Researching Distressing Topics

Qualitative researchers who explore sensitive topics may expose themselves to emotional distress. Consequently, researchers are often faced with the challenge of maintaining emotional equilibrium during the research process. However, discussion on the management of difficult emotions has occupied a peripheral place within accounts of research practice. With rare exceptions, the focus of published accounts is concentrated on the analysis of the emotional phenomena that emerge during the collection of primary research data. Hence, there is a comparative absence of a dialogue around the emotional dimensions of working with secondary data sources. This article highlights some of the complex ways in which emotions enter the research process during secondary analysis, and the ways in which we engaged with and managed emotional states such as anger, sadness, and horror. The concepts of emotional labor and emotional reflexivity are used to consider the ways in which we “worked with” and “worked on” emotion. In doing so, we draw on our collective experiences of working on two collaborative projects with ChildLine Scotland in which a secondary analysis was conducted on children’s narratives of distress, worry, abuse, and neglect.

It has been widely acknowledged that qualitative researchers who explore sensitive topics may expose themselves to emotional distress (Bloor, Fincham, & Sampson, 2007;Dickson-Swift, James, & Kippen, 2005;Dickson-Swift, James, Kippen, & Liamputtong, 2006Johnson & Clarke, 2003;Lee-Treweek & Linkogle, 2000). Hence, qualitative researchers are often faced with the challenge of maintaining emotional equilibrium during the research process (Coles & Mudaly, 2010;Woodby, Williams, Wittich, & Burgio, 2011). Nevertheless, discussion on the management of difficult emotions has occupied a peripheral place within experiential accounts of qualitative research practice (Holland, 2007). There are, however, some notable contributions to our understanding of the emotional lives of qualitative researchers and the strategies researchers use to manage emotional states. For instance, Kleinman and Copp's (1993) Emotions and Fieldwork is a seminal text on these issues. However, with rare exceptions, the focus of published accounts is concentrated upon the analysis of emotional phenomena that emerge during the collection of primary research data. Such accounts are focused upon the emotional challenges of research encounters that require interaction with research participants. By comparison, there is an absence of an academic dialogue around the emotional dimensions of working with secondary data sources (Fincham, Scourfield, & Langer, 2008). This may reflect an assumption that secondary analysis lacks the interpersonal context within which emotional distress can arise (Fincham et al., 2008), the possibility that secondary analysis is considered to be less emotionally demanding (Woodby et al., 2011), and the lack of a research culture in the secondary analysis of qualitative data more generally (Corti & Thompson, 2004).
Given the absence of interaction with participants, it is not difficult to understand why there has been a neglect of the emotional dimensions of secondary analysis or why secondary analysts, unlike fieldworkers, have not been presented as 490705S GOXXX10.1177 1 University of Dundee, UK 2 University of Edinburgh, UK "emotional agents" (Kleinman & Copp, 1993, p. 54) in research accounts. This article highlights some of the complex ways in which emotions enter the research process during secondary data analysis. In doing so, we draw upon our collective experiences of working on two projects in which secondary analysis was conducted on both quantitative and qualitative data extracted from the call records of ChildLine Scotland 1 (CLS), a confidential telephone counseling service for children/young people aged 18 and younger in Scotland, wherein children/young people's narratives of distress, worry, abuse, and neglect are represented. Our experiences of working with the qualitative data are the primary focus of this article.
The first project examined children/young people's concerns about the health and well-being of their parents and significant others, and the second investigated children/ young people's concerns about their sexual health and wellbeing. Both projects involved us reading and analyzing children/young people's accounts of physical and sexual abuse. The analysis of narrative accounts in which children/young people disclosed these experiences were particularly challenging, and it is with the emotional challenges of analyzing these that we reflect upon in this article.
During the course of the two projects and beyond, we discussed, wrote, and reflected upon our experiences of working with CLS data, and we draw upon these in this article. We approach this article as both novice and experienced researchers. Kathryn was the principal investigator on both projects and had extensive prior experience and expertise in qualitative research in the area of children and families. Sharon was employed as the research fellow on the first project, and then as the research fellow and coinvestigator on the second project. Elinor joined the research team during the second project as a research fellow when Sharon left her post to take up a permanent post in another institution. Sharon remained involved in the second project as a coinvestigator. Sharon was responsible for the data collection and analysis task in the first project and the data collection task during the second project. Elinor had primary responsibility for the data analysis task during the second project. Kathryn was involved in many of the analytical discussions that characterized our "analytical meetings," and was involved in reading and working with extracts of the data. For both Sharon and Elinor, these projects represented their first academic positions.
Before moving on to describe and reflect on our experiences of working with the research data, this article will connect with key strands of thought in relation to emotions and qualitative research that have influenced our analysis. We begin by locating emotion as a dimension of reflexivity. We do so because the research process that we undertook required an active engagement with emotional reflexivity, and an acknowledgment that the research process was disrupted by the emergence of distressing emotional states and the requirement to manage these. We follow this by considering Arlie Hochschild's conceptual thinking on emotional labor, which has provided us with the "analytic tools" (Kleinman, 2002, p. 381) to conceptualize our experiences. Finlay (2002) reminds us, As qualitative researchers, we understand that the researcher is a central figure who influences the collection, selection and interpretation of data. Our behaviour will always affect participants' responses, thereby influencing the direction of findings. Meanings are seen to be negotiated between researcher and researched within a particular social context so that another researcher in a different relationship will unfold a different story. Research is thus regarded as a joint product of the participants, the researcher, and their relationship: It is co-constituted. (p. 531)

Emotional Reflexivity
The importance of accounting for the multiple ways in which we as qualitative researchers influence our research is considered fundamental to achieving and preserving integrity and trustworthiness in qualitative research (Finlay, 2002). Consequently, qualitative researchers are required to invoke a visible reflexive analysis that both evidences and articulates their positionalities in relation to their epistemological and ontological positions, and their consequent methodological and theoretical frameworks (Malacrida, 2007). More than this however, qualitative researchers are also required to think about and explicitly acknowledge the various ways in which their own unique biographies enter into the research process. Thus, their lived experiences, understandings, and perspectives, and their social positioning in relation to factors such as class, gender, race, and so forth should also form part of their reflexive endeavors (Finlay, 2002).
