Doubts,dilemmas and decisions: towards ethical research on gender and schooling in South Africa

Introduction

This article describes how a project involving university researchers, school teachers and students confronted as a dilemma the very issues it was seeking to research. As the quote above indicates, South African schooling features high levels of sexual harassment, including male teachers taking sexual advantage of their female students. Concerned better to understand gender inequality, violence, poverty and its relationship to high levels of HIV, we set out to examine gender relations in a number of South African schools. The team expected to find evidence of gender inequality, violence and poverty. It did not expect to have the suspicion raised that a member of the research team was himself engaged in sexual relations with school girls.

Our project examined school regimes and was explicitly committed to contributing to a better understanding of gender inequalities in order to inform the development and implementation of policy for and in schools. The team also wished to contribute to achieving better gender relations in schools, whilst still prioritizing the data gathering and knowledge production processes.

Research that explicitly seeks to engage with vulnerable groups (schoolgirls) and focuses on sensitive issues (violence and criminal acts – sex between an adult and a person under the age of 16 years is statutory rape) is likely to generate awkward choices. Researchers may find, for example, that their respondents have been involved (as perpetrators or victims) in criminal actions and are likely to encounter countless transgressions of policy, in spirit if not in formal rule breaking, in our case those promoting gender equality and school safety. Consequently, research is normally undertaken upon agreed conditions of anonymity with an express commitment to use pseudonyms.

Our research was qualitative, involving ethnography, focus groups and individual interviews as well as a gender equality intervention. As managerial approaches to research in universities have grown (Boden and Epstein, 2006), so has the management of researchers particularly to ensure ethical practice. The assumption is that rigorous ethical oversight can prevent harm or at least anticipate it (and therefore limit its impact). We demonstrate below that, while attention to ethics in designing research is important, the complexities confronted during research are unpredictable. Conditions in the research site, particularly when the focus is on sensitive issues, will inevitably call on researchers to exercise their ethical judgement (Hammersley, 2009). Quite how researchers might do this and what sorts of choices they might face is the substance of this article.

We conceptualize ethics as the conscious practice of navigating between various imperatives in a morally informed way. Drawing on the work of Pels (1999), we discuss how researchers are forced into a position of ‘duplexity’ where, in the process of seeking ‘the truth’, they must confront their own political power. Ethics are the means by which they might do this with as little harm as possible to those whose lives they touch, but Pels (1999: 101) insists that situations involving the ‘unintentional use of “double standards”’ are unavoidable. This is largely because anthropologists (and sociological ethnographers) have a disciplinary obligation to represent their subjects and in this process, cannot but confront dilemmas that require the avowal of power and acknowledgement that research has a purpose beyond representing and supporting those who become constituted as research subjects. On the one hand, the researcher is expected to become ‘part of’ the community being studied but, on the other, that very community will be suspicious of ‘ulterior motive’ (1999: 107). In the project and situations we describe below, our dilemmas and choices flowed from the fact that we wore at least two hats – one as members of the school community trying to work collectively to pursue gender equality and the other as researchers dedicated to uncovering the gendered dynamics of school life.

In this article we describe two dilemmas. The first was what to do with ‘stories’ suggesting that a member of the research team – a teacher acting as a gatekeeper and helping with data collection – was having sex with schoolgirls. The second was how to report the first dilemma in light of the condition of anonymity we had agreed with participants.

One quandary concerned compromising the identity of participants who assisted with the research process but were also members of the school-based research constituency. That is, they were involved in two overlapping roles – facilitating the research process and providing their participant perspectives (as respondents) on their school’s gender regime. These people participated in this way on the understanding that their anonymity would be respected and protected. However, we found ourselves in a situation where anonymity was potentially at loggerheads with the ethos and aims of the research process. We confronted the question: could/should we safeguard anonymity while at the same time protecting the integrity of research that included a commitment to supporting national, regional and school policies and practices that promote gender equality?

