert agency within the research process. These phenomena are often at odds as safeguarding-the work of protecting participants through limiting their exposure to upsetting stimuli-appears to circumscribe participant agency, or the ability of participants to make informed choices for themselves that respond to and change the structures in which they are situated (Emirbayer and Mische 1998). In response to positivist research paradigms driven by white male scholars, feminist research highlights the subjectivity of everyone involved in the research process, which includes recognizing research participants as actors in their own right, not as objects to be studied (Gorelick 1991;Hesse-Biber 2007). The ability to exert agency is tied closely to the subject status of participants, and is significant for survivors of sexual violence: Due to the dynamics of sexual violence-in which someone strips away another's ability to say no as well as their bodily autonomy, therefore removing the survivor's agency at the moment of violence-it is particularly important that those of us researching with survivors do not infringe on their agency in our methodological design lest we replicate power dynamics similar to those which occurred in the initial experience of violence. While researchers have noted this contradiction (Burgess-Proctor 2015; Gillies and Alldred 2012), there has been little examination of what navigating these nuanced constraints looks like in practice.
In what follows, I analyze this tension between safeguarding and agency in feminist research on gender-related violence through a review of the aspects of the methodological framework I created for interviewing student survivors of sexual violence in English and US universities. The article begins by presenting an overview of feminist research epistemology, ethics, and methodological interventions before introducing the context of my doctoral research. I then address the ethical considerations of trauma research and of researching sensitive subjects. Following this discussion, I detail several safeguarding measures I have implemented for participants as well as how and where I attempted to build in agential engagement for survivor participants, and whether or how often survivors took up these options. The article concludes by suggesting ways gender-related violence research more broadly can reflect on and continue to interrogate how researchers balance safeguarding requirements with the need to enable survivors to assert their agency in the research process.

Feminist Research Epistemology, Ethics, and Methodological Interventions
Despite much of feminist theory critiquing binaries-for example, between man and woman, public and private, personal and political-feminist methodology and some of its foundational ontological and epistemological positions did not appear on their own, but rather in opposition to male-dominated knowledge regimes. Descartes and his separation of mind from matter or subject from object, known as Cartesian dualism, are a focal point of feminist challenge to Enlightenment thought. Working within the binary system this philosophy creates, if men are reasonable, then women are emotional. This distinction has allowed men to drive public scientific inquiry while pushing women towards the private domestic sphere (Ramazanoglu and Holland 2002). In extreme positivist approaches, such a stance positions the researcher as subject and the researched as object and therefore "studying human beings is, in principle, no different from studying things" (Gorelick 1991, p. 460).
To challenge hegemonic Enlightenment thought, particularly the assumption that men are neutral objective researchers, feminists have called attention to researchers as people informed by their positioning in social hierarchies, as opposed to detached and therefore (falsely) objective. In this view, what traditional non-feminist researchers have put forth as "objective" is not universal, but rather a reflection of the dominant class of white male researchers; Haslanger sums up this feminist critique of objectivity as the recognition that "'objective' reality is rather 'male' reality" (Haslanger 2015, pp. 91-2). Feminist researchers encourage the use of reflexivity, or acknowledging how we ourselves are positioned. These positions include both the power hierarchies inherent in conducting research that privilege the researcher over the participant (Oakley 1981) and where we are situated within social structures (e.g., gender, race, class, sexual orientation). Incorporating reflexivity into analysis highlights how the various positions and identities we occupy affect interactions with our participants as well as how we approach analyzing data (Hesse-Biber 2007).
Beyond reflexivity, there are several other main features of feminist research ethics. Despite differences in specific standpoints, as feminism is a diverse and fragmented ideology (hooks 2015), feminist researchers overall tend to prioritize studying lived experienceparticularly the experiences of women and those occupying other and additional marginalized identities-for an explicitly liberatory purpose (Burgess-Proctor 2015;Kirsch 1999). Feminist researchers also advocate for reducing the implicit hierarchy between the researcher and research participants by emphasizing collaboration with participants (Burgess-Proctor 2015; Campbell et al. 2010;Oakley 1981). As part of reducing this hierarchy, many researchers employ a feminist ethic of care "centred around an orientation of support and respect" for research participants (Burgess-Proctor 2015, p. 126; see also Campbell et al. 2010). Perhaps most significantly, feminist ethics aim to address urgent issues of inequality, change harmful cultures, and further social justice (Burgess-Proctor 2015;Hesse-Biber 2007;Reinharz 1992).
There are, however, limits and complications when putting these ethics into practice. Gillies and Alldred (2012) assert that, The notion of empowering women through the research process is appealing to many feminists. However, the associated ethical dimensions are complex . . . However participant-led the research may be, the researcher plays a crucial role in initiating, facilitating, and constructing meanings-a point that is often played down in the emphasis on democratic rapport and participant empowerment. Simplistic ideas of participation and empowerment can obscure other aspects of the researcher's power and responsibility. (p. 52) These particular issues of power and agency are heightened when working with a vulnerable research population, such as survivors of gender-related violence. Burgess-Proctor argues that "the safeguards feminist researchers often are required to employ in order to protect our participants may actually serve to reinforce participants' disempowerment, thus replicating the very problem these safeguards seek to remedy" (2015, p. 125). This core issue, balancing safeguarding measures with participant agency and empowerment, is something I attempted to mediate in designing my feminist methodological framework for interviewing survivors of sexual violence. I now offer an overview of my doctoral thesis research before analyzing part of my safeguarding framework and drawing attention to opportunities and constraints for expressing agency within it.