This process of reflexive analysis is considered to involve "personal exposures" (Davis, 1998, p. 331) by the researcher that facilitate the recognition of the role of the self in the research process (Ely & Anzul, 1991;Strauss & Corbin, 1998). Reflexivity, however, goes beyond a simple accounting for dimensions of the self. Reflexive research practice also requires that researchers both acknowledge and critically analyze how these dimensions of the self enter into and influence their encounters with participants and the data that are produced. This aspect of reflexivity is connected to social constructionist understandings that consider the products of qualitative research to be co-constructed through the intersubjective dynamics of the researcher-researched relationship (Finlay, 2002). Hence, researchers are required to consider not only how the self can influence and shape the process and outcomes of their interactions with participants but also how their participants contribute to this shaping.
It has become increasingly common for qualitative research accounts to contain reflexive narratives that incorporate "personal exposures" (Finlay, 2002). However, despite the increasing trend toward "personalizing" research accounts in such ways, Blackman (2007) contends that the acceptance of emotion as a dimension of reflexivity has been slow. As a consequence, the emotional self is rarely attended to in research accounts (Coffey, 1999). Our own review of the literature reveals that while the last two decades have seen a peppering of articles that have sought to address emotions and fieldwork, reflections upon the multiple ways in which emotion is experienced and encountered remains most firmly at the margins of scholarship. Thus, issues of emotionality are only tangentially understood (Coffey, 1999), and research accounts remain largely cleansed of such detail.
Despite the neglect of emotions as a dimension of reflexivity, it has become increasingly acknowledged that reflexive engagement with the emotional dimensions of research production should be incorporated into our reflexive activities. This is considered particularly crucial where researchers engage in the examination of topics that can potentially disrupt the emotional equilibrium of researchers (Woodby et al., 2011). Central to this call for emotional reflexivity are concerns about the extent to which researcher emotion influences the production of knowledge (Malacrida, 2007). Here, as in other reflexive modes, there is a danger that the activity of emotional reflexivity may become reduced to a confessional act that involves the telling of emotional tales with a view that this in itself constitutes reflexivity (and results in the reduction of subjectivity). There is a danger that such narratives can reduce reflexive endeavors to apologies that situate emotion as antithetical to the achievement of objective knowledge. An unintended consequence of this is to position knowledge generated through interpretivist ways of knowing as less reliable than that which is positivistically produced.
Emotional reflexivity, however, is about more than just simply acknowledging phenomenon that may impede the production of an "objective" knowledge (or at least a less subjective one). It is also about more than the retrospective "telling" of fieldwork tales to others or the sharing of what may more properly be labeled as "reflections." Emotional reflexivity in qualitative research is an active process that requires researchers to invoke a reflexive agency throughout the research process. Quite often, in their narratives of emotions and fieldwork, researchers implicitly discuss active reflexive processes that occur throughout the research process. That is, researchers frequently discuss how they "work on" and manage (or sometimes fail to manage) emotion in the field through reflexive processes.
In his discussion of reflexivity and emotions, Rosenberg (1990) makes the point that reflexivity fundamentally "refers to the process of an entity acting back upon itself" and that emotional reflexivity is epitomized by the harnessing of emotions and/or emotional experiences as "objects of one's own reflection and control" (p. 3). Consequently, Rosenberg argues that the "work" that is done on emotions is achieved through reflexive processes or reflexive work. Thus, he postulates that it is via reflexive processes that the internal arousal states through which emotions are experienced (for instance, physiological or bodily sensations) come to be "worked over" (Rosenberg, 1990, p. 3). This "working over" of emotion through reflexivity is considered to happen in three ways: first, through interpretive processes wherein the identification of emotion occurs; second, through the display of emotion that may involve elements of work on the self to produce intended effects on others; and third, through the invocation of emotional experiences perhaps to induce preferred emotional states (for instance, this may happen when the emotion being experienced produces a state of discomfort or distress).

Emotional Labor
Despite the marginalization of emotion as a dimension of reflexivity, some accounts that attend to these issues derive analytical insight from Arlie Hochschild's (1983) Goffmanian informed cultural analysis of emotions. While many accounts do not explicitly situate Hochschild's thinking as a device through which emotional reflexivity can be understood and theorized, Hochschild provides a rich conceptual toolbox through which this can be achieved. Moreover, she provides a framework that resonates with Rosenberg's thinking on emotional reflexivity.
At the heart of her work has been a focus upon how professionals control the expression of their emotions in their attempts to conform to cultural or emotion ideologies around emotional responses to social phenomena and the expression of these responses in various social contexts. She argues that the acquisition of these emotion ideologies is achieved through the socialization processes that characterize the different spheres of activity in which individuals interact. Hochschild (1979) theorizes that within any given society, the totalizing effect of multiple emotion ideologies is the presence of an emotion culture at the macro level (Turner & Stets, 2005).
