Volume 19, No. 3, Art. 12 – September 2018
Confidentiality in Qualitative Research Involving Vulnerable Participants: Researchers' Perspectives
Abstract: In this article, I analyze the ways researchers manage the issue of confidentiality in studies with vulnerable research participants in the Polish context, which is characterized by a relatively low degree of ethics and data protection legislation in the social sciences. I am primarily interested in what approach researchers take on confidentiality and how this informs their research practice. I also consider how they attempt to protect confidentiality in their research. The analysis is based on in-depth interviews and e-mail correspondence with 42 Polish qualitative researchers who deal with vulnerable groups and individuals. In addition, I discuss what my interviewees themselves consider to be confidential in the context of my research with them.
Key words: confidentiality; anonymization; vulnerable research participants; ethics; qualitative research; Poland
Table of Contents
2. Literature Review
4. Confidentiality Issues in this Study
5. Confidentiality in Research Practice
5.2 Protection of information
6. Researchers' Comments on Confidentiality in my Study
Confidentiality is frequently discussed in the literature, especially in the context of research involving vulnerable groups. Although there are many definitions of vulnerability, in the context of research with humans, this term often indicates groups or individuals susceptible to harm or risk (ALDRIDGE, 2014). Some codes of ethics (e.g., AMERICAN ANTHROPOLOGICAL ASSOCIATION, 2012; BRITISH SOCIOLOGICAL ASSOCIATION, 2017) and many researchers point to the need for special protection against harm and exploitation of such groups (HURST, 2008). One way to protect them (as well as the other participants) is maintaining confidentiality. This seems particularly important in situations where participants hide their membership in a stigmatized group or when they are critical of persons or institutions on which they depend. However, revealing the identity of study participants based on their request may empower them in certain circumstances because their voice can finally be heard. Only very few studies of confidentiality draw on a systematic examination of researchers' practices in a specific context. Relying on in-depth interviews with Polish researchers, I seek to address the following questions: "How do researchers understand confidentiality and what shapes that understanding?" and "What confidentiality strategies do they use at various stages of the research process?" I also discuss issues that my interviewees find confidential in the context of my study. I understand confidentiality as 1. not disclosing what the participants said or did during research unless they consent to do so and only in ways they agreed, and 2. concealing the identity of the participants. 
In this article I present a case study that expands our knowledge about confidentiality in a context characterized by a rather low degree of institutional and legal regulation on the matter. In Poland, unlike in English-speaking countries (e.g., the United Kingdom, Australia or the United States), there are only a few ethics review boards in social science, which usually are not mandatory. Nevertheless, there are some guidelines for social researchers provided by the code of ethics (e.g., "Code of Ethics" of the POLISH SOCIOLOGICAL ASSOCIATION, 2012) and by the data protection legislation. Thus, the situation in Poland seems similar to that in Germany (VON UNGER, DILGER & SCHÖNHUTH, 2016). However, in Poland, legal norms regarding the protection of personal data in the context of qualitative research are rather general and not very restrictive.1) Furthermore, this article may also contribute to increasing the ethical awareness of researchers. I assume that research ethics cannot be reduced to universal principles. Ethical decisions are always contextual and socially situated. Therefore, awareness of how the researchers manage ethical issues can help us make our own informed and reflexive ethical choices. 
I begin with a literature review on confidentiality to set the context for the rest of my article (Section 2). This will be followed by an account of the methodology of my study with Polish qualitative researchers (Section 3). After that, I will describe how I managed the issue of confidentiality (Section 4). Next, I will present the results of my study with division into anonymization (Section 5.1) and protection of information (Section 5.2). Afterwards, I discuss researchers' comments on confidentiality in my study (Section 6). This leads on to discussion and conclusion (Section 7). 
2. Literature Review
Three types of work on confidentiality in qualitative research are discussed in the literature. Type 1 pertains to normative codes and guidelines. These studies focus on models of good practice and include ethical guidelines for sociologists and anthropologists. Second, there are considerations of researchers concerning securing confidentiality in the research. For example, many authors point to the difficulties linked to ensuring confidentiality, which—as some emphasize—are not always fully understood by ethics committees. These difficulties may result from the applied research methods, as in visual studies (e.g., MILLER, 2015), the aim of the research (e.g., emancipatory research which involves participants taking part in a public event) (e.g., BARRY, 2017), or the nature of research sites (e.g., KAY, CREE, TISDALL & WALLACE, 2003). In addition, some research participants do not want to be anonymous (MILLER, 2015). The issue of protection of confidentiality in the context of data archiving and their re-use was also discussed in detail. The main question is how to anonymize research material to protect participants on the one hand, and not to undermine data quality and integrity on the other (e.g., CORTI, DAY & BACKHOUSE 2000; ORSATTI, 2000; SLAVNIC, 2013). This is a problem, among others, for qualitative researchers conducting research involving vulnerable participants (e.g., BAEZ, 2002; KAISER, 2009). Thus, although guaranteeing confidentiality is generally an ethical standard in qualitative research, it still raises controversies, especially among ethnographers and researchers working in the emancipatory and participatory research paradigm. 
The ethnographer's experience shows that it is very difficult to hide the participation of some people in research, especially in the case of small communities whose members know each other. This applies both to the stage of research, because researchers attract attention to themselves and to their informants (VAN DEN HOONAARD, 2003), and to dissemination of the results. The use of pseudonyms (for people or places) does not guarantee anonymity, because the research results are full of ethnographic detail that could be used to identify research participants. Therefore, some ethnographers wonder whether it would be better to give up anonymity (e.g., SCHEPER-HUGHES, 2000; STEIN, 2010). In turn, researchers from the emancipatory and participatory research, who share the idea of cooperation with the research participants, emphasize that confidentiality should not be imposed but negotiated with participants. Some believe that in certain circumstances, revealing the identity of the participants may have a strengthening and empowering effect, especially in the case of marginalized or vulnerable people whose voice has not yet been heard (ALDRIDGE, 2015). 
The third type of work on confidentiality in qualitative research is still rare and concerns the investigation of researchers' practices (e.g., WILES, CROW, HEATH & CHARLES, 2008). Based on interviews with British researchers working with vulnerable groups or simply interested in research ethics, the authors investigate the ways researchers manage confidentiality and the circumstances in which accidental and intentional breaches of confidentiality occur. Although the authors focus mainly on the issue of breaking confidentiality, they also analyze the ways in which researchers assure confidentiality. Their results show that researchers protect the identities of respondents primarily through anonymization. Depending on the approach, the researchers either thought they were solely responsible for ascertaining the anonymity of the participants or that the participants should be involved in the process of anonymization (mainly researchers working in a participatory paradigm and conducting research with children, young people or in palliative care). This influenced how and what researchers anonymized in research publications. In general, the main ways to anonymize were: 1. change key characteristics if it did not affect the integrity of the data, and 2. not publishing data if the identity of the participants could not be hidden. 