Research Questions and Methods
Sexual violence in US and English universities affects a significant number of students. The (2010) Hidden Marks report found that 25% of female respondents experienced some form of sexual assault while studying in further or higher education and one in seven experienced a serious physical or sexual assault (National Union of Students (NUS) (2010)). In the US, a large-scale survey of 27 higher education institutions found that 21.2% of all final year undergraduates had experienced sexual violence while studying. Certain groups of US students were more vulnerable than others, as this number rose to 33.1% for final year female students and to 39.1% for transgender, genderqueer, or questioning students (Cantor et al. 2015). While universities in the US and England have comparable sexual violence victimisation rates as well as similarities in student demographics and culture (Phipps and Smith 2012), frameworks for institutional response vary dramatically across the nations: The US has a legalistic national framework for university response in Title IX, whereas England has no such regulation and thus enables a multitude of policies across universities. My doctoral thesis therefore sought to understand how a selection of universities in these countries respond to student reports of sexual violence.
The overarching research question asked: How do some universities in the US and in England 1 respond to student disclosures of sexual violence? In investigating this main question, I also asked the following sub-questions: (1) How do national policies and guidance in the US and England conceptualize sexual violence and sexual harassment?
(2) How do university staff in response roles (e.g., Title IX Coordinators or student welfare staff) perceive and navigate their university's response to student disclosures of sexual violence? (3) What are student survivors' experiences of university responses to disclosures of sexual violence in the US and England? These sub-questions sought to explore the overall issue of institutional responses at three levels: At the structural level of policy to understand intended responses, at the institutional level of staff implementation of policy, and at the subjective level of student experiences of institutional responses. While this was a qualitative multi-method study that used both policy discourse analysis and feminist interviews with university response staff and student survivors, this article focuses on semi-structured feminist interviews with student survivors of sexual violence, and the complexities of attempting to balance safeguarding with participant agency.