Our first dilemma was how to broach the issue with our researcher-participant, ‘Josiah’. It was hard to believe that he was a sexual predator. He was well versed in the importance of gender equality through involvement in our project and participation in university courses. It was difficult to broach the issue because our ‘evidence’ of his alleged predation was itself open to question. We had only indirect claims of sexual relationships. The accusation on its own would besmirch his name, might lead to his loss of employment and damage future work prospects and the material well-being of a very large extended family in a situation of extremely high unemployment and widespread poverty. Additionally and contextually, in many South African schools, sexual relationships between older men and younger, school-going girls were regarded as legitimate by most of the local community and so there was possibly a mismatch between law and social norms (Wight et al., 2006).

Below we discuss how we navigated these treacherous ethical waters in the context of a collective research project based in South African schools that spanned nearly 10 years. We show that we were aware of the dangers and realized that we needed to make decisions.

The article is organized in seven sections. Section 2 examines how existing approaches seek to ensure ethical research conduct. We pay special attention to how context impacts on ethical choices, particularly where qualitative research methods are being used. Section 3 situates our work and dilemmas within their wider contexts. Section 4 describes the dilemmas bringing them together into a coherent story. Section 5 situates these events within the complicating factors produced by local cultures, material and everyday practices. In section 6 we turn to our difficulties in writing about such dilemmas. A particular problem has been to protect anonymity and we consider the costs of protecting the anonymity of some of our participants when confronting the possibility that they might have breached ethical codes and undermined commitment to gender equity. The last section pulls the article together, asking how ethical practice can be maintained in such circumstances and arguing that bureaucracies and informed consent do not provide protection for either participants or researchers. Finally, we suggest that ethical choices cannot be pure or predetermined, but involve compromise borne of a careful weighing of competing ethical and pragmatic imperatives. This article appeals for situated ethical practice rather than mechanistic approaches so often prescribed by the ‘new ethical bureaucracies’ (Boden et al., 2009).

Research ethics

A major justification for developing research ethics guidelines is that they prevent the exploitation of those being researched. The dictum ‘do no harm’ is therefore a guiding principle in both the formal ethical requirement process and in conducting research itself. Since confronting the horrors of Nazi eugenic and medical research after the Second World War, research, particularly in medicine, has become ever more regulated by protocols. The arrival of the AIDS pandemic accelerated the elaboration of ethics guidelines and these have now become a central and generally mandatory part not only of medical research, but of all enquiry that involves human subjects (see Angell, 1988; Boden et al., 2009).

A key mechanism deployed is ‘informed consent’. Research subjects are frequently required to sign a form indicating that they agree to participate in the project. In reality, this mechanism works better in some circumstances than others and in ethnographic research it is impossible to get the consent of every person who may come into contact with the researchers. Over and above these practical difficulties, consent is just the start of conducting research ethically. In this regard, Guillemin and Gillam (2004) distinguish between procedural ethics (frequently associated with formal ethics committees) and ‘ethics in practice’. They argue that ‘procedural ethics cannot in itself provide all that is needed for dealing with ethically important moments’ (p. 262). They call attention to the need for reflexive research, particularly when qualitative, which involves, among other things, ‘acknowledgment of microethics’ and ‘sensitivity to what we call the “ethically important moments” in research practice’ (p. 276).

Calling for reflexive research is neither novel nor a guarantee of ethical practice (Pillow, 2003). Nevertheless, we argue that reflexivity is more likely to ensure ethical research practice than pre-research ethical compliance. The biomedical model that informs most approaches to research ethics is particularly unsuited to projects with a qualitative research design (Van den Hoonard, 2001), which requires that trust be created, developed and maintained over time and ‘[consent] needs to be re-established on a regular basis’ (Ramcharan and Cutcliffe, 2001: 363).

This trust is based on a relationship that is negotiated under conditions that shift over time and are likely to contain contradictions and ambiguities, particularly when participants may have an interest in presenting themselves in ways that obscure or conceal elements they know will not ‘fit’ the research design. Such people may well want to continue to be involved in the research process for reasons of interest, prestige and, sometimes, remuneration. For these reasons ‘a culture of ethical awareness’ needs to be developed (Shaw, 2003: 24).