In this article, I explore the third type of approach to confidentiality. The article is novel in two ways. Firstly, I have gone beyond the existing literature by focusing on the Polish social and cultural context. Poland is a special case, owing to the fact that the institutionalization of control over research ethics is relatively weak in sociology and cultural anthropology (so far, a code of ethics in cultural anthropology has not even been developed). There are the "Code of Ethics" developed by the POLISH SOCIOLOGICAL ASSOCIATION (2012) and "The Code of Ethics for Research Workers" prepared by the SCIENCE ETHICS COMMITTEE OF THE POLISH ACADEMY OF SCIENCE (2017) for all researchers from various scientific fields. It should be noted however that the latter document only in a small part refers to the issues relating to conducting research. There are also some legal regulations regarding confidentiality. According to the Personal Data Protection Act (SEJM RZECZYPOSPOLITEJ POLSKIEJ, 1997a), any person has a right to have his/her personal data protected (1§1). Personal data shall mean any information relating to an identified (mostly name, surname, address) or identifiable (e.g., someone's photo) natural person (6§1).2) The data protection legislation obliges the researcher to obtain consent for the processing of someone's personal data3) (23§1) (unless, e.g., the data after being used are immediately removed or rendered anonymous, 2§3). Furthermore, the researcher should implement technical and organizational measures to protect the personal data being processed, and in particular to protect data against their unauthorized disclosure, being taken over by an unauthorized person, processing with the violation of the Act, or any change, loss, damage or destruction (36§1). In turn, according to the Polish Penal Code (SEJM RZECZYPOSPOLITEJ POLSKIEJ, 1997b, 266§1), whoever, in violation of the law or obligation s/he has undertaken, discloses or uses information with which s/he has become acquainted with in connection with, inter alia, scientific activity pursued shall be subject to a fine or penalty (however there are some exceptions, e.g., knowledge about murder). Thus, Polish researchers enjoy much autonomy in making decisions concerning confidentiality, although they are limited by the rules of law. Second, my study is novel—compared to the work by WILES et al. (2008), which mainly focuses on the issue of breaking confidentiality—because of its particular emphasis on the approach that researchers take to confidentiality and the strategies they employ to keep information confidential. I analyze topics still unexplored in the literature, such as researchers' attitudes to the protection of information in conversations about the study and its participants that they might have with others. 
This study is based on in-depth interviews and e-mail correspondence with 42 Polish qualitative researchers working with vulnerable groups and individuals.4) I assumed that the term "vulnerable groups and individuals" refer to people who are particularly susceptible to harm due to their unprivileged position. Many international guidelines for research ethics (e.g., Belmont Report) contain the list of specific groups that should be treated as vulnerable (HURST, 2008). However, this approach has been criticized by researchers both for not considering some groups and for including too many categories. Moreover, it was also pointed out that vulnerable group reasoning is associated with paternalism (by denying agency), reductionism (it does not take into account situational vulnerability), and essentialization and stereotypes (all persons belonging to a given category are treated equally as vulnerable) (PERONI & TIMMER, 2013). In addition, some researchers have also noticed that study participants do not necessarily see themselves as vulnerable (ALDRIDGE, 2015). Taking this into account, the critics of this approach propose to treat vulnerability not only in static categories (e.g., as belonging to a stigmatized group or having particular characteristic), but also as a category that is variable in time (e.g., unemployment) and dependent on the context (e.g., questions about mourning after recently death of a baby) (LEVINE et al., 2004). From this point of view, there may be different types of vulnerability depending on the cause (which may co-exist and sometimes strengthen each other). Potentially, this can affect the confidentiality management strategies in research―the choices of researchers regarding what, when, and why to conceal or disclose and how to do so. For example, a researcher may hide information related to a person's vulnerability while conducting research because no one else is aware of his/her stigmatized sexual orientation. 
I qualified the research participants as vulnerable on the basis of knowledge about the research and research participants (e.g., by reading research results or research projects). I also took into account the opinions of researchers with whom I conducted interviews. However, I did not have the opportunity to talk to people involved in their research. It is possible that their perception of themselves differs from how they are perceived by researchers or myself. Researchers with whom I interviewed were conducting research with the homeless, sex workers, the poor, terminal patients, migrants and refugees, ethnic minorities, transsexual people, gays and lesbians, people with disabilities, residents of nursing homes, prisoners and former prisoners, and individuals from dysfunctional families, as well as children and teenagers who are present or former residents of educational care facilities and correctional facilities. I conducted the interviews between February and December 2017. 
The study sample was recruited using maximum variation sampling (PATTON, 1990). My interviewees are researchers at various stages of their academic careers (professors, post-doctoral researchers, and PhD students), including 29 sociologists and 13 anthropologists from various research centers and universities in Poland. In the recruitment process, I looked through websites of universities, used my private contacts, reviewed the literature, conference and seminar programs, and followed advice of other researchers. I reached out to my potential interviewees by e-mail, providing basic information about my research project and asking for their consent to be interviewed and the interview to be recorded. The average duration of the interview was 2 hours (ranging from 1.5 to 6 hours). I—a white, female, Polish postdoctoral researcher with an academic background in sociology and cultural anthropology and experience in conducting research among sex workers—conducted all interviews. In the case of selected interviewees, certain substantive details were further elaborated on by e-mail. I conducted all interviews in Polish (all quotations from interviews are my translation). I transcribed and encoded each interview using meaning condensation and meaning categorization (KVALE, 1996). 
4. Confidentiality Issues in this Study
Bearing in mind the agency and autonomy of my interviewees, I asked them whether and to what extent they wished to remain anonymous. Eventually, none of the researchers decided to reveal their identity. Some of them believed it would have compelled them to have a greater degree of self-censorship. I also made it possible for them to edit the interview transcripts. I did this for several reasons. First, I thought that this increases the reliability of my data, because researchers are able to correct errors and also to clarify or deepen certain issues. Second, I thought that they had the right to do so, since the transcript is based on their knowledge and experiences. Third, I hoped that researchers could help me to anonymize interviews, especially regarding internal anonymity. I was also curious about what the researchers would do with this possibility. 
Relatively few researchers in Poland conduct qualitative research among vulnerable individuals and groups, making them easily recognizable. Moreover, my interviewees talked about their own research practices, dilemmas and problems, and at times expressed criticism of certain studies or institutions. Making this information publicly available could have affected their professional relations and further careers (WILES, CHARLES, CROW & HEATH, 2006). Therefore, I decided to refrain from providing more detailed information on my interviewees or their work. When quoting my interviewees, I only reveal whether the author is a sociologist (S) or an anthropologist (A). I discuss their comments and requests concerning confidentiality in Section 6. 
5. Confidentiality in Research Practice
Having analyzed the interviews, I distinguished between two components of confidentiality: anonymity and protection of information. First, I describe the general approach of my interviewees to anonymity (and its scope) and protection of information. Next, I discuss the strategies of anonymization and the maintenance of confidentiality. 
5.1.1 Researchers' approach to anonymization
I rely on the assumption that anonymization is one of the forms of confidentiality, which consists of concealing the identity of study participants (SAUNDERS, KITZINGER & KITZINGER, 2015). The practice is recommended in numerous textbooks on qualitative research and codes of ethics developed by associations of sociologists. For example, the code of ethics developed by the POLISH SOCIOLOGICAL ASSOCIATION (2012, §19) recommends that "the principle of anonymity should be obeyed at all times" (my translation). Apparently still prevailing in social science, this approach is based on the assumption that anonymity protects study participants from harm (VAINIO, 2013). LIAMPUTTONG (2007) suggests that concealing the identity of the participants is especially important in research with members of vulnerable groups because it prevents them from further marginalization or stigmatization. BAEZ (2002) notes that some vulnerable individuals will not participate in the study without guarantees of anonymity and confidentiality due to the fear of retaliation or embarrassment. However, he points out that although such practices protect respondents against the risk of injury associated with the disclosure of their identity, they simultaneously prevent the change of oppressive structures that generate this risk. In addition, anonymization has a methodological justification. For example, concealing the identity of participants may serve as a guarantee of authenticity of their statements (TAYLOR, 2015) and one of the conditions of critical presentation of study results (VAINIO, 2013). However, an increasing number of researchers, especially those dealing with participatory research with vulnerable groups, take a critical approach to the universalization of the anonymity principle (CONNOR, COPLAND & OWEN, 2017; GIORDANO, O'REILLY, TAYLOR & DOGRA, 2007; TILLEY & WOODTHORPE, 2011). One of the arguments they raise is that disclosing the identity of participants in participatory research (participants are partners and co-researchers) and at their own demands might lead to their empowerment—a factor of particular importance in work with vulnerable groups. Supporters of this approach believe that many research contexts require that participants be asked whether and to what extent they want to be anonymous. Thus, they oppose the imposition of the principle of anonymity by ethics committees and data protection legislation, which restricts the autonomy of both the researcher and study participants (however, some exceptions are possible, e.g., based on the explicit, informed consent of the person who wishes to be named). 