Student Population Demographics
Following approval from University of York's Department of Education Ethics Committee, I engaged in data collection from September 2018 to June 2019, which involved carrying out semi-structured interviews that lasted between 45 min and one hour, either in-person or via Skype. My final study population included 19 survivors who self-selected into the research following a call for participants I posted on Twitter. The call asked for current students and graduates within one year of completing their degrees, specified that they had to have experienced sexual violence while a university student, and had to have tried to disclose this experience to their universities. I explicitly noted that the interviews would focus on how their universities responded to their disclosures, and that I would not ask them about the experience of sexual violence that they had (tried to) report. Since my research questions did not ask about experiences of sexual violence, not asking participants about sexual violence was the ethical choice to avoid them engaging in the emotional toll of giving me information that was not directly relevant to the study; collecting the minimal amount of data necessary is particularly important in studies that cover potentially distressing topics as it minimizes the risk of retraumatisation.
Self-selecting participation for student interviewees resulted in a more homogenous sample than I had hoped. The vast majority of student survivor participants were white cisgender women, and scholarship on sexual violence in universities already tends to centre these experiences, and as I myself am white, that also impacted how participants responded to me in the interview and shaped how I analysed the data. Of the 19 student participants, seven came from English universities: Of these seven, four were cisgender women and three were nonbinary; all were white; five disclosed during undergraduate education and two disclosed during postgraduate study; and five came from researchintensive universities while two came from teaching-intensive universities. Five interviews occurred in person where I was based as the researcher, in private study rooms of the University of York library, and two occurred over Skype. Twelve students had attended US universities: Of these, 10 were cisgender women, one was a transgender woman, and one was a cisgender man; 10 were white while one was Black and one was Asian-American; eight disclosed during undergraduate education and four disclosed during postgraduate education (including a borderline early career researcher who experienced harassment both as a postdoctoral research fellow and as a medical school resident); and eight attended public state universities, three attended private universities, and one attended an Ivy League university. 2 Unlike my English student interviews, which mainly occurred in person, all interviews with US students happened via Skype. Some of these happened while I was still in England before I returned to the US to conduct staff interviews at case study institutions, so it would not have been possible to speak any other way; for the few remaining student interviews that occurred when I was back in the US, participants and I were at large geographical (and time zone) differences within the country, so Skype again was a necessity. Beyond the geographical expanse of my research necessitating online interviewing technology, however, I also used Skype in an attempt to increase survivor agency, which I address in a later section.

Trauma Research and Researching Sensitive Subjects
Sexual violence is a distressing topic to discuss, especially for those who have experienced it. As both trauma victims and qualitative interviewers, Brzuzy et al. (1997) explain, in conducting research with female trauma survivors, common participant responses have included heightened anxiety before the interview, becoming emotional in the interview when speaking about the trauma, and emotional aftermath following the interviews such as nightmares. Despite knowing that engaging in trauma research may trigger negative emotional responses, many victims participate out of a desire to make things better for others, to work through some of their own experiences, and/or to take back control over situations in which they had none (Brzuzy et al. 1997). Even with self-selected participants in trauma research, however, Brzuzy and colleagues warn of "[t]he tension between the need to gather information and the possible revictimization of survivors" (1997, p. 79). Fontes (2004) highlights the way in which the specific nature of research into violence against women can exacerbate this tension because of the dynamics of shame and disbelief that accompany speaking out about such stigmatized experiences.
Although trauma researchers agree that there are risks in the work they do with survivors, not all agree that trauma research is inherently retraumatizing to participants. Members of this latter faction charge that there is a necessary and important distinction between evoking emotion from participants and retraumatisation, since too often people conflate trauma research with retraumatisation. As Newman and Kaloupek (2004) state, [u]se of [the word 'retraumatise'] is unwarranted in the research context because it equates recounting a traumatic experience with the actual occurrence of traumatic exposure. It essentially ignores the distinction between distress that emanates from recall of an experience and, for example, the 'intense fear, helplessness, or horror' . . . that emanates from direct experience with a traumatic stressor. (p. 390) One significant way in which trauma research differs from a traumatic experience is the presence of control: Trauma researchers give their participants control in recalling a situation in which others took their control away from them (Newman and Kaloupek 2004). Good ethical practice in interviewing survivors of sexual violence is therefore not about minimizing expressions of fear, anger, or upset, but rather ensuring that there are supportive measures in place for both the participants and for the interviewer to safely work through whatever emotions arise during the research.