Drawing upon the dramaturgical sociology of Erving Goffman, Hochschild (1979Hochschild ( , 1990 argues that within any specific context, two sets of norms frame the different ideological spaces individuals inhabit. These norms are also considered to reflect the wider emotion culture of a society. The first of these norms she labels as feeling rules, which govern the kinds of emotions that are considered appropriate for individuals to feel and experience in particular situations. The second norm she refers to as display rules, which govern the parameters for the appropriate expression or performance of emotion in particular situations. These feeling rules are integral to what Hochschild (1983) refers to as display rules.
Here, the display of appropriate emotion by an individual is considered to be crucial for meaningful or successful social interaction. The central issue in Hochschild's (1979Hochschild's ( , 1983 thesis, therefore, is anchored in her observation that the management of emotion to conform to socially prescribed norms or emotion ideologies requires work that she describes as emotion work or emotion management. To invoke an analytical framework for the analysis of how individuals accomplish their feeling performances, Hochschild (1983) presents the concepts of surface acting and deep acting. In surface acting, an individual engages in the regulation of their public expression of emotion, whereas deep acting refers to the process whereby individuals consciously attempt to alter internal states (feelings) to achieve the appropriate outward expression of emotion.
The central focus of Hochschild's body of work has been the management of emotion in organizational/commercial life, and it is this locus that has given Hochschild's work its enduring potency. Developed out of her ethnographic study of flight attendants, Hochschild (1983) eloquently outlined the processes whereby the management of emotion by flight attendants was intimately connected to and determined by the emotional ideology of the organization (airline). Here, the management of emotion was conceptualized as a vehicle through which airline employees invoked expressive devices to achieve organizational goals such as customer satisfaction.
The management of emotion for organizational or commercial gain was described by Hochschild (1983) as emotional labor. For Hochschild, however, the process of emotional labor is conceptualized as ultimately producing states of alienation within employees. She proposes that the consistent effort required by individuals to control their emotions results in burnout and stress particularly where individuals experience continued emotive dissonance (Hochschild, 1983) between their internal feeling states and private emotion systems, and the [organizational] emotion ideologies that dictate the performance of altered public expressions of emotion (Hochschild, 1983).
The concept of emotional labor has enjoyed widespread currency within discussions of emotions and research. It has entered the reflexive narratives of qualitative researchers seeking to theorize their experiences with emotion. Within such accounts, the concept of emotional labor has primarily been invoked as a conceptual device to explore the emotional dimensions of fieldwork experiences. As a consequence, it has become somewhat commonplace to conceptualize the labor of qualitative researchers as emotional labor.
The central focus of much of this work is centered upon the analysis of the emotional work that underpins encounters with participants and in particular the extent to which researchers are required to offer emotional performances within these encounters. For instance, Malcolm (2012) in her recent account of the emotional labor of the contract research fellow researching adult learning discussed how she found that emotional labour was performed in the detailed activity of co-constructing a life, of asking, listening to, and reflecting back the emotions of the interviewee. (p. 8) Malcolm (2012) goes on to describe how, during her fieldwork experience, the empathy she was required to display required her to engage in a considerable level of deep acting to perform as "good" researcher. She describes how she had to suppress feelings of sadness and enthusiasm when interviewing a participant and how deeply conflicted she felt when her own ethics of care were placed in a tense and contradictory relationship with the requirements of data collection and the performance this required. In doing so, Malcolm (2012) discusses how she experienced the requirement for a complicit emphatic performance to elicit data as emotionally challenging.

Secondary Analysis and Emotional Labor
Commentary of the kind Malcolm (2012) offers are commonplace within accounts of emotion and fieldwork in qualitative research, and the concept of emotional labor is widely drawn upon to provide analytical power. Moreover, where the concept is not invoked, similar experiential phenomena are frequently described.
However, not only has emotion been marginalized and neglected as a dimension of secondary analysis, but to our knowledge, the concept of emotional labor has also not been applied to understand the emotional work of secondary analysts. This may be because the conditions in which emotional labor is considered to be performed appear to exclude a consideration of the labor of the secondary analyst as emotional labor. In her definition of emotional labor, Hochschild (1983) describes emotional labor as work that 1. Requires face-to-face or voice-to-voice contact with the public 2. Requires a worker to produce an emotional state in another person 3. Allows the employer, through training and supervision, to exercise a degree of control over the emotional activities of employees. (p. 147) The first two of these three dimensions immediately situates secondary analysis as out with the analytical reach of emotional labor for quite obvious reasons-the absence of interpersonal encounters with research participants. In most instances, secondary analysts interact with documents or other nonhuman artifacts and not research participants. This would suggest, therefore, that there is no requirement for secondary analysts to produce emotional performances and thus no requirement for them to invoke such devices as deep or surface acting.
It is our contention, and it has been our experience, that the research procedures that characterize secondary analysis can create the conditions in which an intimate and deeply felt interpersonal connectedness with participants arises. Here, we would argue that the narrative or text as the object of analysis should not be considered as a passive artifact. We experienced the narratives that we analyzed as possessing an affective agency wherein the "narrated participant" (the child or young person) was communicated, was emergent, and was brought into an interactional and dynamic relationship with ourselves. In this sense, analytically and conceptually, the narratives we worked with could be understood as mediating devices through which the relationships between ourselves as researchers and our participants was co-constructed. Thus, it was through analytical interaction with each individual narrative that we were brought into a relationship with the child/young person the narrative (re)presented. It is through these narratives that our participants spoke and shared with us their most acutely felt distress and worries.