In their study, WILES et al. (2008) sampled British researchers working with vulnerable groups; the authors pointed to two opposing approaches to anonymization. One of them was represented by researchers who anonymized participants on the grounds of feeling responsible for them. The other approach, assuming the engagement of participants in the anonymization process, was adopted by researchers working in emancipatory or participatory paradigms, who also supported the idea of collaboration and non-hierarchical relations between the researcher and the researched. WILES et al. believe that these two orientations should be considered to be the extremes of a wide spectrum of approaches to anonymity, rather than as constituting two separate categories. 
Most participants in my study considered anonymity of the researched to be an ethical standard. In the words of one such researcher: "One of the basic principles of qualitative sociology is ensuring the anonymity of participants, so I did not have any second thoughts about it" (S).5) The researchers believed they should protect the identity of participants, which to a large extent translated into control over both the content to be anonymized and the strategies of anonymization. They took responsibility for the participants and the research process. A few of the researchers who conducted participatory research allowed their participants to decide whether and to what extent they wanted to remain anonymous. Thus, their approaches seem to resemble the two aforementioned attitudes to anonymization known from the literature. However, as I discuss in greater detail later (Sections 188.8.131.52 and 184.108.40.206), the approach of my interviewees to anonymization was slightly more complex for two reasons. First, their views only partly overlapped with the research method they adopted, and second, those who assumed anonymity of study participants differed with respect to the importance they attached to anonymization. 
220.127.116.11 Approach to anonymization and methodological perspective
Only six out of 42 participants regarded themselves as participatory researchers (or as working in an emancipatory paradigm). Three of them gave their participants some freedom, at least initially, as to whether and to what extent they wished to be anonymous. They mainly consult the research results regarding anonymity with their participants due to ethical reasons (giving recognition to the informants). However, one of these researchers no longer proposed the disclosure of identity to the participants, having observed that it affected their responses. Moreover, another researcher restricted collaboration with study participants concerning anonymization in order to protect their wellbeing:
"It turned out during the study that it is a closed group, everybody knows everybody, there is gossip, et cetera. I did not want to make them a target of personal attacks. For the same reason, I decided to change names, even if someone agreed to be named under each statement. At one occasion, a girl changed her mind and no longer wanted to be identified as the author of all her statements. Therefore, when I sensed that some information was intimate and might turn into an object of mockery, I changed the name" (A). 
Unlike in the study by WILES et al. (2008), other interviewed participatory researchers a priori assumed the anonymity of the researched individuals, considering it a pillar of professionalism and research ethics. They did not consult either transcripts or the research results with their participants. Such discrepancies in the approach to anonymization among researchers who work in the participatory paradigm may result from the poor institutionalization of research ethics in Poland. Initial ethical assumptions may be freely adjusted to the development of the research process, and diverse concepts of ethics in social studies may be mixed and mingled at will. 
Paradoxically, researchers who opted for non-participatory paradigms do not differ too much with regard to collaboration with participants on anonymization from participatory researchers. Many of my interviewees, both sociologists and anthropologists, would send study reports before publication to at least some of the participants—in contrast to KAISER (2009), who found that anthropologists, more often than sociologists, consult study results with participants. However, most of them consult with the participants because of methodological rather than ethical reasons, since it would enhance the accuracy of their interpretations (in contrast to the three afore mentioned participatory researchers). Moreover, I observed no links between the interviewees' identification with sociology or anthropology, or between the area of their research, and their approach to anonymization. By contrast, only rarely did the researchers working in non-participatory paradigms allow researched individuals to choose whether and to what extent they wanted to remain anonymous (this applied mostly to the participants who are public figures). Some of them would not even agree to disclose the identity of a participant at their explicit request. They tend to argue that this would violate the ethical standards or their personal beliefs. Moreover, the researchers who applied non-participatory methodology rarely consulted transcripts with the researched. This was not necessarily motivated by sheer reluctance—some of them did not consider it necessary or simply observed a lack of interest on the part of the participants. Still others did not even consider such a possibility, although they admitted that it might be worth doing so in future studies. Conversely, those who were indeed reluctant to consult transcriptions with participants expressed their concern about two basic issues. First, they were afraid that the researched might choose to modify the transcript and thus strip it of its authenticity, and, second, that the informants might want to redact certain significant parts of the text or even go so far as to withdraw their consent for participation. SAUNDERS et al. (2015) pointed to other limitations of this anonymization strategy, although they considered it valuable and used it themselves. In their opinion, such a practice takes a lot of time, which can be problematic in the context of time constraints of research projects. In addition, they noticed that participants in the study do not necessarily have the knowledge and skills associated with anonymization that the researcher has. To sum up, nearly all my interviewees outside the participatory paradigm adopted anonymization of study participants. Their decision to conceal the identity of the informants was based on ethical standards that were taken for granted, not on a reflexive consideration of pros and cons. 
18.104.22.168 Two approaches: Protective and balanced
Concealing the identity of participants poses the challenge of finding the right strategy of anonymization that protects study participants, on the one hand, while ensuring accurate reporting, on the other (BAEZ, 2002). Excessive anonymization may reduce the quality of the data and undermine the value of the results. For example, NESPOR (2000) argues that anonymization of the research site leads to the omission of the historical and geographical context of information, which may affect the interpretation of the research findings (about the consequences of the anonymization in the context of re-use of data, see e.g., SLAVNIC, 2013). On the other hand, insufficiently concealing the identity of the informants may expose them to harm (e.g., SCHEPER-HUGHES, 2000; STEIN, 2010).The research participants might be identified by some members of their community/group or by outsiders (e.g., journalists). This is particularly important in research with vulnerable groups and individuals in which the risk of harm is greater, for example because of their low social status or dependent position. This challenge was also experienced by those of my interviewees who assumed anonymity of participants. Some of them prioritize the participant protection―I refer to this approach as protective. Others looked for a balance between protecting participants and maintaining the integrity of the data―an approach I refer to as balanced. The former entails rigorous anonymization of study results (anything which is not of key importance to the analysis), mainly not to abuse the trust of participants or expose them to harm and discomfort:
"It is a principle of not causing harm. I would hate to discover that someone read my report and guessed who the participant was, while the participant still struggled with the same problem, such as dissatisfaction with their job or a family conflict. People share such information with me and I sometimes include it in my text" (A). 
Some researchers adopted the protective approach because they found it difficult to discern what others might consider to be sensitive knowledge. Therefore, they anonymized "just in case." As one researcher stated, by adopting such an approach, the researchers may want to hedge themselves against the lack of consent on the part of the participants for a certain interpretation or way of presenting them in the text. Those who took the protective approach typically anonymized data in the interview transcript. Others concealed personal data of participants in the study results. Several researchers admitted that when in doubt as to whether a piece of information might facilitate identification of the participant, they chose not to include it in the text, even for the price of undermining data integrity. Several of my interviewees noted that they wanted to make use of more anonymization, but the research project manager did not permit them to do so. This type of conflict is rarely discussed in the literature (SURMIAK, 2016). On the other hand, one researcher stated that she would not anonymize so much data if it were not for the study participants, who "sometimes are not only concerned, but experience irrational and unjustified anxiety about being identified" (S). The researcher did not want to say that her research participants are not vulnerable. On the contrary, it seems that she considered them—using GOFFMAN's category (1963, p.4)—as "discreditable" persons, who try to conceal certain potentially stigmatizing features or facts. It was rather that in her opinion, the risk of identifying the participants in research publications was smaller than they thought. 