Safeguarding Measures
Following best practice guidance from feminist and trauma researchers, I enacted safeguarding measures for participants. Before discussing specifics, I want to briefly address a choice I made very early on in designing the interview schedule for survivors. This choice was to not ask about their experience of sexual violence. My working assumption was that participants would have experienced trauma both from the incident(s) of sexual violence and from their university's response to their disclosure. Given the focus of this research on institutional responses and to minimize the emotional toll of the interview, I focused the questions only on the university response. It is important to note, however, that I did not prevent participants from discussing their experience of sexual violence if they chose to do so, and some of them did decide to talk with me about this.

Participant Safeguarding before and during Interviews
Safeguarding measures before the interviews took the form of alerting participants to the potentially distressing nature of the research and creating an individual safety plan with each participant. Based on Newman and Kaloupek's (2004) suggestion, I included a clause in my informed consent sheet which acknowledged that people may find participation distressing; this action respected the autonomy of my potential participants, as only they knew the level of upset with which they were comfortable and could decide for themselves if they would have liked to participate despite possibly becoming distressed. This is one of the first ways in which I attempted to balance safeguarding and agency in my methodological design.
Perhaps the clearest example of attempting to enact this balance is the individual safety plans I created with participants prior to the interview. Fontes (2004) suggests that, before beginning an interview, interviewers and interviewees should work together to make a clear guide on how interviewers should proceed depending on interviewee reactions. My adaptation of this plan involved asking participants several questions before the start of the interview: how they would like me to respond if they started crying, how they would let me know if they did not want to respond to a question, and what they planned to do if they realized they wished they had not participated following the interview. This plan was tailored to every participant and, as such, acknowledged the different stages of recovery survivors could be in when speaking to me, which Campbell et al. (2009) found to be important in their research about interview practice with rape survivors. Many participants-staff and student-were slightly taken aback by these questions and awkwardly laughed it off before answering, but several students thanked me for the consideration; in these latter cases, I think this guide created rapport between us before the interview even began. In the former cases, I do not think that the safety plan negatively impacted how participants and I related to one another: Nobody expressed discomfort or appeared unsettled, and all participants did go on to answer the questions. In light of this, I believe these reactions were because these questions are not commonly asked and it took people a few minutes to process them; some also said that they had never considered what is helpful for them when they start crying, so they had to stop and think before answering.

Feminist Interviewing Techniques as Support Measures
During the interviews, I used feminist interviewing techniques to make the conversation as equitable and supportive as possible. Alongside their (2009) research, Campbell and colleagues also conducted a (2010) "qualitative metastudy" that examined feminist interview techniques in interviewing adult female rape survivors, and these findings greatly influenced my interview conduct. Feminist interviewing attempts to topple the implicit privileging of the interviewer over the interviewee-which results from the structure of the interview method itself (Oakley 1981) by positioning the researcher as subject and the participant as object, or some thing to study (Gorelick 1991)-by actions such as encouraging mutual dialogue (Campbell et al. 2010). In interviewing survivors, Campbell et al. (2010) emphasized that the participants-not the researchers-controlled the conversation, and interviewers reiterated this fact to them often. Enabling participants to drive the interview is particularly important in sexual violence research, as it gives them power in speaking about an experience in which another person (or persons) took away their power. The claim 'driving the interview' should perhaps be mediated in the context of my own research as I used a semi-structured interview schedule: Although I was flexible in the order of the questions I asked, so that I could follow the lead of the participant, it was still I myself as the researcher who began the interview by asking them questions, asking the same questions to all participants, and signalling when I thought they might need a break or would want to skip a question. While I would like to say that participants had more power in the interview than I did because I attempted to mitigate positivist disempowering interview practices by using feminist interview techniques, it is important to acknowledge that the very structure of academic research still prioritizes the researcher over the research participant (Gillies and Alldred 2012). Fontes (2004) also addresses the need to stress participant control and suggests that researchers should give clear points during the interview that offer participants the ability to skip a question or stop the interview altogether, which makes it less probable that researchers will coerce interviewees into participating beyond that with which they are comfortable. I utilized both of these techniques to stress to my participants that they were in control of our conversation, and to try to make them feel as safe as possible speaking with me. Several students did decline to answer questions, but this was a small portion of the 19 who spoke with me. While I did not get the sense that students answered questions that they did not want to, by virtue of participating in a research interview they may have felt like they should answer all my questions in any case. I shared many attributes in common with most participants as I was around their age, white, a woman, and a student; though this could have increased rapport between us if they saw me as a member of their group, our shared identities also could have potentially made it harder for them to decline to answer questions (Hesse-Biber 2007).

