Both Insider and Outsider: On Conducting Social Work Research in Mental Health Settings

The mental health clinic poses unique challenges for social work scholarpractitioners. The familiar setting, the nature of mental health data collection, and the researcher’s clinical training and experience all complicate efforts to maintain a reflexive stance in research. Additionally, conducting research in a clinical environment risks replicating a hierarchical medical model in the research relationship. Using a theoretical framework of critical realism, two doctoral-level scholar practitioners analyzed the advantages and challenges of conducting research in a clinical setting. Audit trails and experiences of peer debriefing from their dissertation research served as the basis for this conceptual analysis. The analysis considers the impact of the clinic setting on the power dynamics of the research process, as well as the researchers’ subjective experiences throughout the process of data collection. The authors discuss the risks of Othering and the challenges of straddling insider and outsider identities as scholar-practitioners in clinic settings. To navigate these dual identities of researcher and clinician, the authors recommend maintaining awareness of power dynamics and discourses, debriefing regularly with peers and mentors, introducing reflexive practices into both interviews and writing, and moving beyond binary identities in order to occupy a “space between.”

Social work research is invariably influenced by the contexts within which it takes place (Houston, 2010;Longhofer & Floersch, 2012). Both real and socially constructed characteristics of the research setting can influence the power dynamics that exist between researchers and participants. In particular, in many social work practice settings, including child welfare agencies, schools, and outpatient clinics, social workers represent authority figures (Shulman, 2016) with power to affect the life circumstances of their clients (Kondrat, 1999).
For practitioner-researchers, conducting research in mental health clinics poses unique challenges. Clinical research settings have the potential to activate specific aspects of both interviewers' and interviewees' identities as researchers, clinicians, and/or clients. In clinic-based research, the familiar setting, the collection of mental health data, and the researcher's clinical training and experience complicate the process of conducting research. Wearing "dual hats" of practitioner and researcher in such a setting can both facilitate and inhibit the researcher's capacity to navigate the power dynamics between the researcher and research participants.
This conceptual analysis examined the context of the mental health clinic as the setting for two research projects, both conducted by practitioner-researcher doctoral students. For this analysis, we reviewed the literature on reflexivity and research in mental health settings. We then applied a critical realist framework in order to analyze how the context of the clinic shaped both the process and power dynamics of the research; specifically, we explored how power and authority were negotiated in the mental health clinic in the arena of Othering (Sands & Krumer-Nevo, 2006). We applied three distinct context-specific approaches to reflexivity, each of which were particularly salient in our experience as social work researchers in mental health clinics. The first analyzed the social and discursive context of the mental health clinic; the second identified the impact of the researcher's own subjectivity, particularly for practitioner-researchers working in mental health settings; and the third perspective examined the intersubjective dynamics between researchers, clients, and service providers in the clinic setting. We identified advantages and disadvantages of holding a practitioner-researcher identity while conducting research in mental health settings, and based on the analysis, we recommend reflexive practices that can aid social work researchers working in this context.

Theoretical Paradigm
This analysis was grounded in the philosophical paradigm of critical realism (Houston, 2010;Longhofer & Floersch, 2012). Critical realism is a philosophy of social science that seeks to discover "the deeper structures and relations that are not directly observable but lie behind the surface of social reality" (Longhofer et al., 2013, p. 158). This stance acknowledges that events and behavior that take place in the real world cannot be meaningfully understood without exploring how individuals make sense of them, based on their assumptions and prior experiences (Maxwell, 2013). In this analysis, a critical realist perspective allowed us to recognize both the reality of mental illness, as well as the social discourses that construe how mental illness is perceived, conceptualized, and discussed by providers, clients, and mental health service researchers (Longhofer & Floersch, 2012).
Social work practice typically takes place in complex open systems, where the occurrence of events depends on all the observable and unobservable conditions produced in a given context or environment (Longhofer et al., 2013). Consequently, Houston (2010) argued that social work researchers should explore the enabling or constraining unseen generative mechanisms (such as social roles, beliefs about authority, or discourses) relevant to social work practice, and consider how these mechanisms interact in a particular context, time, and with human agency. In this way, social work researchers can explore the interface between the "micro" and the "macro" in a given context (Houston, 2010). In any institution, unseen mechanisms function to reproduce social positions and relations (Houston, 2010). In this conceptual paper, we analyze how the institutional context of the mental health clinic influenced the process of social work research in that setting. As part of this analysis, we explored the intersubjective and systemic power relations inherent in the mental health clinic, and the impact of these dynamics on research conducted in clinic settings. In keeping with the emancipatory goals of critical realist research (Houston, 2010), we explored ways that the power dynamics of the mental health clinic are both reinforced and challenged by participants in social work research. We also found that holding dual practitionerresearcher identities had us straddling the insider-outsider divide in the clinic setting, which had both constraining and facilitating effects on our efforts to navigate these dynamics with client and provider participants. See Table 1 for a summary of our efforts to analyze the impact of the clinic context on the research process. 3. The clinic context influenced intersubjective dynamics in research relationships with providers and clients.