Researchers who adopted the balanced approach also bore in mind the wellbeing of their researched; however, they did not try to anonymize the maximum possible amount of research material. On various occasions, they just relied on common sense to decide what data should be anonymized and at what stage of the research. In their opinion, anonymization makes sense only when faced with a real risk that the participant could be identified. The researchers noted that the risk of identification is minimized by the restricted readership of scientific publications. One person pointed to the fact that some researchers tend to anonymize an excessive amount of material: "I have observed it working on bachelor's theses or talking to students. They would go to any lengths to conceal the identity of their informers, believing that anything could harm them. I think it's gross exaggeration, hypersensitivity" (A). 
Some researchers adopted a mixed view based on both approaches—protective and balanced. It seems that both perspectives may be, to a greater or lesser degree, dominant, and at the same time prone to alteration, depending, for example, on the concrete research problem. 
5.1.2 Anonymization strategies
The results of numerous studies point to the fact that guaranteeing confidentiality to study participants is not easy, especially if they are part of small communities (e.g., ELLIS, 1986; STEIN, 2010). Insiders may easily recognize other insiders who participated in the study―a situation referred to by TOLICH (2004) as a threat to internal confidentiality, and by KAISER (2009)―as deductive disclosure. Many textbooks on social research, as well as codes of ethics, recommend that arrangements relating to anonymity constitute part of the informed consent. By contrast, any guidelines concerning concealment of participants' identities—apart from not using their real names and surnames and removing any other data which might help identify them—are rarely to be found. In this context, it is worth referring to anonymization strategies used by SAUNDERS et al. (2015) in research with people who have a relative with catastrophic brain injury. The researchers considered anonymity of participants very seriously due to the fact that they obtained sensitive and personal information from a specific and rather small group of participants. They distinguished six key areas of anonymizing: 1. people's names; 2. places; 3. religious or cultural background; 4. occupation; 5. family relationships; and 6. other potentially identifying information. Thus, they chose pseudonyms that do not reveal the ethnic/cultural backgrounds of participants. To avoid a situation in which the reader recognizes a research participant by combining various information about that person, they sometimes dropped a pseudonym or created a smoke screen by using two pseudonyms for the same person (particularly in the case of sensitive information or in a context that can make a person recognizable). Researchers also commonly replaced identifying places with numbers or generalized descriptions. However, to avoid entirely decontextualizing research settings, they kept some information about the place of research (e.g., the names of countries) that was important for interpretation. In the other areas of anonymization, if it was not significant for the analysis they replaced some information (e.g., about religious beliefs) with a similar one or with generalized descriptions. In addition, they sometimes removed identifying information or altered it, for example, if the information was both important for the analysis and very identifying (e.g., the name of an illness). Their decisions on what to anonymize was based on a certain balance between the protection of the participants' identity and the maintenance of data integrity. To reach that balance, they discussed what to anonymize with each other and with the interviewees. As SAUNDERS et al. write: "We have shown that anonymization is not, in practice, something that can be done on automatic pilot with 'one size fits all' or 'find and replace' approach, and have highlighted some of the challenges we faced in one particular dataset" (p.627). 
My interviewees informed their informants about study participation being anonymous when requesting their consent, typically in the oral form. In many Western European countries, the researcher is required to receive written consent from the participants, although this seems to be less frequent in ethnographic research. In Poland, on the other hand, study participants are rarely offered to sign a written informed consent in qualitative academic research in social science unless, for example, the research is sponsored by a European institution. One of the reasons for the lack of this practice may be the aforementioned poor institutionalization of ethical control. Moreover, as follows from my analysis, some researchers themselves, especially ethnographers, are also reluctant to formalize ethical aspects of the research process. 
My interviewees were aware of the fact that ensuring full anonymity to the researched individuals is difficult, and this awareness affected to a lesser or greater extent their research practice. As mentioned before, the researchers who adopted the protective approach anonymized more data than did those who applied the balanced approach. Furthermore, the researchers with a protective approach were more likely to give aggregate information about all participants and, instead, avoided attributing demographic features to specific individuals—in some cases, even gender. The researchers' decisions concerning the data to be concealed depended on the study area, the significance of certain information for the analysis, and in several cases, also depended on the decision of participants. Some researchers prioritized anonymization of individuals that could potentially suffer the most harm if their identities were disclosed—for instance, sexual minorities. Researchers of small communities or groups were in a particularly difficult position. One of my interviewees stated: "Certainly, for someone who knows this community, there is no anonymity; this is a small group of people. For many, this is not a code which could guarantee one hundred percent anonymity to anyone" (A). Many researchers seemed to ignore the problem of fictitious internal anonymity status in their research, attaching more importance to external anonymity. 
The anonymization strategies applied by my interviewees referred to six areas of anonymization discussed earlier, which were distinguished by SAUNDERS et al. (2015): people's names, places, religious or cultural background, occupation, family relationships, and other potentially identifying information. My interviews show that many researchers distinguish one additional area of anonymization: information related to the participant's biography, for example, place and date of birth, schools attended, and important life events, like the specific circumstances of death of a loved one, etc. 
Researchers in my study usually do not use their participants' surnames, and in most cases, they do not provide their first names (however, some make exceptions for public figures or―like some participatory researchers―for the participants who want to be named). This also applies to third parties appearing in research, especially if they are from a family circle or close friends of participants. Some of the researchers used only the numbered interviews or informants. The others used pseudonyms.6) One researcher who was working in a participatory paradigm gave the participants the opportunity to choose their own pseudonyms. The researchers working in research teams used only first names of research participants or pseudonyms in their communication with each other (personal data were in separate tables). Unlike SAUNDERS et al., my interviewees did not create a smoke screen, but rather prevented the reader from recognizing the research participant through quoting only short statements (to avoid including too much information) and compilation of stories told by the researched (especially in biographical studies). The latter strategy (not used by SAUNDERS et al.) also involved some fictionalization. For example, one of the researchers described her own story as a story of another person―a fictitious participant in her research. In addition, if it was not significant for the analysis, the researchers in my study used a few other strategies of anonymization in the areas mentioned, such as removal, change, and generalization. 
My interviewees removed information about informants or their relatives, which were specific in a given context, for example: occupation, place of work, nationality, religion, names of schools, hobbies, military rank, sometimes details of certain life events, or the architectural style of the building, etc. As the researchers said, sometimes the participants of their research asked for the anonymization of some information. For example, in studies involving veteran soldiers who were injured, the interviewees asked the researcher to anonymize the part of the body that they lost. 
The researchers in my study also often used generalization. Similar to SAUNDERS et al., this strategy of anonymization was applied in order to avoid entirely decontextualizing research results and to not completely give up some data. The researchers usually generalized some sensitive, intimate, or very identifiable stories told by participants. They also provided an approximate age of participants or gave only general information about place of work (e.g., "small company") or place of research (e.g., "village in the eastern part of Poland"). One researcher noted that such anonymization of the place of study was necessary in the context of discussing the marginalization or exclusion of the researched person, as "the term 'excluded' is not a compliment. ... It is stigmatizing ... and may offend not only a particular individual, but the entire group or community" (S). This statement indicates that the researcher anonymized the place of research to prevent stigmatization not only of research participants, but also other members of the vulnerable and marginalized community who lived in that place. 
Researchers in my study also changed some characteristic features, for instance, gender or age of participants or participants' children. Some of them also replaced the name of the research location with a name of place of similar size (although they often left the names of big cities),7) or slightly changed the context of the story told by the participant (e.g., by changing the actors involved in the events). Importantly, the researchers tried to ensure that these changes did not affect the meaning of the content. In the literature, the change of non-relevant details is a frequently used strategy to protect anonymity. BAEZ (2002) notes, however, that in a certain context, for example, research on discrimination, altering data is not adequate for two reasons. First, it decreases the significance of the problem analyzed. Second, it "undermines critical agency because it can accomplish the same thing as confidentiality: keeping oppressive power arrangements hidden" (p.41). 