Those with whom research is conducted, the conditions under which and the places in which it occurs have profound implications for the kind of ethical dilemmas confronted. This has increasingly been recognized in critiques – which recognize the importance of ethical researching – of standardized or uniform ethical requirements (Benatar, 2002; Ijsselmuiden and Faden, 1992; Singh et al., 2006). The goal of research in developing contexts shapes ethical choices because research is often aimed at solving problems. Thus, the resources and influence of the researchers may throw into relief the relative vulnerability of participants. Ethical choices need to acknowledge the relative privilege of the researcher (with secure employment, social security of one sort or another, access to health care and so on) and the difficulties of respondents who may be grappling with substantial material and life challenges (Benatar, 2002). Under these circumstances a researcher may come to occupy an exploitative position whether she or he has complied with formal ethical requirements or not (Benatar, 1998).

Recognizing the unequal status of the various parties involved in a research project is an important starting point that allows recognition that ‘research subjects’ might want something from the process. For example, it is frequently assumed that participants wish to be anonymous. However, having one’s picture taken or one’s name in print may be desired. In our work, a frequently desired outcome was to be listened to and believed. These features were necessary constituents in building relations of trust.

Contexts

This article is based on the work of a mixed gendered and raced research team of South Africans and ex-South African working with several research assistants (all from the local area) and MA students (all from the UK, studying gender, education and development and doing small scale ethnographies in one of our schools). In different formations, the team has worked together since 1996 investigating questions of gender and sexuality in South Africa in the context of HIV/AIDS. ¹ We used a variety of mostly qualitative methods, including small scale ethnographies, group and individual semi-structured or conversational interviews with learners, teachers, principals and provincial government officials with a brief for ‘life skills’ or ‘life orientation’, ² and life histories as well as a survey of learners aimed at establishing a baseline of knowledge about HIV/AIDS and a review of reports of HIV/AIDS interventions across South Africa aimed at young people.

In order to protect research participants, it is necessary to be much less specific and detailed about our research sites than would normally be our practice. We would prefer to provide ‘thick description’ (Geertz, 1973), but this would risk revealing the identities of research participants. Nevertheless, some context is necessary and what follows provides this in general terms in relation to the schools where we worked.

Schooling in South Africa still bears the marks of apartheid, though the divisions between them are now related to racialized class rather than race on its own (Chisholm, 2004). Older apartheid-driven racialized categories – ‘white’, ‘African’, ‘coloured’, and ‘Indian’ – continue in daily and official use in the new South Africa. In the richer, suburban middle class areas, the formerly white only schools are now mixed race but almost entirely middle class, while schools in the townships, those serving informal settlements ³ and those in rural areas, remain entirely African as far as their learners are concerned (though a few teachers may be from other groups). The particular school in this article is a township with only African teachers and learners. For most, English was their second language and many were embedded in local custom and practice particularly in relation to gender and sexuality. Such schools serve a diversity of working class and unemployed people, many of them living in multiply deprived and extremely poor populations with high levels of HIV infection (Taylor et al., 2003). Infection levels are particularly high among young people (aged 15–24) with a strong gender bias (15.5% among girls and 4.8% among boys) (Pettifor et al., 2005).

South Africa is one of the countries most affected by HIV/AIDS worldwide. It has the highest number of children living with HIV in the world with an estimated 280,000 children under 15 living with the infection (UNAIDS 2008). In tandem with this are high levels of gender-based violence (Coombe, 2002; Jewkes et al., 2001; Varga, 1997; Wood and Jewkes, 1997). Some researchers (e.g. Whiteside, 2002) link sexual violence with the prevalence of HIV, while Dunkle et al. (2004) report that women at antenatal clinics who have experienced violence in intimate relationships are significantly more likely to be HIV+ than those who have not. Nearly three-quarters (72%) of pregnant teenagers and nearly two-thirds (60%) of teenage girls who had never been pregnant reported coerced sex (Jewkes et al., 2001). ⁴ While many young women receive and refuse sexual advances, nearly three-quarters unwillingly succumb to sexual demands (Varga, 1997).

Human Rights Watch (2001) claims that sexual liaisons and coercive sex between learners and teachers are widespread in South Africa and, although their methodology has been widely criticized, their findings have been substantially confirmed by other studies (Jewkes et al., 2002: 319). More recently, the 2008 National HIV Survey found that 33.6 per cent of 12–14 year olds reported that males always/often/sometimes sexually harassed females, while 8.1 per cent reported male teachers proposing relationship with female learners in school (Shisana et al., 2010). As we show in section 5, however, not all sexual liaisons between teachers and learners are violent or, indeed, stigmatized within local communities. Furthermore, due to the relatively high repetition rates, particularly in grade 11 (see below), some of the young women still at school are well over the age of consent (16 years in South Africa) and many are already mothers (Barbarin and Richter, 2001).