Reflexivity in Research
For several decades, the insider/outsider debate in the sociological literature has considered the significance of researchers' social locations in terms of their claims to knowledge (Griffith, 1998). Researchers with shared lived experience, tacit knowledge, and a high degree of familiarity with the people they study operate as "insiders"; those without the intimate knowledge, shared lived experience, or membership in the community they study approach their research from an "outsider" perspective (Dwyer & Buckle, 2009;Griffith, 1998). Researchers' social locations also influence the assumptions and paradigms they bring to interpreting data (Berger, 2015;Collins, 1986). The researcher's knowledge is thus necessarily partial and situated within the researcher's personal and cultural history (Collins, 2000).
Each position carries unique implications for a researcher's work; for example, researchers who share a lived experience with participants may have an easier time understanding nuanced reactions and thus enlisting participants (Berger, 2015). Participants may perceive researchers with insider status as more sympathetic and may disclose more accordingly. However, these investigators may also incorrectly assume similarities in experiences, and may risk blurring boundaries or projecting their own beliefs or perceptions onto participants (Berger, 2015). Researchers who are outsiders are less likely to experience this kind of role confusion and may have an easier time approaching their work with "eyes open" (Asselin, 2003, p. 100). Conversely, an outsider's unfamiliarity with a group may lead them to miss subtle nuances in the research and could increase the likelihood of interpreting data through dominant stigmatizing discourses (Berger, 2015;Collins, 2000;Tew et al., 2006). Patricia Hill Collins' concept of the "outsider within" (Collins, 1986, p. S14) challenges the either/or dichotomy of the insider/outsider debate. Collins identifies the unique vantage point of Black women in sociology, who have acquired the "insider" perspective of sociological training (which has typically reflected the experiences of White men) while retaining the standpoint of an oppressed social location at the intersection of racism, sexism, and classism (Collins, 1986(Collins, , 2000. Researchers' positions throughout the research process are often fluid, and researchers may find themselves moving between insider and outsider status at various points in the research process, depending on the ways their identity as researcher is constructed in the interactions of research relationships (Berger, 2015;Dwyer & Buckle, 2009;Griffith, 1998).
Reflexivity, defined as "our human capacity to consider ourselves in relation to our context, and our contexts in relation to ourselves" (Longhofer & Floersch, 2012, p. 513) can help researchers make sense of these complex relationships, and is a key component of a critical realist analysis of social work research (Longhofer & Floersch, 2012). Reflexive practices help investigators to recognize the ways in which they function as an integral, influential part of the research setting, process, and interpretation (Probst & Berenson, 2014), and assist researchers in identifying their biases and reflecting on the emotional impact and power dynamics of the research. By engaging in this process, researchers are better able to illuminate their assumptions and eliminate the presumed neutral, authoritative stance (Krumer-Nevo & Sidi, 2012). Building this awareness allows researchers to learn from these moments of vulnerability during the research process (Goldstein, 2017). Both researchers hold identities as clinical practitioners and as mental health researchers. Throughout this analysis, we used multiple practices of reflexivity to explore how both the context of a mental health clinic and our dual identities affected the process and dynamics of social work research in two different studies.

Reflexive Research in Mental Health Settings Power Relations in the Clinic Context
Conducting reflexive research in mental health services is well-served by an acknowledgement that the concept of mental illness itself is a contested domain (Tew et al., 2006). Numerous approaches to understanding emotional and behavioral challenges exist, ranging from sociological theories of social or political deviance to neurochemical disease (Lacasse & Gomory, 2003). Research in mental health clinics must contend with the dominant biomedical paradigm, or "medical model" of mental illness, which is grounded in a hierarchical approach to expert knowledge and relies on terminology of psychiatric disorder that is frequently stigmatizing (Tew et al., 2006).
The biomedical paradigm situates the mental health researcher within a position of authority and power relative to their participants (Kondrat, 1999). Yet a researcher's work is also reliant upon participants' willingness to share their substantive and experiential knowledge and expertise about the topic of interest (Brinkmann & Kvale, 2005). While a researcher's power is grounded in their choice of research topic, the direction of the interview, and the interview questions, respondents also make conscious choices about their disclosures during the research process, including decisions to withhold information or terminate participation (Anyan, 2013). If this duality of power and control remains unaddressed and unresolved, research may merely replicate the prevailing hierarchy of "expert knowledge" by failing to authentically capture the lived experience of the research participant. Avoiding this risk requires careful attention to complex dynamics of power, control, and dominance between the investigator and participants, as well as an awareness of how these dynamics influence the knowledge-building process (Riley et al., 2003). Conducting reflexive research is facilitated through acknowledging participants' expertise and affirming their right to decide what information to share with the researcher -skills that are often overlooked or underdeveloped within current hierarchical approaches to the researcher-participant relationship.