Some of my interviewees also anonymized photos included in their publications. First, they asked informants which photos they could publish on the project's website or in publications and which photos could only be used for their analysis. Second, while selecting photos for public audience research, participants were asked to consider their own privacy and that of other people. The researchers did not edit photos, but they did not publish photos that would show someone's face. In addition, they also concealed the information about which participant was linked to which photo. They selected pictures from photos that the interviewees have agreed to publish to make sure that research participants' identities remain hidden. However, sometimes this caused problems. For instance, one of the researchers asked the participant who was living in poor social conditions to take a picture of his/her biggest dream. The participant took a picture of his/her destroyed house, because the biggest dream of this person was the renovation of this house. The researcher doubted whether it was possible to anonymize the photo of someone's home. Finally, the researcher chose a photo that contained only a part of this house, which illustrated the condition of it, but was not directly identifiable. It seems that the researchers anonymized photos despite the consent of the participants to publish them, because they thought that the photo (especially published on the Internet) could be used to trace an identity of the research participant. Two interrelated reasons appear to influence the researcher’s decisions: on the one hand, the sense of the researcher's responsibility, and on the other, the vulnerable situations of research participants whose identification could, for example, expose them to stigmatization. However, several researchers did not attach any importance to photos. For example, they would publish a photograph of a participant (not captioned with his or her name) without the consent of the individual appearing in the photo, but with the consent of the leader of the researched group or community. Those researchers adapted themselves to the existing social hierarchy rather than acting in accordance with the ethical research standards. 
5.2 Protection of information
I assume that information protection involves the researcher not disclosing information unless researched participants consent to its disclosure and even then, only in ways to which they agreed. This means that the researcher only uses some information (obtained during the research) in a specific context (mainly scientific), in a specific way (to some extent anonymized or not), and in a specific form (e.g., scientific publications, lectures). In addition, the protection of information also applies to the safe storage of information (e.g., anonymized, with passwords) and, if so, how and to whom it will be made available. I assume that the researcher is responsible for protecting information both during research and after its completion. This involves, inter alia, such questions as: What can be disclosed and to whom? Who should be protected? Why? What strategies should be used to protect information during various stages of research? How should we safely store material from research? These questions seem particularly significant in the case of vulnerable participants because of their susceptibility to harm. Therefore, I wonder in what way (if any) conducting research with vulnerable participants affects the protection of information. 
In the literature, protection of information is discussed mainly in relation to two issues. First, some textbooks, especially those on qualitative studies, provide guidelines on protecting research material. The number of publications on securing data confidentiality from re-use by other researchers is also growing (e.g., CORTI et al., 2000). In this context, major considerations pertain to the individuals who may have access to the data and the conditions that must be fulfilled to obtain that access. A second, equally important issue is the breach of confidentiality and the related consequences for the researcher and the researched. Despite the existing legal or institutional regulations, the decision on whether to disclose research material or personal data of the researched individuals may be problematic and spark controversies (e.g., GIBSON, BENSON & BRAND, 2013). An interesting analysis of the latter was made by WILES et al. (2008) on the basis of the study of British researchers. However, the authors paid little attention to the attitude of researchers to information protection or the strategies used. I analyze these issues below (in Sections 5.2.1 and 5.2.2). 
5.2.1 Attitude of researchers to protection of information
22.214.171.124 Protection of information vs. research methodology
Most of the researchers in my study took responsibility for protecting the information collected from the researched and for the decision concerning the methods of protection. Some researchers who work in the participation or empowerment paradigm were more open to consulting the researched individuals on this matter. For example, they negotiate with the participants whether and what of their statements they would be allowed to publish. However, most of them at least sometimes decide to protect certain information about the researched without seeking their opinion and they―similarly to the researchers using other methods―explained that by doing so, they took responsibility for the participants and their wellbeing. One of the researchers commented as follows on her research in visual anthropology:
"It was an extremely poignant story and would have done its job in the film, but I did not use it anyway. He didn't even ask me for it. He told me about his son on many occasions. I took the decision myself because I knew that ... it was not something I wanted to play with just to add some drama to the story" (A). 
In addition, if the researchers working in the participatory or empowerment paradigm consult with the researched individuals about the issue concerning information protection, they usually mean the content of publication. Generally, they themselves decide what and to whom they could talk about their informants and research field and how to store and secure data. 
Other qualitative researchers in my study also make such decisions on their own. Moreover, unlike researchers from the participatory research or empowerment paradigm, their decisions also concerned what to publish and how. However, the researchers respect the participants' requests not to analyze or publish some information. For many of them, the challenge was how to treat contextual knowledge (i.e., that acquired outside of official research procedures): can they use it, and if so, where and how? This was particularly important in the case of research carried out in the group or community to which they belong. Researchers felt that the use of contextual knowledge could be regarded by some members of their group/community as a "betrayal" of loyalty and a breach of confidentiality. Many ethnographers also pointed out that the protection of information during research is extremely processual and contextual. Some of them also emphasized that the knowledge of intimate secrets of research participants and various conflicts of interests in the studied community/group requires the researcher to be mindful and cautious at every stage of research. Often, this also means refraining from publishing intimate knowledge about other people. Some ethnographers claimed that they knew their research participants enough to know which information they would consider publishable and which they would not. 
The researchers' attitudes to protect information were complex, and therefore it is difficult to analyze them in terms of the two types of approaches distinguished in relation to anonymity: protective and balanced ones. The attitudes and strategies of researchers depend, among others, on the type of information, research situation, research area, and the form and circumstances of their (potential) disclosure. In the latter section of this article, I discuss an issue which is rarely analyzed in the literature, but which was prominent in my interviews and is ethically significant, i.e., disclosing information in casual conversions with other people. 
126.96.36.199 Revealing information in casual conversations
As part of the informed consent, researchers commonly communicate the methods of storing, using, and possible access to the research material. However, the question arises of whether the consent covers casual conversations about the researched, the research itself or sharing particular information with other researchers, friends, or family. As noted above, the literature rarely discusses or even touches on the issue. It seems, however, that there is a difference between sharing experiences and emotions from research with other researchers and with "lay" people (e.g., family, friends). Sharing research information is often part of the research work (e.g., in the case of team work, biographical method) and a necessary practice for researchers working with vulnerable groups. 
Overall, my informants ensure confidentiality to the researched, that is, they explain to the researched who might access the data, how it might be accessed, and how it would be used (mainly in publications). However, they rarely considered confidentiality in the context of talking with others about the researched or the information obtained from them (with the exception of information about other members of the research team). My analysis reveals that it is an undefined aspect of research ethics that is open to various interpretations. The approach of my interviewees to the issue depended predominantly on four factors: 1. the receiver of information; 2. the way of providing the information; 3. the purpose of providing it; and 4. the research method (to a certain extent). 
My interviewees approved of researchers talking about their study subjects or events that occurred in the study process if they did not share any personal data and talked to other researchers with the aim of consulting or gaining more knowledge. In fact, it was part of the research process if a study was conducted by a team. Moreover, many researchers reported an inner need to talk about their experience with others to release the emotional load they accumulated during their research. Most of them confided their problems and dilemmas to their co-workers, colleagues, or close friends and family. However, they felt a lack of professional supervision. One of my interviewees stated, "there is room in sociological studies for someone who may be dubbed a supervisor―someone you could turn to for advice when in trouble, when you can no longer handle it on your own" (S). Many scholars agree that professional counsel offers particularly useful support for researchers dealing with sensitive topics (e.g., CORDEN, SAINSBURY, SLOPER & WARD, 2005; JOHNSON & CLARKE, 2003). It seems that supervision could help researchers not only to manage their emotions, but also to make decisions concerning ethics by giving them the opportunity to discuss the difficulties and dilemmas they experienced. 