In these conditions, the management of emotion emerges primarily as a private experience within the analytical and interpretive task of making sense of these narratives. As affective artifacts, these narratives act upon the secondary analyst inducing emotional states that require private emotional performances "backstage." These emotional performances require both deep and surface acting in order that the analyst can perform as a researcher and accomplish the analytical and interpretive tasks necessary to produce the research outputs required through their emotional labor.
It is our argument, then, that secondary analysis of the narratives we worked with required us to invoke an emotional reflexivity. We turn now to our discussion of these processes across the two projects to illustrate our experiences of doing so.

Working With CLS Data
Many of the records we analyzed across the two studies revealed details about the different forms of abuse that children/young people were suffering at the hands of their parents and others who held a duty of care toward them. The nature of the sometimes multiple abuses they described was overwhelming in terms of the severity of the abuse and its frequency. The accounts given by many of these children/ young people can only be described as harrowing. The data analysis task involved us reading record after record after record that represented, often in the child/young person's own words, lives that were shattered by incomprehensible acts of cruelty, violence, and neglect. The detail given was often graphic, containing explicit and vivid descriptions of what children/young people were experiencing, and the often dire circumstances in which they lived. A typical example is given below: Dad tries to put his hands down my trousers when mum goes to bed. He told me not to tell anyone. He hits me, slaps me and punches me. I have bruises on my face and two black eyes. (Female child,11 years) The data that were analyzed were unique in being unmediated by researcher interests insofar as they represented children/young people's self-identified concerns. Research that has investigated the experiences of people who have been abused is usually "survivalist" in the sense that the abuse experienced can be viewed as historical. As such, the data gathered usually represents the retrospective accounts of adult survivors. The records we were analyzing were collected from children/young people who were often experiencing abuse at the time that they placed the call to CLS. The records tended therefore to be written in the present tense reflecting the immediacy of the child/young person's circumstances. The impact of this cannot be understated. In reading these accounts, there was a sense that you were bearing witness to the child or young person's trauma-as it was happening. The team's emotional experiences of working with this data resonates strongly with Moran-Ellis's (1996) reflections on reading accounts of child sexual abuse and her observation that she felt "much pain by proxy" (p. 181).
It is extremely unusual for researchers to access the accounts of children/young people who are currently experiencing abuse or who have recently experienced such abuse. Children are not empowered to speak about such things either in a research context or in the wider sociocultural context in which they live. There are also innumerable ethical reasons why many researchers would not pursue or would be prevented from collecting primary data on these issues from children and young people.
Over the course of the two projects, we analyzed thousands of records. Coming to terms with the numbers of children/young people describing abuse in their lives was challenging. The scale of the suffering was enormously difficult to comprehend. Both research fellows were exposed to this data for prolonged periods of time and had to work with the data intensively to deliver the research findings within the project time frames. The emotional impact of intensively analyzing a substantial number of narratives of this nature was cumulative. Elinor wrote that after reading hundreds of accounts of sexual abuse, she felt horror, anger, and a deep sadness that mere statistical accounting cannot convey.

Deep Acting
Managing the sadness, the horror, the anger, and the realization that children and young people were reporting their experiences in their thousands was incredibly difficult and felt lonesome. While we were able to enter into a dialogue with each other about the emotional demands of the data analysis task during formal team meetings, for the most part, the emotional labor we each performed was a deeply private actively. For reasons of confidentiality, we were required to work with the data primarily within CLS premises. We were therefore disconnected from each other during the substantive part of the data analysis task. Managing these emotions while working with the data in isolation required both Sharon and Elinor to engage in an ongoing process of emotional reflexivity. This involved us engaging in a process of deep acting to perform the analytical task. Here, we had to actively alter the internal emotional states we were experiencing so we could carry out the cognitive work required for data analysis.
Invoking altered emotional states was not an easy task and required a considerable level of what Rosenberg (1990) labels as "mental self-manipulation" (p. 11). On an almost daily basis, we found ourselves attempting to try to overcome and control difficult emotions in an effort to rid ourselves of them or to at least mute their intensity. Sharon recalls the enormous cognitive effort this internal work required to regulate, control, reduce, and/or displace feelings of immense sadness and bewilderment, and to quite often fight back tears. Sometimes, this involved active attempts to replace these kinds of emotions with other more pleasant sensations. To do so often required breaking from the data analysis task to provide temporary relief.
Attempting to alter emotional states is considered by Rosenberg (1990) to be a clear manifestation of emotional reflexivity. However, the act of trying to feel something different or to feel less can produce an ethical uneasiness. On one hand, there is an impetus to perform the task, to be as Malcolm (2012) states, "a good researcher" (p. 11). This connects to concerns that the analysis may be impeded or distorted by meanings produced in distressful conditions. Moreover, for Sharon and Elinor who were novice researchers, there was a strong desire to perform as "a good researcher" to other members of the research team. Demonstrating analytical progress and dumbing down personal struggles is one of the ways through which this was sometimes achieved. On the other hand, there is a personal desire to numb unpleasant feelings as an act of self-caresomething that Dickson-Swift et al. (2008) suggest researchers working on sensitive topics should be proactive about.