Personal Reflexivity in the Clinic Context
Researchers are both affecting and affected by the research conducted (Few, 2007;Probst & Berenson, 2014). Research that involves active listening to stories of pain, abuse, and neglect takes an emotional toll on the researcher (Coy, 2006). Practitioner-researchers have training and experience in dealing with emotionally distressing disclosures; yet, if not addressed properly, these can lead researchers to emotionally withdraw from the research out of self-protection (Konrad, 2007).
This process may be further complicated for those whose clinical social work training gives them insider familiarity and lived experience as a provider of mental health treatment (Dobson, 2009;Dwyer & Buckle, 2009). In this case, clinical training and practice experience can function as a form of habitus (Longhofer & Floersch, 2012). Bourdieu's concept of habitus describes the embodied personal dispositions that guide individual action as a result of socialization (Bourdieu & Wacquant, 1992), such as the conventions associated with professional social work training. Having a shared sense of language, values, or culture carries with it an existing set of assumptions that can impede the researcher's ability to process stories that contradict expected narratives, maintain a critical perspective, or encourage participants to speak in ways that challenge an accepted discourse (Asselin, 2003;Dobson, 2009;Van Heugten, 2004). The structure of social relations in a mental health clinic can unconsciously activate the institutional habitus of practitioner-researchers and influence interpretations throughout the research process (Longhofer et al., 2013).
As social work clinicians and researchers, the authors also faced the challenge of maintaining dual identities of practitioner and scholar. Instances where the priorities or normative behaviors associated with each identity diverge can create role conflict for the researcher, as she must decide which part of her professional self to draw upon in that moment (Colbourne & Sque, 2004). Personal reflexivity increases one's awareness of this conflict and requires the researcher to reflect carefully on how her own values and assumptions carry forward into the research process.

Intersubjective Dynamics in the Clinic Context
Reflexive practices in clinic-based research require attention to the interpersonal dynamics that occur when the dual subjectivities of researcher and participants meet (Goldstein, 2017;Jensen, 2012). When researchers invite participants to speak from a particular subject position, they in effect invite them to occupy a specific identity, which may be stigmatized (Jensen, 2012). These invitations carry the risk of Othering, defined as perceiving people who have been marginalized as foreign, one-dimensional, or a fundamentally different them, in contrast to us who are not suffering, devalued, or marginalized in this way. In research, Othering can reflect an effort to seek distance from distressing narratives (Konrad, 2007;Krumer-Nevo, 2002), or a researcher's tendency to focus on a single dimension of a person's identity (Sands & Krumer-Nevo, 2006).
Research with users of mental health services carries unique risks of Othering. Dominant narratives around health and disability create a binary between those who are healthy and those who are sick (Butler et al., 2007;Read, 2016). The medical paradigm has created normative expectations for recipients of care to behave as cooperative and compliant patients (McCreaddie & Wiggins, 2009), while privileging the perspectives of social workers' expert opinions to define the experiences of others (Butler et al., 2007;Tew, 2008). Practitioner-researchers who have been socialized into these models of understanding, such as the medical model, may risk unintentionally subordinating or pathologizing users of mental health services throughout the process of data collection and analysis (Tew et al., 2006). Researchers may also find that clients have learned to speak in institutional discourses and may share only the stories they believe interviewers want to hear (Butler et al., 2007). For these reasons, research in mental health settings ideally involves both the standpoints of those with insider knowledge along with research frameworks that allow for rigorous analysis (Tew et al., 2006).
Power relations in research are not static; both the researcher and research participants may feel powerful and powerless at different points (Goldstein, 2017;Jensen, 2012;Krumer-Nevo, 2002). One sign that Othering has been taking place during research is the presence of a shock. Such shocks, which can be intentional or unintentional, are unexpected statements or actions by a research participant that generates a cognitive and emotional reaction of surprise in the interviewer (Sands & Krumer-Nevo, 2006). Sands and Krumer-Nevo (2006) offer several examples of shocks in research interviews, including unexpected emotional reactions (such as participants crying, flirting, or becoming hostile) and research participants who switch roles and ask the interviewer questions. The interviewer's surprise indicates that her implicit expectations have been disrupted and reveals a clash between the narratives of individuals occupying different positionalities (Sands & Krumer-Nevo, 2006). By puzzling or disarming interviewers, interviewees who deliver shocks gain a chance to renegotiate their stories and resist being depicted in stereotypical ways (Sands & Krumer-Nevo, 2006).

Case Summaries of Research Projects
Sapiro has previous clinical experience working as a social worker with adolescents in community-based outpatient mental health services. Her dissertation focused on how older adolescents living with mood and anxiety disorders experience their relationships with friends, family, and helping professionals. The dissertation project was part of a larger pilot study of an intervention designed for transition-age youth receiving mental health services. The study included multiple forms of data collection, including quantitative surveys and qualitative interviews with both clients and service providers at an outpatient mental health clinic. Data analysis consisted of thematic analysis of interview transcripts. Sapiro used several methods to ensure the rigor of her qualitative data analysis, including keeping an audit trail of memos throughout the process of data collection and analyses, consulting with both peers and mentors during the process of analysis, and triangulating analysis by having a second coder analyze code transcripts and review the findings. Throughout the analysis, consulting with the second coder helped Sapiro become aware of blind spots in her interpretation of data because of her own clinical training and experience. During coding, Sapiro also coded for references made by participants to the research relationship; these observations serve as the basis for the current analysis.
Matthews previously worked as a clinical practitioner in a community health center. Her dissertation examined how electronic health records were being used in therapy and tested the relationship between in-session computer use and the client-provider relationship. This mixed-methods project used video-recorded therapy sessions and postvisit surveys completed by therapists and clients in order to explore the relationship between the frequency and manner of in-session computer use and the therapeutic alliance. Matthews used several methods to ensure the reliability and rigor of her coding and analysis. This included co-coding of therapy sessions by a doctoral level research assistant, paired with routine peer debriefing. The audit trail detailing these processes provided the basis for the present study. Both studies were approved by the authors' institutional review boards.