The participants of my study considered it an obvious breach of confidentiality if the researcher were to disclose facts from someone's life, identifying the individual by their name and surname, without their consent, especially for his or her own benefit, and in spite of the existing risk of harm. Less extreme cases were more contentious. I discuss two of them below. Some researchers, especially those who conduct ethnographic studies, did not consider it necessary to anonymize data if the identity of the researched could not be concealed for other reasons. They would talk about the researched with their key informants (or gatekeepers) and did not consider it a breach of confidentiality. It was part of the research, or even more―part of the researched community's life. In such cases, the decision as to what does and does not constitute protection of information depended on the context. The researchers also applied the principle of consequentialism, that is, evaluated the potential consequences of their actions, and avoided taking action which would have put the researched at risk. On the other hand, some of my interviewees claimed that anonymization of study participants is not enough to protect information. In their opinion, talking about someone may be unethical even if personal data is not revealed and information is being shared with other researchers. As an example, they referred to a situation in which a researcher judges the behavior of a study participant or uses information obtained from them as an anecdote. This approach seems to be based on a deontological theory, according to which a given behavior is wrong when it violates fundamental moral principles—in this case, the principle of respect for other people (i.e., the researched). 
I do not think it to be necessary, or even possible, to find a universal rule that would regulate this issue. However, the question is important and warrants further analysis because decisions concerning confidentiality in informal conversations may have important consequences for the informants (e.g., cause harm), the researchers (impact their reputation), or the research itself (e.g., informants may be reluctant to take part in future research if they feel that a researcher reveals too much information about them to other people). 
5.2.2 Methods of data protection
My interviewees protected the information they obtained at various stages of the research process. They mainly focused on two issues: concealing the information (e.g., information about hidden sexual orientation, criminal activity) and potentially interfering in the social settings in which their researched individuals lived, exposing them to harm and protecting the research data from unauthorized access. The first issue was often associated with the positivist belief that the researcher should be impartial and neutral. In addition, it resulted from a sense of responsibility for the consequences of the disclosure of certain information and an ethical obligation to avoid harming the participants. My participants try to conceal this type of information in various ways. For example, some of them understand the protection of information to be non-disclosure of what the researched said or did. Consequently, they faced the challenge of concealing the information from those interested in what a particular participant had to say. They apply various strategies, one of them being the reciprocity of perspectives. In short, the researcher justifies their refusal to reveal information and encourages the asking person to imagine themselves in the situation of the person s/he is asking about. For instance, this strategy is successfully used by researchers dealing with institutions, such as nursing home or correctional facilities, where they sometimes felt pressure to "report back" to staff (JOHNSON & CLARKE, 2003). Some researchers mentioned that it is impossible to predict negative consequences of revealing information about participant's behaviors, because they depend on staff and management. The staff could use information about its residents against them, for instance, the information that a resident of a nursing home keeps an animal against the regulations. Because of this, the researchers did not breach confidentiality even when some residents’ behaviors seem to be harmful to them, like taking drugs (though most researchers believed that they would intervene in a life-threatening situation or where there was serious harm). However, sometimes even a suggestion or comment from a researcher may harm someone, and that is why when you approach the relevant institutional authorities, as one of my interviewees said: "you have to be very careful with what you say, who you speak with, who you drink vodka with, what is being said during such meetings, because everything can have meaning" (S). 
In some cases, it was necessary that the researchers supported the false image of the researched in their closest circle. For example, one researcher commented as follows on her experience with a transsexual person:
"I suppose it was the most difficult experience for me. While talking to that person, I had to learn to switch from the feminine to the masculine form, or vice versa, depending on whether we talked one-to-one or in the presence of other people, who did not know what that person was sharing with me. To make things even more difficult, the daughter showed up, to whom that man was the mother" (S). 
According to this quotation, the research participant concealed his sexual identity even around his daughter. The participant was biologically female, but he felt like he was a man. The researcher hid this information (e.g., using appropriate personal forms), not wanting to expose him to harm and disruption of his relationship with his daughter. It must be added that transsexual persons are vulnerable due to being susceptible to stigmatization in Polish society. 
The researchers protected information from unauthorized access and when transcribing interviews or publishing study results. For example, one of my interviewees undertook to transcribe interviews herself due to the intimate and private character of the collected data. Some researchers, who commissioned other people to transcribe interviews, imposed on them the obligation of confidentiality, in some cases even in writing. When publishing study results, many researchers decided not to disclose material they considered too personal, or attempted to anonymize it to the largest extent possible. 
The researchers needed to decide on their own how to protect the study material. They regarded this as an important issue, especially given that some of their research participants were afraid that their interview would be publically available on the Internet or could be read by authorities and, as a consequence, that they might be punished for who they are or for their views. The researchers typically recorded the data on external disks that were then stored at home. Occasionally, they used additional security, such as passwords to files containing the material. One researcher noted that she/he had an external disk with study data stolen from the university. My interviewees underlined a lack of relevant systemic solutions or recommendations: "There is nothing ... nobody even stores it, nobody talks about it, and nobody brings this up" (S). This statement well illustrates the gap between the formal requirements set for researchers in Poland and the available assistance in their implementation. The legal regulations related to the protection of personal data oblige the researcher to secure the research material. Similar requirements are contained in codes of ethics. For example, according to the "Code of Ethics" developed by the POLISH SOCIOLOGICAL ASSOCIATION (2012, §21): "Appropriate steps must be taken to guarantee the safe storage of research data. If it is possible, techniques of protection of anonymity should be used, such as removing identifiers, using pseudonyms and other technical measures concealing the relationship between data and identifiable persons based on them" (my translation). However, apart from these general guidelines, there are still a few recommendations on how to secure data safely (including how to store them safely during and after research). Researchers in my study who collaborate, for example, with researchers from Germany or Ireland believed that in Poland, the data protection standards are much lower compared to these countries and that the Polish academic community is less aware of this matter. In fact, a discussion about storage and archiving of qualitative data has a rather short history in Poland. Only recently have some qualitative data archives been set up (e.g., the Qualitative Data Archive at the Institute of Philosophy and Sociology of the Polish Academy of Sciences, which was created in 2013). However, this is still a relatively poorly recognized issue, requiring broader discussion and attention (STRACZUK & FILIPKOWSKI, 2014). 
6. Researchers' Comments on Confidentiality in my Study
Nearly three quarters of my interviewees wished to see the transcript of their interview, even despite the fact that most of them―like the researchers studied by WILES et al. (2008)―did not show transcripts to those participating in their own studies. Additionally, several interviewees requested that I send them the quoted pieces of their interviews before they are published in a paper or book. This shows that some of the researchers treated themselves differently than they treat their study participants. The application of double standards may result from the awareness of recognizability in their own professional circle and the necessity to control information and their own image. Perhaps it also results from the fact that they are researchers themselves and know how the process of analysis proceeds, which makes them less trusting towards another researcher (e.g., they know that the researcher can use certain statements in various ways, not necessarily in accordance with the intentions of the participant). 
My interviewees were given the opportunity to modify the transcripts, that is, add, remove, and change the text, as well as indicate the issues that, in their opinion, required further anonymization. I also asked them to mark sensitive data and pieces that I was not allowed to quote (if any). Half of the researchers who were sent the transcripts made some comments on the text. The comments, like those in the study by WILES et al., mostly concerned anonymity and image protection, although in a different sense. Participants in the WILES et al. study modified the content of the transcripts, which only occasionally happened in my study―perhaps because I was more interested in the researchers' experience rather than their opinions on research ethics. 