As necessitous as self-care strategies and deep acting might seem, there were times when Sharon considered that in trying to feel less, she was minimizing the pain that was being expressed by the children and young people. There were also concerns that a failure to engage emotionally-to empathize fully-with the voices of these children/young people could in itself produce a distorted analysis.
This confusion over exactly how one should work on emotion to perform the analytic task and function on a personal and professional level was salient throughout the research process. Despite phases of intense emotional disruption, there were also periods when we felt disconnected from the data. Far from being viewed as problematic, these instances of disaffection brought some much-needed relief, allowing us to "get the job done." However, at times we worried that this emotional disconnection from the data would influence the interpretations we were reaching. We had concerns that we had become inured to the pain of these children/young people and worried about the implications of this.
For both research fellows, their involvements in the projects represented not only their first academic research posts but also their first experiences of working with child abuse data. Fincham, Scourfield, and Langer (2008) in discussing how researchers experience their first encounters with disturbing secondary data highlight that unless a researcher has previous experience in working with similar kinds of data, their response to being exposed to such data may be similar to that of a lay person. That is, the emotional impact can be profound. Drawing on the commentary of Cohen (2001), they argue that while researchers may have been exposed to the particular topic of analysis (in their study suicide and in ours child abuse) on an almost daily basis through, for example, media images, we cultivate a denial of the human suffering involved. In the case of suicide, for instance, they argue, We might observe that there are particular social processes associated with the routine denial of mental or emotional suffering that is in our midst. We are not thinking here of faraway trauma such as famine and war, but personal suffering that is in a sense near at hand. (Fincham et al., 2010, p. 860) None of us were unaware that child abuse represents a serious social problem for children/young people. However, and as we found, there is a vast difference between a social awareness of these issues and having to confront the abuse of children/young people through being exposed to their trauma in the ways that we were.

Embodied Experiences
Sharon and Elinor both experienced challenging visual and auditory sensations. They found that at the time of reading the accounts and latterly reflecting upon them, there was an alternation between imagining or picturing the events described in the narrative and hearing the voices of children/ young people. It is incredibly difficult to describe or explain these sensations in the written word. The visual sensations were akin to commonly experienced ways in which we conjure up imagery when reading a book. In many ways, therefore, we were "imagining the text." In reading the accounts, Sharon and Elinor found themselves "seeing" the abuse happening. For Elinor, these visual disturbances were so profoundly invasive and distressing that it had a temporary impact upon her capacity to continue with data analysis-She described this as "hitting the wall." The auditory sensations were experienced as hearing the voices of children/young people narrating their experiences. Although we never directly heard the voices of the children/ young people whose accounts we read, we felt as if we heard them speak to us through these auditory experiences. We ascribed each of them with a unique voice. Where these voices came from we do not know. For Sharon, these voices can still be heard. She can still hear children/young people "talking." There are certain narratives that remain embedded within memory that can be accessed or played back, or which invade the consciousness whenever this research is remembered. There is one voice in particular that has never leftthe words of a young girl contained in a record that detailed her sexual abuse and that ended with the capturing of these words: Where can I go, who can help me? Sharon recalls that when she read these words, she was left feeling utterly impotent, and it is our experience that engaging with this kind of research never leaves you-The stories of the participants remain and the emotional impact of witnessing their trauma leaves its traces.
Embodied experiences such as these are identified by Rosenberg (1990) as being central in relation to the phenomena of emotional identification. Emotional reflexivity, he proposes, requires an active interpretation of identified signifiers, and the regulation of such embodied states requires emotion work. When people experience such states of emotional arousal without engagement in interpretive process, these are considered to be nonreflexive states. We found that we actively attempted to regulate these emotional states through a process of emotional labor, and through the strategies of deep and surface acting to induce preferred states.

Being Disempowered
Feelings of "impotency" were a marked feature of the analysis stage. This perhaps signifies where secondary analysis of this kind differs from research that involves direct communication with participants. Often, in reading children/young people's accounts, we were left with a sense of utter powerlessness. We felt that there was nothing we could do to help or make a difference to these children/young people's lives. We were party to their disclosures and their acts of helpseeking, yet we could not act, and we could not help. We were not as CLS counselors were or indeed as any other "trauma worker" who encounters children/young people in difficulty in a position to act. Working with this kind of data strips you of agency. It positions you as helpless.
As an adult, your primary moral response to a child/young person (or any other person in distress) drives you to act. The impotency Sharon felt in relation to her inability to act-to either offer comfort or counsel (as CLS counselors were able to do) resulted in her experiencing a very acute emotive dissonance between her personal moral and ethical ideology, and the professional demands upon her to engage in the analytical and interpretive task of a qualitative researcher. The ability to "sit with" these tensions required considerable resolve. At the same time, she felt angry about her perception that society had failed these children/young people and powerless in her ability to do anything about it.

Dissemination and Surface Acting
Throughout the analysis stage, we had to engage in deep acting to manage our emotional states. However, the dissemination phase required us to invoke surface acting as a strategy to reduce the expression of difficult emotions during public events. However, for Sharon speaking publicly about the experiences of these children/young people required her to manage her personal emotional states in front of an audience "on stage." Both projects had a planned dissemination phase that aimed to bring the voices of these children/young people to policy makers, practitioners, and others we identified as important stakeholders in the lives of children/young people who experience such difficulties. In the first project, we held a media launch of the key findings to invited persons from relevant agencies, governmental departments, and members of the print and television media.