The Context of the Mental Health Clinic
As an institutional social setting (Houston, 2010), the functioning of the mental health clinic depends on reliable patterns and practices among administrators, service providers, and clients, much of which were familiar to the authors as former clinicians. This social structure had both enabling and constraining effects (Houston, 2010) on our efforts to conduct research in the clinic setting.

Advantages of the Clinic Setting
As part of participation in both research studies, clients evaluated their experiences receiving mental health services at the clinic. Those who felt positively towards the clinic often transferred these warm feelings to the researchers, making the clinic context an enabling factor in some circumstances. In Sapiro's experience, several participants commented that they felt comfortable participating in the study because of the positive associations they had with the clinic and its staff. One participant reported, "I love talking to people here, because every person I've met here is a pleasure." Similarly, Matthews found that when providers expressed comfort with research participation, clients were more accepting of the study as well. Further, clinicians often brokered Matthews's initial introduction to clients. In these conversations, the established trust and rapport in the clientprovider relationship were extended to Matthews, allowing her to seamlessly initiate a discussion of research participation.
For Sapiro, conducting interviews in therapeutic offices also benefited from this association, as participants were used to talking about themselves in this setting, especially about topics they may not easily discuss with others. One youth participant made this explicit when she reported that her experience being interviewed felt similar to her experience in therapy: "I guess it was nice 'cause I never like, I mean I guess I've spoke about it with [my therapist]. I never really like talk about it with anyone else." When asked about her experience participating in the interview, another youth participant reported, "It's the same thing as talking to my therapist."

Challenges of the Clinic Setting: The Discourse of Mental Illness
A critical realist analysis identifies potentially oppressive dynamics of power in a given setting (Houston, 2010). In addition to the advantages of conducting research in the clinic, this setting meant that the power dynamics of the therapeutic relationship were highlighted in research interviews. In many ways, qualitative interviews mirror therapeutic interviews: both involve empathic listening, bearing witness to disclosures, and allowing space for participants to develop new understandings of themselves (Coy, 2006;Goldstein, 2017). The asymmetrical power dynamics of research interviews also mimic those of therapeutic interviews, in that the researcher is the identified "expert" or authority figure who sets the agenda, asks the questions, determines what constitutes data, and retains privilege in interpreting, defining, and writing up the interview (Brinkmann & Kvale, 2005;Krumer-Nevo, 2002). For example, when Sapiro asked a youth participant at the conclusion of the interview if there were other questions she should have asked but did not, the participant replied, "Oh, I have no idea 'cause I don't know what you're supposed to ask," implying that the researcher alone held responsibility for directing the research interview. Interview dynamics manifest differently depending on the research questions and study design. In participatory action research, for example, researcher authority is more explicitly addressed compared to traditional research interviews (Padgett, 2016). Regardless of how dynamics of power and authority are ultimately negotiated between the researcher and participant, the presence of this asymmetry persists. In this way, the research relationship echoes but does not disrupt the medical model of expertise and authority in the mental health clinic (Tew, 2008).
In each of our experiences, we identified ways that participants engaged with or resisted the dominant biomedical discourse of mental illness. Our research in mental health clinics seemed to bring to the forefront participants' feelings about their providers as significant medical authority figures. Within the health care context in particular, the construct of a good patient places expectations on individuals to demonstrate a sense of gratefulness towards providers, and a willingness to adhere to prescribed treatment (McCreaddie & Wiggins, 2009). Much like social desirability can shape participant responses (DeVellis, 2003), expectations of good patients are salient to clients, who in turn may suppress behaviors or emotions in order to embody the attitudes and characteristics desired by their providers (Campbell et al., 2015). In this way, conducting research within the clinic itself complicated and in some ways constrained the process of uncovering clients' authentic feelings about providers and their care, as the context seemed to activate their desire to perform the good patient role.
For Matthews, this was apparent in several interviews, where clients offered unprompted reassurances about their loyalty to their provider while completing survey questions relating to their relationship quality and satisfaction. Being situated within the therapeutic setting, clients appeared reluctant to betray their providers by exposing ruptures in their relationship or desires for a different type of care. In three separate instances, clients penciled unprompted positive statements into the margins of the paper survey after endorsing an item that could be considered unfavorable to a provider's behavior during session. For example, one client wrote "these are not important to me, I trust [provider]" after endorsing two items that suggested her provider was not transparent or collaborative in her note-taking. This need to justify or offset their negative rating with praise seemed to highlight the client's unwillingness to appear critical of their provider. In response, during the rapport-building stage, Matthews was careful to review confidentiality of the data, to normalize the experience of having both positive and negative reactions to therapy, and to distinguish her role from an administrator or evaluator of the clinician's work.
Sapiro's dissertation was part of a larger study described to participants as a way to improve services for young people receiving mental health treatment. Perhaps inadvertently, this framing of the research emphasized the power that social workers possess to label, diagnose, and interpret behavior (Butler et al., 2007). Despite researchers' efforts, by conducting research in a mental health clinic, the roles and power of the clinician and the researcher were often blurred in the minds of participants, adding to the power differential in the interview.
In Sapiro's experience, the biomedical model of mental illness was reinforced through some of the language used by the larger study, which surveyed youth participants about their symptoms and asked about previous hospitalizations. Some participants referenced the stigmatizing language of mental illness, while others found ways to reclaim it as a point of pride. Following consultations with her second coder, Sapiro identified the habitus of clinical social work in her tendency to use clinical terms (such as defenses, coping mechanism, or social anxiety) to describe research participants. In a memo from the data analysis, Sapiro wrote, [Second coder] notices that just because someone self-identifies as very shy, does not automatically mean that they have social anxiety -it may be regular adolescent shyness. My bias may be towards diagnosis, because of my background and because the interviews are taking place in a clinical setting.
This framework undoubtedly contributed to Sapiro seeing participants through the dominant or "insider" lens of mental illness and mental health services (Read, 2016), in spite of all of her efforts to the contrary.
In this way, conducting research in a mental health clinic poses risks for Othering. One research participant referenced this idea explicitly. In describing her general unwillingness to participate in therapy, she described how she did not want to be treated in any clinical or other encounter: It's like they look at all the clients or patients, whatever, the same, you know. She's someone who went through this or this happened, you know. I have more to my life than being this little like a lab rat that you're testing.
Throughout the research process, we found that participants actively engaged with the norms, logic, and dominant biomedical model of the mental health clinic.