The transcripts of interviews I sent to my interviewees for approval were anonymized. I deleted most of the proper names, including: first and last names of the participants in my study and other researchers or scholars they mentioned (also the titles of their publications); last names of other people they talk about, excluding the names of persons who held public functions; names of universities (and also names of faculties and research departments); names of villages, small towns, neighborhoods, or organizations in which they conducted research8) (sometimes also the names of cities if the research site was very specific, e.g., prison); and the names of the places in which my participants currently work or live. The researchers additionally requested me to conceal some items in their academic biographies (e.g., topics of their earlier research, the name of the journal in which they published the research results), names of towns and cities in which they had conducted research, facts from biographies of their study participants (e.g., schools attended by somebody's children), and the real names of the researched, if they had happened to use them. Moreover, they marked as sensitive or deleted statements that were critical of other researchers or institutions or violated internal anonymity (e.g., information about co-workers' genders, some stories from the field that are well known among other academic researchers or could reveal the identity of the study participants). Some of my interviewees also removed from the transcript information about themselves (e.g., zodiac sign, being a vegetarian, getting sick during their research, information about family, etc.). One researcher asked me not to provide in my future publications the exact area of his research to prevent identification. More researchers who adopted the protective approach than those with the balanced approach wished to see the transcript; they also anonymized more information. However, not all the researchers with the protective approach asked to have the transcript sent to them or made any comments. Many factors might have played a role here, such as the trust they put in me as a researcher, a different study area, the content of the interview (how much they revealed in the conversation), or the time they spent or did not spend on reading the transcript. 
Modifications on account of the protection of image applied to both the researchers and the researched. Some of my interviewees made stylistic corrections to their statements or even deleted certain interesting pieces of the text. Certainly, they had the right to do so, and I respected all such requests for deletion. In this way, they sometimes materialized their own concern that returning transcripts to the researched may affect the authenticity of text and result in a loss of data. On the other hand, some of my interviewees declared they would not edit the style or content of the text to preserve its truthfulness and reliability. I faced the problem of bringing together the two approaches―quotations in their original wording would put their authors in a less favorable light than those whose statements were edited. I decided to edit style in all the transcripts to avoid the difference. Moreover, some of my participants asked me not to quote or use in the analysis the pieces of statements that could reflect badly on study participants. They were concerned about confirming the stereotypes of individuals who belonged to certain social categories (e.g., the homeless) or acted illegally. I consider it an expression of responsibility and loyalty toward the researched on the one hand, and distrust about the way I may use such information, on the other. 
Building on the existing literature, I extend currently existing knowledge by adding a discussion of the ethical aspects of confidentiality in qualitative research with vulnerable groups. I examine how Polish sociologists and anthropologists conducting research with vulnerable groups manage issues of confidentiality. The results of my study show that researchers take responsibility for ensuring confidentiality to their participants. They are in a unique situation, considering that research projects in sociology or anthropology in Poland are not typically reviewed by ethics committees. Little institutional control of research ethics leaves researchers unsupported in their ethical decisions on the one hand, although gives them freedom to adjust decisions concerning ethics to the research context, method, and the research process, on the other hand. What is more, it creates space for various approaches to confidentiality in qualitative research among vulnerable groups and individuals. 
The results of my research are particularly interesting in the context of ethics committee recommendations found in Anglo-Saxon countries. For example, institutional review boards (IRBs) in the United States, which often relies on the clinical, biomedical, and quantitative research model, usually stress the maximum guarantee of the informants' confidentiality and anonymity. This means, for example, that the identities of research participants should be hidden not only in research publications, but also during research (i.e., they should not reveal their identity to other participants and outsiders (MILNE, 2005). IRBs also require the researcher to plan in advance how they will meet these guarantees. The practice of qualitative researchers who took part in my research indicates that such recommendations are problematic mainly for two reasons. First, some researchers working in the emancipatory and participatory research paradigm share control over the research process with the participants. For example, they allow them to decide whether or not they want to be anonymous. Second, some researchers, especially those conducting ethnographic research, approach maintaining confidentiality in a contextual and dynamic manner. They treat initial assumptions about confidentiality flexibly and modify them during research. For example, during research they get know their informants and they often assume that they know them well enough to decide which information they can publish. This also applies to conversations about the research participants with other informants or gatekeepers. Additionally, the results of my research indicate that setting requirements in advance regarding confidentiality can threaten validity of data and does not necessarily protect participants from harm. Confidentiality is a very contextual process that is simultaneously dependent on the approach adopted by the researcher. 
The approaches of my interviewees to ensuring confidentiality to the researched varied both with reference to anonymity and data protection. With respect to anonymity, the study revealed two main categories of researchers: those who let their participants decide on whether and to what extent they wanted to remain anonymous, and those who made the decision to anonymize in advance. The latter group either took a protective or a balanced approach, which had an impact on when and to what extent they anonymized research material and how they presented the findings. Further analysis is required of how taking one or the other approach affects the research process and whether researchers who adopt different approaches could cooperate with each other, for instance, as part of the same research team. Furthermore, the participants of the present study expressed different opinions on the research content (information about the researched and the research itself) that they approved to be shared and the circumstances in which it could be shared. The issue is worth exploring not only in the context of vulnerable research participants but also in the context of qualitative research in general. For instance, it would be worth discussing whether the fact of having shared research material with others should be communicated to the researched and which manner of communication to choose. 
Funding for this research was provided by a grant from the National Science Centre, Poland (2016/20/S/HS6/00435). I would like to thank both editors for their insightful comments on the article, as these comments led me to an improvement of the work.
1) Current data protection legislation remains in force until May 25, 2018, when it will be replaced by new European Regulation. <back>
2) A piece of information shall not be regarded as identifying where the identification requires an unreasonable amount of time, cost and manpower (6§3). <back>
3) There is no requirement for written consent, unless the data is sensitive, which means for example information about health, sex life, addictions, etc. However, this does not apply to the situation in which the researched is impossible or difficult to identify (since there is no guidance on how to assess it this issue is open to different interpretations). <back>
4) My project is still in progress, the results that I present in this article are preliminary and partial, although they provide some novel insight. <back>
5) "(S)" means that author of the quote is a sociologist, "(A)" that s/he is an anthropologist. <back>
6) Unfortunately, I did not discuss this issue with the researchers in detail, and therefore I have little information about the basis on which they chose the names. However, one of the researchers mixed the initials of the participants, and in earlier studies he sometimes used real pseudonyms, which, however, were quite commonly used, e.g., "fatty." Another researcher gave names to participants from a different culture than theirs. Several researchers mentioned that they did not have any pseudonym selection system. <back>
7) Not all researchers have anonymized the research site. One researcher said that the local context is an important framework for the interpretation of research results. Besides, few researchers treated leaving the name of the town or the institution where they conducted research as a way to authenticate their findings. <back>
8) One researcher, after checking the transcription of the interview, wrote to me that she was surprised that I removed the names of some towns and neighborhoods. <back>
Aldridge, Jo (2014). Working with vulnerable groups in social research: Dilemmas by default and design. Qualitative Research, 14(1), 112-130.
Aldridge, Jo (2015). Participatory research: Working with vulnerable groups in research and practice. Bristol: Policy Press.
American Anthropological Association (2012). Statement on ethics: Principles of professional responsibilities, http://ethics.americananthro.org/category/statement/ [Date of access: April 19, 2018].
Baez, Benjamin (2002). Confidentiality in qualitative research: Reflections on secrets, power and agency. Qualitative Research, 2(1), 35-58.
Barry, Ben (2017). Enclothed knowledge: The fashion show as a method of dissemination in arts-informed research. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 18(3), Art. 2, http://dx.doi.org/10.17169/fqs-18.3.2837 [Date of access: April 19, 2018].
British Sociological Association (2017). Statement of ethical practice, https://www.britsoc.co.uk/media/24310/bsa_statement_of_ethical_practice.pdf [Date of access: April 19, 2018].
Connor, James; Copland, Simon & Owen, Jill (2017). The infantilized researcher and research subject: Ethics, consent and risk. Qualitative Research, Online First.
Corden, Anne; Sainsbury, Roy; Sloper, Patricia & Ward, Bernard (2005). Using a model of group psychotherapy to support social research on sensitive topics. International Journal of Social Research Methodology, 8(2), 151-160.