At such events, professional competence is under intense public scrutiny, and one wishes to perform well. We all to a greater or lesser extent had experience of presenting research findings to various audiences; however, neither Sharon nor Elinor had previously presented research findings that spoke to the abuse experiences of children/young people. Presenting the findings was not considered problematic to the team. Previous presentations had taken place with a select audience of policy makers and had passed unremarkably as these things often do. However, in this instance, Sharon found it extremely difficult to control visible signs of emotional distress and regulate her emotional display when she was (re) presenting the experiences of these children/young people The findings are hard hitting and harrowing when orally presented. We have noted how the presentation of the data from both projects affects audiences-It perhaps disrupts their own routine denial. When the presenter speaks, the reality of these children/young people's lives arouses a stunned and somber silence that leaves the air heavy. It is almost impossible to be unmoved by what is heard. The presentation for this media launch required Sharon to read aloud the following extract from a young boy's account of the abuse he was suffering and what was happening in his family life: Always hitting him, when gets in dad usually drunk. Also takes drugs-hash, acid, ecstasy. Has bruises often. Also battering mum. Mum told him not to tell social worker about what's happening. He would like to move out or run away. It's been happening for months. Mum told him not to tell anyone out of the family. (Male child,14 years) The research fellow struggled to read this extract without breaking down in tears. In this instance, Sharon was unable to "work on" her emotional expressional-Her attempt to engage in surface acting failed, and she worried that her professional performance had exposed her as incompetent in this academic task. She has never read an extract out loud again. She is unable to, preferring instead to invite the audience to take a moment to read the extract presented. Each time it is encountered, it is experienced as distressing. Sharon has reflected upon the reasons she cannot read this or any other extract aloud without emotional display but is unable to provide an explanation. Perhaps it is because the act of speaking on behalf of these children/young people-giving voice to their distresses-transforms these narratives from text to reality. Perhaps it is because the individual narrative of each child is a representation of the pain of thousands of others. We did, however, feel that dissemination gave us back an effective agency. It allowed us to feel that in some small way, we were able to render visible the struggles of these children/young people and make a difference.

The Importance of Emotional Support
It is important, however, to recognize that a significant part of our task also allowed us to access some of the most inspiring and humbling examples of human resilience, courage, and compassion that we have had the privilege to bear witness too. Many of the children/young people's accounts we read contained rich detail on the strategies they used to get by in difficult circumstances, of the actions they took to help or protect parents, siblings, friends, and other significant persons in their lives. It also revealed to us the depth of their own emphatic understanding of the difficulties that even abusive parents faced in their daily lives.
Much of what we encountered here challenges perceptions of children/young people as lacking in agency or as entirely helpless in the face of adversity. That is not to dismiss the very real difficulties that many of these children/ young people faced nor is there any intention here to suggest that any of these children/young people were not in need of adequate levels of care and protection-Rather, it is a testament to the strength of character that many of these children/ young people appeared to have. Support from others-particularly supervisors-is considered to be extremely important in militating against the negative effects of witnessing the trauma of others (Conrad & Keller-Guenther, 2006). Here, there is considerable research to support the hypothesis that individuals are less susceptible to both the shorter and longer term consequences of exposure to trauma if they have an adequate opportunity to discharge the emotional distresses they experience (DePanfilis, 2006). In academia, support structures for researchers who work on emotive topics and who encounter emotional challenges in the course of their professional work are all but absent despite the efforts of others who have highlighted this as an area of concern (see, for example, Dickson-Swift et al., 2008). There is however little recognition that secondary analysts can also be emotionally affected by the work they undertake, and we would suggest that a dialogue on these issues is overdue. Sharon and Elinor were privileged in being able to access support when working with the data within CLS premises through the mechanism of formal debriefing sessions available to all CLS staff, and we were also able, as a team, to disclose and discuss the nature of this data, and our experiences with it among ourselves. Many researchers are perhaps unable to do so.
It is also important to recognize that children's emotional distress can also be considerably alleviated where there is a receptive context (Focht-Birketts & Beardslee, 2000). To some extent, each child presented to us in the narratives we analyzed had found a receptive context-ChildLine. We are in little doubt that the service has made a considerable contribution to the lives of hundreds of thousands of children who otherwise would have had no one to share their own emotional burdens with. Moreover, when we experienced such phenomena as emotive dissonance and we felt impotent to act, we were able to remind ourselves that these children/ young people had sought and received help. This assisted us in moving forward with our task and muting the distance between the emotions we encountered and our professional objectives.

Emotional Reflexivity and (Re)presentation
Thus far, we have considered the kinds of ways in which emotions entered the research process and the strategies that we used to manage our emotional states at different stages in the process. In doing so, we have commented on our concerns that various emotion states may have led to a distorted analysis. However, while our reflexive dialogue has communicated an acknowledgment and understanding that emotions affect researchers and require management, our analysis of how emotions affect the products of research requires further consideration.