Researcher Experience in the Clinic Setting
Our second analysis of the impact of the clinic context involved reflection on our subjective reactions and interactions during the research process, focusing specifically on how clinical training shaped our thoughts and behaviors as mental health service researchers (Longhofer & Floersch, 2012). For both authors, our past experiences as clinical social workers contributed to role conflicts and challenges with managing researcher relationships during data collection and analysis.

Role Conflicts
As both of us brought our past experiences of clinical social work to our research, an initial task was to negotiate a new role, as researcher and doctoral student, within a familiar setting. Being trained as social workers required effort to ensure that our research curiosity was not limited by a clinical lens. Clinical training can serve as a kind of habitus, or disposition, that can affect a researcher's experiences with participants and with interpretations of their narratives (Longhofer & Floersch, 2012). We each encountered instances where our dual identities came into conflict, leading to dilemmas during data collection. Matthews video-recorded therapy sessions and coded them for particular behaviors related to providers' computer use. Viewing these recordings provided opportunities to witness both exceptional and problematic therapeutic work. In one instance, she observed a session in which a client disclosed suicidal ideation and the clinician failed to adequately address the risk. In spite of a strong sense of professional obligation to protect the safety and well-being of the client, Matthews recognized that her role as a researcher, rather than a clinician, placed new limitations on her opportunities to intervene. Matthews used peer debriefing to process and consider this conflict. Through gentle probing, despite her lingering discomfort with this scenario, Matthews recognized that, particularly as the client in question was actively engaged with both a therapist and a psychiatrist, asserting any type of clinical expertise into this scenario would have overstepped her boundaries as an outsider and academic researcher.
Sapiro experienced a different version of this conflict when interviewing clients in a mental health clinic. Interviewing participants in acute emotional distress engendered feelings of sadness, anxiety, and exhaustion, even when she was able to help connect these research participants to mental health services. During the interviews, Sapiro was aware of the potential need to stop the research and contact a supervisor in the event of a participant who appeared to be at risk of harm to self or others. Although the need never arose, staying alert to this possibility meant an inability to fully disengage from clinical assessment during the research interview, and may have contributed to the use of clinical terminology in interpreting data as described above. Coy (2006) and Colbourne and Sque (2004) have described similar conflicts, noting that, when conducting research with people in distress, researchers with practitioner experience may struggle to divorce themselves from the role of a clinical professional. Though the skills learned as a clinician may be helpful for gaining insight into respondent behaviors and establishing rapport during interviews, they can also obfuscate the goal of the research project and the role of the researcher. Careful reflection must be used to determine when the employment of clinical skills is a useful tool, as when fostering rapport, or broaching sensitive topics, and when role conflict is arising due to an unwillingness to "let go" of one's clinical role. For both authors, "letting go" was uncomfortable, though peer debriefing helped us reflect on the challenges of being a researcher, rather than a clinician. In these situations, we were reminded that in spite of our clinical training, we were still "outsiders" in these clinics.

Managing Research Relationships
Matthews conducted her research within a mental health clinic that was a former place of employment, which raised specific challenges in establishing and maintaining research relationships with the agency. Matthews's lingering feelings of affiliation with the clinic created loyalty tugs (Brannick & Coghlan, 2007), where she felt compelled to prioritize the desires and expectations of the organization over those of the study, even when they clashed with her priorities as a researcher. For example, during the project, she needed to both ask for logistical assistance from already overburdened staff members, and tactfully decline administrators' requests to view study data for their own purposes. In these instances, she struggled to balance the desire to maintain a positive working relationship with the organization, while upholding the standards of ethics and rigor in her research.
Fortunately, during this time Matthews was simultaneously involved in another study involving primary data collection with a different group of mental health clinics. The two studies served as a useful exercise to compare and contrast approaches towards establishing a working relationship with familiar versus unfamiliar organizations and staff. This reflection helped Matthews identify instances where her discomfort was triggered by an inclination to operate according to the chain of command that defined her former role as a clinician (insider), rather than her current role as an independent researcher (outsider).