Flexibility and Implementing Participant Feedback
I remained flexible in implementing participant feedback during data collection. I owe a significant part of my interview process's final form to Dylan 3 , a non-binary student participant from an English university. In our interview-which was only my second student interview and fourth interview overall-they challenged the phrasing of a question which asked how confident they were prior to reporting that their university would be able to help and support them. They responded, "I thought before I reported that they would be able to help me. After I reported, I think they are able to help me but I don't think they're willing to help me, and I think that's a useful distinction to make." Based on their feedback, I changed the wording of the question in all subsequent student interviews to "able and willing to help and support." Dylan also mentioned at the end of our interview that they were very nervous beforehand because, despite the call for participants mentioning that I would not ask anyone to discuss their experience of sexual violence, they were not entirely sure that that would be the case. Their case as well included a highly convoluted paper trail with different university offices, and they said that having the option to see the questions beforehand would have eased their worries because not only would they have seen that I would not ask about sexual violence, but it also would have enabled them to get all relevant emails in order prior to the interview instead of searching their computer throughout it. I took this advice onboard and subsequently offered each student participant the option to see the questions beforehand. Most students did then look at the question list before our interviews, and some even came in with timelines and relevant notes like Dylan wished they could have had. While flexibility and following survivors' lead reflects feminist interviewing, I also wanted to take this space to recognize Dylan's contribution to this research from an ethical perspective that sees feminist research as collaborative. Thank you, Dylan.
Sharing interview schedules prior to the interview enabled survivors to prepare themselves, but again challenges the idea of survivors driving the interview: Even though I may have followed the direction of each participant in terms of asking follow-up questions when they organically made sense in the conversation, I was still driving the interview by setting the interview schedule and in turn participants' expectations of what we would discuss. This contradiction, however, does not make the practice less valuable, as it was implemented following survivor feedback and I continued to receive feedback from other participants about this practice alleviating their nervousness around the interview. In other words, even though implementing Dylan's feedback and offering to share interview schedules with survivors did not fundamentally eradicate power differentials between me as the researcher and my participants, it is still good trauma-informed practice.