Reflexivity is routinely used in qualitative accounts to address questions of representation-that is, to illuminate upon the ways in which the subjective elements of the researcher are implicated in the process of knowledge production and the representations of the participant worlds our analytical endeavors create. For the most part, accounts that attend to this rarely move beyond acknowledgment. Hence, the concept of reflexivity that is invoked is necessarily impoverished and rarely extends beyond considerations of the co-construction of data in participant-researcher interchanges. Much less attention is given to the examination of reflexivity as it appertains to the interpretive process of (re) presentation that characterizes data analysis, and here, the ways in which reflexivity is operationalized-how it is practiced-appear elusive (Mauthner & Doucet, 2003;Pillow, 2010). Therefore, while we may acknowledge emotional contaminants, and speak to and describe their identification through reflexive interpretive processes, and the management of these emotions during analytical processes, we are much less likely to extend our treatment of reflexivity to consider how emotions act upon the analysis of our data and the creative enterprise of representation. Failure to attend to this reduces emotional reflexivity to a dimension of emotional labor-that is, as a vehicle through which emotions are interpreted, isolated, and dealt with through management strategies. At the very best, this places limits on reflexivity as something peculiar and separate from data analysis and, at worst, suggests that once worked on through emotional labor, they are somehow "hoovered out." Moreover, there is a danger in this version of emotional reflexivity that researchers can somehow "free" data from emotion.
This brings to the forefront the question of how emotional reflexivity in data analysis is practiced and incorporated into analytical procedures, and how it may come to shape (re) presentation. Here, there is little to guide the researcher on how emotional reflexivity (or any other version of reflexivity) is carried out that goes beyond acknowledging the (emotional) context in which knowledge is produced through analysis. A notable exception to this is to be found in Mauthner and Doucet's (2003) account of reflexivity in qualitative data analysis. Here, they describe a method of data analysis that they consider facilitates the operationalization of reflexivity during data analysis that draws upon Brown and Gilligan's (1992) "voice-centred relational method of data analysis." This method of data analysis is grounded in feminist methodology. Based on the work of Gilligan (1982), this approach to analysis facilities the acknowledgment and analysis of both the participant's story and the researcher's role in relation to the interpretation of that story. This recognizes that (re)presentations of participant worlds are co-constructed during data analysis and brings into the analysis the voice of both participant and the researcher bringing to the fore the relational positioning of the researcher to the text (Paliadelis & Cruickshank, 2008). This method requires three separate but related readings of the text. The first involves "reader response," wherein the analyst "reads for himself/herself" in the text identifying his or her positionalities in relation to it. This requires a reflexive analytical consideration of how he or she is responding emotionally and academically to the text and how he or she is interpreting the data in light of this. This reflexive analysis requires the researchers to isolate how their personal biographies, social and cultural assumptions, academic knowledge, and emotional responses as sources of knowledge connect with their interpretation of the data at the time that analysis is performed. This is considered to reveal the act of interpretation as multilayered and informed by these different and interacting sets of knowledge.
The second reading of the text is focused upon the participant narrative. It places as central the voices of participants, their thoughts and feelings, their perceptions of themselves, and their life worlds. This cycle of analysis is focused on allowing the participant to speak "before we the researchers speak of them" (Paliadelis & Cruickshank, 2008, p. 1449. The third reading of the text focuses on the relationships of participants, and the fourth on the contexts surrounding their relationships (Brown & Gilligan, 1992). These third and fourth readings provide the researcher with understanding how participants locate themselves in relation to others and the wider sociocultural contexts of their lives (Paliadelis & Cruickshank, 1998).
In adopting this method of analysis, the role of the research is rendered more explicit as their values, beliefs, and emotions are situated as a central part of the analytical process and are reflexively considered in relation to the other three interpretive readings of the text (Brown & Gilligan, 1992;Hewitt, 2007;Mauthner & Doucet, 1998). This approach to data analysis has been successfully used in a number of studies to operationalize reflexivity during the analysis cycle (see, for example, Mauthner, 2002;Mauthner & Doucet, 2003).
In our studies, we did not use a structured method of analysis such as the voice-centered relation method; however, the processes we went through in invoking emotional reflexivity in our analytical endeavors resonates strongly with the processes involved in the first cycle of the method. Our first readings of the narratives of children/young people were very much centered on our emotional reactions to the texts. The voices of the children/young people were latterly heard. We were able to consider during the process of analysis how our emotions and biographies as both women and academics were inextricably woven into the fabric of our interpretations and how in light of our internal and shared reflexive dialogues around this, our interpretations could shift and change.
A very good example of this was our exploration around perpetrators of abuse. In our study, the vast majority of perpetrators were identified as known to the child, and many of these were parents. Our horror at the level of abuse being reported and our abhorrence in relation to the extent of intrafamilial abuse clouded how we considered parents in relation to this. As a parent, Sharon struggled in comprehending the level of parent perpetrated physical and sexual abuse. As a parent herself, understanding how a parent could harm his or her child in such ways restricted the extent to which she could conceptualize the circumstances that can predispose parents to abuse. In particular, as a mother, female-perpetrated abuse was experienced as particularly problematic to grasp. Through reflexive analysis, Sharon was able to identify that her interpretations of the data related to sexual abuse perpetrators was limited, first, by her absorption of a cultural discourse around the "perpetrator as stranger"; second, by a cultural discourse that portrays a stereotype of perpetrators of sexual abuse as sadistic pedophiles; and third, by a conceptualization of motherhood that emphasizes gendered norms of care and romanticizes motherhood and femininity. Hence, her sensibilities around whom sexual abuse perpetrators are were considerably disrupted, and her interpretive abilities to analyze the context in which abuse may occur were constrained.
The research questions and our ontological positioning of children also obscured our analytical gaze in relation to perpetrators. In the first instance, our analytical focus guided by the research questions was to understand how children/ young people's perspectives and how the adversities they experienced affected their lives. Thus, we were attempting to place their voices as central. In doing so, we were focused on their struggles and therefore narratives of the "self." Our gaze therefore was not initially extended to a consideration of perpetrator struggles. In exploring the assumptions that underpinned her reactions to the data around perpetrators, Sharon was able to shift her perspective from one that initially predisposed her to considering perpetrators as abominable beings and dislocate their being from their acts of abuse to reconsider how parents become perpetrators and more thoroughly examine the contexts in which abuse arises.