Intersubjective Dynamics in the Clinic Setting
Power is omnipresent in social interactions (Houston, 2010). The final application of critical realism to these studies involved a focus on how the stratified social world of the mental health clinic, with its particular power dynamics, manifested in the research relationships we developed with both service providers and clients. These examples illustrate how the research relationships were affected by the researchers' identities as both insiders and outsiders.

Relationships with Providers
In Matthews's experience, the existing trust and rapport clinicians had with clients meant that providers functioned as gatekeepers to client access. Providers served variably as a bridge or buffer between the researcher and the clients, either facilitating or precluding access to potential respondents depending on their own assessments of clients' suitability for research participation. In several instances, providers buffered access by preempting attempts to recruit client participants if they assessed the individual to be acutely distressed or in a delicate stage of treatment engagement. Coy (2006) identified a similar dynamic, whereby researchers assumed the task of evaluating a potential respondent's fitness to participate. Though a well-intentioned effort to protect vulnerable clients from undue harm, by determining their ability to participate on their behalf, these decisions ultimately minimized client autonomy and reinforced providers' power over participants. This placed Matthews at the intersection of two principles of ethical research: autonomy, in which the respondent's decision authority is respected, and beneficence, which demands research participants are not harmed by the research process.
For Sapiro, the power dynamics in interviews with providers offered several examples of shocks that suggested these interviews were taking place in the arena of Othering (Sands & Krumer-Nevo, 2006). As a clinical social worker, Sapiro saw herself as an insider: she identified with the clinicians and expected these interviews to proceed without friction. However, a number of staff members seemed to feel nervous, uncomfortable, or somewhat reluctant to be interviewed, and expressed this in a variety of ways: by joking, by presenting as standoffish, or by acknowledging a wish to have prepared for the interview in advance. For example, one provider reflected on her experience responding to Sapiro's questions: It makes me have to think because, you know, I didn't know what the interview was gonna be about, right…. So I had no way to mentally prepare, which is something I always like to do, right, is mentally prepare for whatever I know is coming. I couldn't mentally prepare, so I definitely had to think more as I was talking to you, which is fine, you know.
These reactions served as a corrective shock to Sapiro's assumption that her status as a presumed insider would facilitate the interview process and reduce interview-related anxiety. Upon further reflection, Sapiro recognized that providers' reactions could have reflected general attitudes about research participation. In addition, clinicians who are accustomed to assuming a position of relative authority in their therapy offices may find being the interviewee, rather than the interviewer, to be uncomfortable or unexpected.
Matthews also had an experience of assuming to be an insider, only to find that she was perceived otherwise. Data collection relied heavily on earning the trust of providers working within the agency. Having the shared experience of being a clinical social worker who had worked within the organization, Matthews began her recruiting of provider participants with the expectation that her shared group membership would award her a degree of cultural capital (Bourdieu, 1986). She anticipated gaining trustworthiness and buy-in from providers through her insider status. Despite the purposeful disclosure of her shared background, Matthews was surprised to discover that she was viewed first and foremost as an academic researcher, rather than a former clinical employee or colleague. Ultimately, the attempt, and failure, to connect with providers using this assumption of "sameness" squandered precious opportunities to gain their buy-in for this study. Being more aware of how providers were perceiving her would have enabled Matthews to forge a more authentic collaborative relationship with providers.

Relationships With Clients
The dynamics of interviews with clients also reflected meetings in the arena of Othering. Research participants knew that they were participating in these studies based on their status as clients of the mental health clinic (Jensen, 2012). With this knowledge, they both engaged with and resisted the stereotypes of mental illness and client-provider relationships in various ways. This played out in the frequent conformation to the role as the good patient observed during Matthews's interviews with clients, where clients offered unsolicited assurance of their appreciation of and connection to their providers. There were also instances of clients subtly bucking these expected norms and behaviors of compliant patients. These behaviors were most apparent during the video-recorded sessions, when several clients engaged in a dramatized process of primping for the camera, in an apparent attempt to take ownership and control of the recording process. In the middle of one session, one respondent took a moment of silence between he and his therapist as an opportunity to make faces at the camera, demonstrating that his awareness of his research participation had not waned, even as the camera rolled silently in the corner of the office. Sapiro interviewed one young woman who was very clear about her intention to challenge stereotypes by representing herself as a young person who lives with mental illness and has a very supportive group of friends: People with mental illnesses, alright, they're portrayed in like a certain way, and I feel that now when I'm telling you about how I have friends and everything like that and how like I live fairly happily for most, for a good part of my life. It's very important to me to talk to somebody else, to let them know that there's that stigma, like what you're talking about that like mentally ill people can't hold friendships, they can't have like interpersonal relationships, things like that. And I think that that's something that people shouldn't look at and they shouldn't think that, you know, someone is like sick so it prohibits them from doing x, y, z.
Before, during, and after interviews, youth participants made a number of comments that referenced the power dynamics of the interviews, reflecting their awareness that as the researcher, Sapiro retained both the audio record and the interpretive privilege over the stories that they had shared. A number of participants commented that it was their first time being audio-recorded for an interview. At the beginning of her interview, after reviewing the purpose of the interview and giving consent to audio-record, one participant commented, in jest, "Wow, I feel really weird. I feel like I'm being interrogated." At the end of the interview, Sapiro asked her to reflect on her experience, and her response reveals both the similarities and differences between participating in research and being a patient at the clinic:

I've never done like part of a research kind of study thing, so I've always wanted to. I don't know why. I've always wanted to be like a guinea pig, you know. I mean, I don't really feel [like one], probably because like, I don't know, you know, like I know I'm a guinea pig, but I don't feel like a guinea pig right now. So it's okay. Like I'm not in a lab getting like blood taken or something.
Another young woman demonstrated acute sensitivity to the potential for adults to abuse the power they have over young people. She described her encounters with certain teachers at her school in this way: There are other teachers who are just like, I'm sorry, I can't do that for you because I don't know, like they don't have to or whatever it is… they're like I'm sorry, I can't do that for you or because you know, they don't feel like it and it's just, like, taking extra time out of their day. And they also, like don't, like, give the student any power; it's just like whether I feel like doing it or not for you.
In her two interviews with this participant, Sapiro was conscious to be as transparent as possible in her role as an interviewer and a collector of survey data. At the end of the second interview, the youth said to Sapiro, "You're really good at your job." Sapiro was grateful for the compliment as a novice researcher and thanked her. The participant continued, "You're not like one of those gynecologists who doesn't tell you what they're going to do and then just sticks you with things." Once again, an interviewee delivered a shock that prevented Sapiro from seeing herself as an innocuous or benevolent researcher who stood apart from the medical bureaucracy of the clinic. It is difficult to imagine a more invasive metaphor for research participation, even for a young person who actively consents to participate.

Recommendations
Using a critical realist analysis, we highlighted how the context of the mental health clinic impacted multiple components of the research process. As early-career researchers, this reflection sensitized us to practices that can enhance our awareness of the role of context in future research projects. Drawing from these experiences, we offer several recommendations for practitioner-researchers looking to maintain a reflexive stance while conducting research in familiar clinical environments. These recommendations are summarized in Table 2.

Recognize Power Dynamics
Critical realism recognizes the omnipresence of power dynamics in social interactions, especially in stratified social settings such as agencies (Houston, 2010). The hierarchies of mental health clinics and the power dynamics of the client-provider relationship manifested as unseen mechanisms in each of our studies in several ways. As researchers, it is incumbent upon us to recognize these dynamics and their impact on research participants. In any social work setting, including schools, child welfare agencies, or other systems of care, people's prior experiences with social workers (as well as researchers and other authority figures) will impact how potential research participants experience their interactions with researchers in these settings. These power dynamics are not simple or straightforward, as our experiences demonstrate; rather, they are constantly shifting depending on the particular individuals and agendas involved. We recommend that researchers partner with service users in the development of research projects, in order to ensure that their valuable standpoint is incorporated into each level of the research project (Tew et al., 2006).

Use Debriefing
For both authors, the process of debriefing with both colleagues and mentors was an important strategy for managing the unavoidable challenge of conducting research with people in pain. Just as regular supervision helps clinicians reflect on their work and identify emotions and dynamics within the therapeutic relationship (D'Cruz et al., 2005), regular debriefing with peers and mentors serves a similar function throughout the research process. Peer debriefing is an established method for maintaining rigor and credibility in qualitative research (Shenton, 2004). For us, peer debriefing also helped us interrogate our assumptions and reactions, process the dilemmas that arose during data collection and analysis, and differentiate our dual and sometimes competing identities as scholars and practitioners. In Sapiro's experience, having a peer coder who was not a clinical social worker helped correct for some of her biases in interpretation owing to her clinician habitus.
Both researchers were also fortunate to have academic mentors with clinical backgrounds. These seasoned scholars, who had both worked as clinicians and conducted research in clinical settings, were clearly able to understand and respond to instances of conflict as described in the preceding sections. For example, when Matthews was surprised that providers perceived her as an outsider, her mentor related by reflecting on specific instances where her own experience as a case manager influenced her research relationships with case management programs. Hearing these experiences, in addition to witnessing how this mentor navigated current research relationships, offered a model for how to resolve tensions between the practitioner and researcher role. Similarly, Sapiro found it invaluable to talk through with her mentors the ethical dilemmas inherent in conducting research with participants in various levels of emotional distress, as well as how to manage situations in which respondent well-being trumps the collection of research data. We recommend that practitioner-researchers in particular deliberately make room for reflection during debriefing sessions, and seek out mentors, colleagues, and peers to help them identify the conscious and unconscious dynamics present in their research.

Be Alert to Discourses
In each of our studies, we identified diverse ways that people conceptualized and made sense of their psychological distress. Suffering that was labeled as mental illness was variously understood by research participants as a response to trauma, a quirk of the brain, or a source of vulnerability and shame. A critical realist approach to mental illness recognizes that the suffering of individuals living with mental health difficulties is real, and at the same time, people can construe this suffering in various ways, each of which can have different effects on individual life experiences. Our experiences as researchers in mental health clinics reminded us of the need to be aware of the discourses that we and others rely on to make meaning of our own and others' experiences. During our research projects, we witnessed clients both resisting and embracing conventional notions of 'mental illness' and 'mental health patients,' each of which uniquely influenced how participants related to and responded to the research process. Drawing from these experiences, we recommend that practitioner-researchers proactively explore how participants understand their suffering, and consider how various constructions of mental illness affect both researchers and participants. We also recommend that social work researchers investigate the models of practice that social workers in various settings use to make sense of their work, and consider if and how these models could be oppressive and/or emancipatory for service users. These efforts can promote more respectful and authentic relationships with clients, increasing the potential for more meaningful and complete study participation from client respondents.