Use of Skype in US and English Student Interviews
As previously discussed, many of the interviews occurred over Skype. Skype is particularly effective at reducing barriers in international research such as the cost and logistics of travel to distant cities or countries (Mirick and Wladkowski 2019), as was the case in this comparative study. Although some scholars have expressed concern about establishing rapport with participants over Skype, using the video call feature on Skype allows participants and researchers to see non-verbal cues and body language from the shoulders up, which fosters rapport similar to that of in-person face-to-face interviews (Mirick and Wladkowski 2019;Weller 2017). This was definitely the case for me, as all participants-save for one (Dylan, the aforementioned English non-binary student)-who spoke with me via Skype used their video and I mine; I did not feel it was more difficult to establish rapport over Skype compared to in-person interviews, and none of my participants expressed or appeared that they felt awkward using a video call for the interview. In fact, in comparing two interview-based studies that used a mix of in-person face-to-face interviews and Skype video call interviews, Jenner and Myers (2019) found that "Skype interviews may actually be advantageous for discussing deeply personal or sensitive topics" (p. 176). Skype enables participants to choose a location where they have the flexibility, privacy, and comfort to discuss difficult topics, and this setting can facilitate more open discussion than face-to-face in-person interviews in public (Jenner and Myers 2019). Although I did not conduct a metastudy of interview responses like Jenner and Myers (2019), I did not get the sense that students speaking with me over Skype were holding back in their responses to me; this may be a result of the closeness in age and/or student status I shared with many participants. Additionally, while technical issues may arise in Skype interviews, such as poor internet connection or dropped calls, I encountered only a few of these and they were quickly fixed, which ultimately did not seem to negatively impact rapport or interview timing.
I found Skype to be an invaluable resource for interviews with survivors of sexual violence. Beyond the convenience and pragmatic necessity informing my usage of it, I had previously conducted interviews with survivors via Skype and was aware of its benefits, especially around increasing survivor agency in the interview process. I use agency here to refer to both choice and control, which are two elements that are fundamentally absent in instances of sexual violence when someone removes a victim's ability to say no and retain control over their body. Skype offers choices to those using it: Participants can choose to show themselves on camera or to call only with audio. I did not request that all participants use video calling and, as mentioned, one participant did speak with me only via audio. I also stressed at the beginning of each interview that they could switch to audio only at any point should they feel more comfortable doing so, yet all participants who began the interview with their camera on kept their camera on throughout the conversation. Although a single participant engaged with the audio-only element of Skype, in using the platform in this way, I attempted to give participants control over what it was that I saw (i.e., where the interview took place, whether I saw them or listened) and when I saw them (e.g., if they began crying and wanted a moment of privacy to gather themselves, they had the opportunity to turn the camera off and recollect themselves if they preferred I did not witness that). Interviews via Skype that allow audio or video calls or a switch to audio calling after a period of video calling can enable survivors to engage in presentation management or exert choice in ways that would not be available in in-person face-to-face interviews. Even when survivors do not engage with these options, the presence of them can contribute to a safer interview environment.
While I had a very positive experience using Skype for interviews, other researchers may not find it as useful. Seitz (2016) notes that the limitations of Skype include technical difficulties, dropped calls and connection issues, inaudible moments, limitations in reading nonverbal cues and body language, and potential loss of intimacy. Beyond these limitations with the technology itself, researchers working with populations who are elderly, without access to or not as familiar with technology, and/or without a private place to speak may not find using online video calling as intuitive as I did in this study.

Participant Engagement Post-Interview
All participants had the opportunity to comment on their transcribed interviews, at which point they helped identify any unintelligible comments on the recording, highlighted any information they felt could be identifying, clarified statements, provided pseudonyms to others named, or updated me on any changes in their cases if they were actively engaged in university response processes. I received either confirmation of the initial transcript's acceptability or additional comments and/or edits from all but one staff member, who was experiencing exigent circumstances in her workplace and allowed me to use the initial transcript without reviewing it, and from all but two student participants, both of whom did not respond after I reached out two times over the course of a month following the initial email. Once I finished transcribing the interviews, I sent the transcripts to every participant without the participants having to request them; upon receiving their transcript, participants had two weeks to review and comment on it before I would use my existing copy for data analysis, and I explained this timeline to them in the consent forms as well as at the end of the interview. This timeline was flexible, as some survivors did say that they were not able to look at it just yet. As mentioned, only two did not respond after several attempts asking if they would like to comment on the transcript.
By offering participants a chance to review, edit, and comment on their transcripts prior to data analysis, I hoped to offer a chance for collaboration with my participants, which is a tenet of feminist interviewing. I did not contest any notes that survivors gave me on the transcripts-which ranged from adding more or new information about their cases if they were ongoing, to correcting instances where they misspoke and could double-check email trails, to asking me to remove certain identifying characteristics of themselves or their universities-and applied all of their comments prior to my analysis of the transcripts. Two survivors expressed, either in email communication about the transcript or in their transcript annotations, that they wish they had the opportunity to comment on the transcripts of their student conduct investigations, and felt that commenting on the research transcripts was a meaningful opportunity for them. As Gillies and Alldred (2012) note, though, researchers are still responsible for meaning-making. In order for this practice to be truly collaborative, I would have had to engage participants in the analysis phase of the research, which I did not. I did, however, regularly check that my interpretation of their account was correct in the interviews themselves: This looked like me paraphrasing what I had heard, potentially framing it in a certain way based on theories informing my research if applicable, and directly asking if that interpretation was accurate and fair. In this way, I engaged survivors in meaning-making in real time during the interviews, even though survivor participation at the stage of transcript review was limited. Checking informal meaning-making does not eradicate the 'problem of speaking for others' when it comes to writing up qualitative data, especially about groups that I as the researcher may not be a member of (Alcoff 2008), but it does mitigate it.