Corti, Louise; Day, Annette & Backhouse, Gill (2000). Confidentiality and informed consent: Issues for consideration in the preservation of and provision of access to qualitative data archives. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 1(3), Art. 7, http://dx.doi.org/10.17169/fqs-1.3.1024 [Date of access: April 19, 2018].
Ellis, Carolyn (1986). Fisher folk: two communities on Chesapeake bay. Lexington, KY: University Press of Kentucky.
Gibson, Susanne; Benson, Outi & Brand, Sarah L. (2013). Talking about suicide: Confidentiality and anonymity in qualitative research. Nursing Ethics, 20(1), 18-29.
Giordano, James; O'Reilly, Michelle; Taylor, Helen & Dogra, Nisha (2007). Confidentiality and autonomy: The challenge(s) of offering research participants a choice of disclosing their identity. Qualitative Health Research, 17(2), 264-275.
Goffman, Erving (1963). Stigma. Notes on the management of spoiled identity. London: Penguin.
Hurst, Samia A. (2008). Vulnerability in research and health care; Describing the elephant in the room?. Bioethics, 22(4), 191-202.
Johnson, Barbara & Clarke, Jill M. (2003). Collecting sensitive data: The impact on researchers. Qualitative Health Research, 13(3), 421-434.
Kaiser, Karen (2009). Protecting respondent confidentiality in qualitative research. Qualitative Health Research, 19(11), 1632-1641.
Kay, Helen; Cree, Viviene; Tisdall, Kay & Wallace, Jennifer (2003). At the edge: Negotiating boundaries in research with children and young people.Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 4(2), Art. 33, http://dx.doi.org/10.17169/fqs-4.2.706 [Date of access: April 19, 2018].
Kvale, Steinar (1996). InterViews: An introduction to qualitative research interviewing. London: Sage.
Levine, Carol; Faden, Ruth; Grady, Christine; Hammerschmidt, Dale; Eckenwiler, Lisa & Sugarman, Jeremy (2004). The limitations of "vulnerability" as a protection for human research participants. The American Journal of Bioethics, 4(3), 44-49.
Liamputtong, Pranee (2007). Researching the vulnerable: A guide to sensitive research methods. London: Sage.
Miller, Kyle (2015). Dear critics: Addressing concerns and justifying the benefits of photography as a research method. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 16(3), Art. 27, http://dx.doi.org/10.17169/fqs-16.3.2434 [Date of access: April 19, 2018].
Milne, Catherine (2005). Overseeing research: Ethics and the institutional review board. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(1), Art. 41, http://dx.doi.org/10.17169/fqs-6.1.531 [Date of access: April 19, 2018].
Nespor, Jan (2000). Anonymity and place in qualitative inquiry. Qualitative Inquiry, 6(4), 546-569.
Orsatti, Cristina (2000). How to constitute an archive of oral memory and identity within the framework of A.P.T.O.: A few methodological proposals. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 1(3), Art. 14, http://dx.doi.org/10.17169/fqs-1.3.1031 [Date of access: April 19, 2018].
Patton, Michael Q. (1990). Qualitative evaluation and research methods (2nd ed.). Newbury Park, CA: Sage.
Peroni, Lourdes & Timmer, Alexandra (2013). Vulnerable groups: The promise of an emerging concept in European Human Rights Convention law. International Journal of Constitutional Law, 11(4), 1056-1085.
Polish Sociological Association (2012). Code of sociologist's ethics, http://pts.org.pl/wp-content/uploads/2016/04/kodeks.pdf [Date of access: April 19, 2018].
Saunders, Benjamin; Kitzinger, Jenny & Kitzinger, Celia (2015). Anonymizing interview data: Challenges and compromise in practice. Qualitative Research, 15(5), 616- 632.
Scheper-Hughes, Nancy (2000). Ire in Ireland. Ethnography, 1(1), 117-140.
Science Ethics Committee of the Polish Academy of Sciences (2017).The code of ethics for research workers, https://instytucja.pan.pl/index.php/kodeks-etyki-pracownika-naukowego [Date of access: April 19, 2018].
Sejm Rzeczypospolitej Polskiej (1997a). Ustawa z dnia 29 sierpnia 1997 r. o ochronie danych osobowych [Personal Data Protection Act]. Dz.U. nr 133 poz. 883, http://prawo.sejm.gov.pl/isap.nsf/DocDetails.xsp?id=wdu19971330883 [Date of access: April 19, 2018].
Sejm Rzeczypospolitej Polskiej (1997b). Ustawa z dnia 6 czerwca 1997 r., Kodeks karny [The Penal Code]. Dz. U. Nr 88, poz. 553 z późn, zm, http://prawo.sejm.gov.pl/isap.nsf/DocDetails.xsp?id=WDU19970880553 [Date of access: April 19, 2018].
Slavnic, Zoran (2013). Towards qualitative data preservation and re-use—Policy trends and academic controversies in UK and Sweden. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 14(2), Art. 10, http://dx.doi.org/10.17169/fqs-14.2.1872 [Date of access: April 19, 2018].
Stein, Arlene (2010). Sex, truths, and audiotape: Anonymity and the ethics of exposure in public ethnography. Journal of Contemporary Ethnography, 39(5), 554-568.
Straczuk, Justyna & Filipkowski, Piotr (2014). Archiwizacja danych jakościowych. Wprowadzenie [Qualitative data archiving. An introduction]. Studia Socjologiczne, 3(214), 161-166.
Surmiak, Adrianna (2016). Wybrane problemy etyczne w naukowych badaniach na zlecenie. Refleksje etnografki [Some ethical problems in academic contract research. Reflections of ethnographer]. Przegląd Socjologii Jakościowej, 12(3), 120-134.
Taylor, Rebecca (2015). Beyond anonymity: Temporality and the production of knowledge in a qualitative longitudinal study. International Journal of Social Research Methodology, 18(3), 281-292.
Tilley, Liz & Woodthorpe, Kate (2011). Is it the end for anonymity as we know it? A critical examination of the ethical principle of anonymity in the context of 21st century demands on the qualitative researcher. Qualitative Research, 11(2), 197-212.
Tolich, Martin (2004). Internal confidentiality: When confidentiality assurances fail relational informants. Qualitative Sociology, 27(1), 101-106.
Vainio, Annukka (2013). Beyond research ethics: Anonymity as "ontology", "analysis" and "independence". Qualitative Research, 13(6), 685-698.
van den Hoonaard, Will C. (2003). Is anonymity an artifact in ethnographic research?. Journal of Academic Ethics, 1(2), 141-151.
von Unger, Hella; Dilger, Hansjörg & Schönhuth, Michael (2016). Ethics reviews in the social and cultural sciences? A sociological and anthropological contribution to the debate. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 17(3), Art. 20, http://dx.doi.org/10.17169/fqs-17.3.2719 [Date of access: April 19, 2018].
Wiles, Rose; Charles, Vikki; Crow, Graham & Heath, Sue (2006). Researching researchers: Lessons for research ethics. Qualitative Research, 6(3), 283-299.
Wiles, Rose; Crow, Graham; Heath, Sue & Charles, Vikki (2008). The management of confidentiality and anonymity in social research. International Journal of Social Research Methodology, 11(5), 417-428.
Adrianna SURMIAK is a postdoctoral fellow in the Department of Sociology of Morality and General Axiology at the University of Warsaw in Poland. She holds an MA in sociology and a PhD in cultural anthropology. Her main research interests include qualitative research ethics, lying, and sex work.
Department of Sociology of Morality and General Axiology
University of Warsaw,
Żurawia 4, 00-503 Warsaw, Poland
Phone: +48 665 509 503
Surmiak, Adrianna (2018). Confidentiality in Qualitative Research Involving Vulnerable Participants: Researchers' Perspectives [55 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 19(3), Art. 12, http://dx.doi.org/10.17169/fqs-19.3.3099.