Recognize Insider/Outsider Status as "Both/And"
Both authors experienced instances of wrongly assuming insider status. Particularly for Matthews, many challenges stemmed from her assumption that she was carrying out her study as both a professional and organizational insider. At various stages of the project, both authors found this expectation to be alternately validated and violated, creating challenges that required careful and explicit reflection. Our experiences confirm those of other scholars who recognize that researchers are constantly moving between insider and outsider status in research relationships (Griffith, 1998). Especially when investigating dimensions of social difference and marginalization, researchers are often both insiders and outsiders (Collins, 2000;Griffith, 1998). To this end, Dwyer and Buckle (2009) suggest that conceptualizing oneself as either "out" or "in" imposes reductive and arbitrary expectations of complete sameness or difference. Instead, these authors offer an alternative framework of occupying the "space between," to fully appreciate the complexity of human experience and recognize more authentic instances of similarity and difference that cut across group membership or affiliation. In keeping with our emancipatory goals of conducting research with members of marginalized communities, challenging the automatic categorization of participants is one way of interrogating assumptions associated with dominant patterns of thought (Collins, 1986;Griffith, 1998). Similarly, Collins (2000) reminds researchers that our knowledge is inevitably partial and situated within our social locations and personal histories. Collaboration with people whose standpoints differ from our own increases the potential for new insights and angles of vision (Collins, 2000).

Interview and Write Reflexively
Reflexivity also requires attention to the representation of the Other in the research, since writing about members of marginalized groups has the potential to either challenge or reinforce stereotypes (Collins, 2000;Krumer-Nevo & Sidi, 2012). In all of the research interviews, we approached participants with the assumption that they were the experts on their own life experiences and worked to listen intently and respectfully while staying aware of what we did not know and wanted to know. When research participants are members of marginalized groups, there can be a tendency for people to present themselves in ways that conform to stereotypes (Sands & Krumer-Nevo, 2006), such as the idealized compliance of a "good patient" (McCreaddie & Wiggins, 2009). Researchers and participants alike may unconsciously expect interviews to follow "master narratives" that reflect the beliefs of dominant groups about groups that are stigmatized or subjugated (Collins, 2000;Sands & Krumer-Nevo, 2006). These narratives often preclude the representation of members of marginalized groups as multifaceted and complex individuals (Sands & Krumer-Nevo, 2006). In our experience, participants engaged with these stereotypical controlling images (Collins, 2000) in various ways. The challenge is to listen in a more complex way that rejects the stereotype, affirms the interviewee, and helps her present herself in a fuller way (Krumer-Nevo, 2002). One way to get underneath this mainstream discourse is to leave room for persistent follow-up questions, which can reveal complexity behind a seemingly simpler narrative (Gilligan, 2015;Krumer-Nevo, 2002). As one example, Sapiro asked each research participant to reflect on their experiences being interviewed, as a way to invite participants to actively shape their own representation in the interview, as well as comment on the dialogue between interviewer and participant (Collins, 2000;Krumer-Nevo & Sidi, 2012).
While the practice of reflexivity was particularly salient throughout the data collection and interpretation phases, the final challenge involved efforts to avoid Othering in the writing stage of this project. As practitioner-scholars who toggled between insider and outsider statuses in the clinic settings, we both were conscious of how our identities influenced the collection and interpretation of data, as well as the ways that they were constructed in interactions with research participants (Griffith, 1998). Given the restricted and frequently pathologizing nature of a clinical lens, Sapiro endeavored to represent participants in their full complexity and inconsistency, in order to ensure that the descriptions did not conform to stereotypical depictions of a problematic, blameworthy, or pitiable person. In the same way, Sapiro attempted to represent the many strengths of research participants without romanticizing them or minimizing the challenges they faced. This effort sought to counter some of the common tendencies by which members of marginalized groups are often Othered in social work research (Krumer-Nevo & Sidi, 2012). While our social locations as practitioner-researchers contributed particular standpoints to this research (Collins, 2000), we were also reminded of the need to collaborate with service users whose perspective provides a necessary corrective to our own blind spots.

Conclusion
Throughout this paper, we have discussed several challenges encountered in our experiences conducting clinic-based research while holding dual identities of both researcher and practitioner. We have also identified ways in which deliberate engagement in reflexivity helped uncover ways that the context influenced our research, through unconscious expectations, assumptions, and pre-existing patterns of social relations. Reflexive practices also helped us process some of the more salient research dilemmas and emotional obstacles we encountered while conducting research with vulnerable populations. Collectively, this process has provided opportunities for growth, demonstrating the limitations of binary conceptualization of insider/outsider status. Peer debriefing and reflexive writing practices helped us identify and reflect on these challenges and will continue to serve as essential tools for rigorous research going forward.