Conclusions
When conducting researching with survivors of sexual violence, two key methodological issues emerge in tension with one another: the need for safeguarding and the preservation of survivor agency. Safeguarding ensures that there are appropriate resources and support in place for survivor participants, while agency refers to participants' ability to make informed decisions for themselves in response to the structures in which they operate (Emirbayer and Mische 1998). As Burgess-Proctor (2015) highlights, safeguarding requirements in research with survivors of violence can be the very measures that circumscribe survivor agency. In this article, I have drawn attention to the ways in which safeguarding measures and agential opportunities for engagement worked with and against each other in part of my methodological framework for interviewing survivors of sexual violence. These parts of the framework included information on the consent form, safety planning prior to interviews, offering to share interview schedules with survivors, emphasizing their right to skip questions or end the interview early, mentioning that they could use Skype however was most comfortable for them (e.g., interview by audio only), and reviewing transcripts before data analysis.
Looking back three years after data collection, I realize that this tension between safeguarding and agency in my research speaks to a much larger issue within gender-related violence research more broadly. Although I attempted to offer opportunities for survivors to become collaborators in the research process, such as through checking transcripts and initial meaning-making, these opportunities were necessarily constrained by not only safeguarding procedures, but also the process of conducting research itself. This experience echoes Gillies and Alldred's (2012) reminder to be aware that, within the current paradigm of western research, no matter how emancipatory or collaborative we attempt to make our methods, we as researchers still ultimately occupy the more powerful position compared to our participants. This is not to say that balancing safeguarding measures with avenues for participants to share power in the research process is a futile endeavour. Merely because agential opportunities are constrained does not abdicate us as researchers from the responsibility to create them wherever possible, or make these opportunities less important for the survivors who share their experiences with us. As long as the western academy is organized the way it is, gender-related violence researchers will continue to grapple with this tension between safeguarding and agency. Ultimately, there is scope for innovative methodological interventions in the future to advance our understanding of these competing needs, and better serve survivor participants in the process.
Funding: This research received no external funding.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Department of Education Ethics Committee of University of York (approved 18 April 2018).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to their sensitive nature and their containing information that could compromise the privacy of research participants.

Conflicts of Interest:
The author declares no conflict of interest.

1
Although it is common to refer to England as part of the United Kingdom, this study focused solely on England as an individual country, and not the other three U.K. member states of Northern Ireland, Scotland, and Wales, as higher education is a devolved policy issue across these four countries (Department for Business Innovation & Skills 2011). 2 I include the descriptor of Ivy League to indicate a level of prestige and intense research activity associated with a handful of US universities. Due to the elite status these universities hold, they receive a lot of media scrutiny from mainstream news outlets, and this scrutiny extends to sexual violence cases, which in turn can impact response. These Ivy League universities are comparable to research-intensive universities in England.