Within the context of both projects, this shift emerged as being vitally important in revealing that most parental perpetrators did not fit the stereotype of the sadistic abuse or the pedophile "out there," but resituated many of them as persons whose capacity to love and care for their children had been diminished through circumstances of adversity that grew a context for abuse to occur. Sharon found that the narratives of children/young people provided rich, detailed, and often empathetic insights into parental troubles. Contained within these were accounts of once loving parents whom in a context of adversity had lost their capacity to act toward their children in loving and responsive ways. For instance, it emerged that parent's struggling with grief, divorce, depression, relationship problems, and other health and well-being issues were often set on a trajectory that ultimately led to abuse. In many instances, such as grief or divorce, this had led to depression sometimes in combination with increasing reliance on alcohol, and this domino effect rendered the child vulnerable as a consequence of diminished parental capacity. Inappropriate sexual relations with children revealed themselves to occur in some instances in the context of comfort seeking from a child that led to the development of an unacceptable level of intimacy with the child and ultimately a sexually abusive relationship dynamic.
While such interpretive insights did not lead Sharon to view abusive acts as less reprehensible, they did allow for the development of an analysis of the context in which abuse occurs that raised the need for parental support at times of parental vulnerability. Moreover, it brought to the surface that children/young people themselves had the cognitive and emotional capacity to enter into a critical reflection of their parents' struggles in their attempts to construct a coherent narrative of how they came to be abused.
This example illustrates how engagement in (emotional) reflexivity during data analysis can profoundly shift our interpretive endeavors and, as a consequence, our (re)presentations. Whether this results in a "better" analysis is perhaps more difficult to identify. At best, we suggest that it can lead to a more informed, nuanced, and often different analysis than that which might occur if emotion and other sources of knowledge are not reflexively engaged. Either way, it reveals that analysis is messy and that findings can be "thoroughly contaminated" through the interweaving of the researcher with the knowledge that is produced (Ellingson, 1998, p. 494).

Reflexive Writing, Ethics, Emotion, and Secondary Analysis
The research process that characterized both of the studies discussed in this article required the research team to bear witness to some of the worst elements of human destructiveness. It is unsurprising, therefore, that we were emotionally affected by our engagement with the data. Yet, we remain perplexed by the absence of academic engagement on the emotional impact of working with secondary data sources of this kind and the lack of published experiential accounts from researchers that describe and analyze how they manage challenging emotions during the research process. Presently, the restrictive focus upon the analysis of emotional phenomena that emerge during the collection of primary research data with participants in fieldwork settings denies the complex ways in which emotions enter the research process during secondary analysis. Secondary analysts who work with distressing data can be emotionally affected by such data in many of the same ways as fieldworkers, and are similarly required to engage in an active emotional reflexivity and the invocation of emotion management strategies to maintain emotional equilibrium during the research process.
It has been our argument throughout this article-and one that we hope we have illustrated-that secondary analysis has its own emotional demands. These emotional demands are, however, not restricted to those who engage in research that explores child abuse and other forms of distress children/young people encounter. This much has been well documented elsewhere (see, for example, Dickson-Swift, James, Kippen, & Liamputtong, 2007). There are many accounts of researcher emotion in other areas of sensitive research where either the topic of investigation or dynamics experienced in the field are encountered as emotionally challenging or distressing. Consequently, the issues raised within this article have wider currency that extends quite considerably beyond the context of the projects we have explored.
Such a consideration, therefore, brings to the fore how researcher distress is addressed in the ethical processes that underpin research studies. While we were able to access support through each other and through formal debriefing sessions, many researchers may not benefit from informal or formal support structures that buttress the emotional impact of working with distress. This inevitably positions researchers as vulnerable. Yet, there is little consideration of this in ethical procedures that tend to be focused on ensuring that harm to participants does not occur or that appropriate protocols are developed for participants who experience emotional distress as a consequence of their involvement in our studies. Many academics have raised concerns about this and have highlighted the role that ethics committees could perform in extending ethical considerations around risk and harm to researchers where there is a likelihood that research may result in risk and/or harm (Dickson-Swift et al., 2007). We would concur with such observations, and suggest that such issues are formally integrated into the ethical underpinnings of studies and that mechanisms are put in place to monitor the well-being of researchers, to facilitate supportive spaces within research teams and institutions, and to ensure that where needed, formal support can be provided and/or accessed.
The writing of this account necessarily leaves one with a sense of vulnerability insofar that the construction of this account has involved "personal exposures" of the kind that we are ill at ease with professionally. Reflexive writing of this kind is unfamiliar territory for academics used to the conventions of academic writing that tend to prescribe the rendering of "objective" accounts somewhat divorced from issues of the self that can be comfortably glossed over, diluted, or more often dismissed. The dominant ideology of academia that privileges rationality and objectivity requires us to "extract out" emotion (Author, 2001, p. 135). There is a sense, therefore, that in speaking about these issues, one is revealing something deeply personal about the self for a public audience. Poignantly, it reminds us that the children/ young people who called ChildLine and whose accounts we worked with each went through the process of disclosing the profoundly personal to others. It allows us to recognize to some extent the courage it took them to do so.