There is a high degree of fluidity of and contestation over gender/sexual norms. While child and baby rape (understood as a violent attack) are virtually universally condemned (Posel, 2005; Richter et al., 2004), there is much less agreement over what constitutes harassment, rape of an adult woman, and appropriate sexual behaviour more generally (Le Roux et al., 2005). This was amply illustrated in the rape trial of Jacob Zuma in 2005–6 (prior to his becoming President) and the alignment of popular support for and against him (Ratele, 2008; Robins, 2008).

Finally, our project was situated within a feminist politics. This commitment influenced our research questions, methodologies, decision making and policy/practice recommendations. Our understanding of gender equity:

entails putting in place the social and institutional arrangements that would secure . . . the freedoms of all individuals, irrespective of gender or other markers of discrimination, to choose actions, aspirations, and attributes that they have reason to value. (Aikman and Unterhalter, 2005: 3)

The dilemmas

In South African classrooms, it is common for there to be a wide age difference among learners in any grade, partly because the system allows schools to hold back learners who do not pass examinations at the end of the year. In secondary schools, probably because schools are usually judged by their pass rates in Grade 12 (De Souza, 2003), the highest repetition rates are found in Grade 11. Thus, the female repetition rate at Grade 11 increased from 11.4 per cent in 1997 to 18.8 per cent in 2003, and from 12.3 per cent in 1997 to 13.5 per cent in 2004 in Grade 12. In Grade 11, the male repetition rates increased from 12.3 per cent in 1997 to 19.7 per cent in 2003, and from 11.1 per cent in 1997 to 12.7 per cent in 2003 in Grade 12 (Department of Basic Education, 2010: 48). Children from poorer backgrounds often start school later than their middle-class counterparts. Girls who fall pregnant often miss a year or more of schooling (Jewkes et al., 2009).

At the time of our fieldwork, it was common for pregnant young women to be excluded from or stigmatized while at school, despite this being illegal under the South African Schools Act 1996 (Bhana et al., 2008; Masuku, 1998). ⁵ One principal told us that he was not happy to have pregnant learners in his school both because of the ‘example’ they gave to others and in case they needed medical attention. Families of pregnant girls sometimes withdraw them from school, and sometimes the girls themselves, due to stigma and prejudice, decide to leave school during pregnancy. Young mothers may drop out of school for reasons of material need (to earn some money in advance of the birth, for example), because of their own health, or because having a baby connotes adulthood and staying at school connotes childhood (Cunningham and Boult, 1996; Kaufman et al., 2004).

Informal conversations between members of the research team and schoolgirls led to discussions about teenage pregnancy. These included speculation about the paternity of babies born to schoolgirls. Josiah featured in this gossip because of his alleged affairs with pupils. Hearing this gossip placed us in a dilemma. Should we take it seriously and, if we did, what should we do? This was compounded because Josiah was a member of the research team: he assisted us in gaining access to the school and participated in our debates (at the school and the university) about how to promote gender equality at the school, specifically through the school-based intervention that was part of our project. Josiah was a younger, very charismatic teacher. From our conversations at the school we were aware that he was an object of fantasy and desire among some of the learners – as are many excellent male and female teachers around the world (Miller, 1996). We were unable to contact any of the young pregnant women, so had no way of confirming whether any of them identified their (former) teacher(s) as the father(s) of their babies.

We could imagine that this young, charming male teacher would find it very easy to seduce his pupils. Equally, we realized that the very same attributes could lead to his becoming an object of desire within the school’s (hetero)sexual economy of shared fantasy. Josiah may well have flirted with some of his students without anything more untoward happening. As researchers, we had little way of checking these possibilities except by raising the issue directly with him. The truth was that we simply did not know what, if anything other than the purely pedagogic, had happened between Josiah and learners. Nevertheless, given the explicit feminist goals of our project it was important for us to attempt to find out.

Complicating factors: local cultures, material conditions and practices

Sexual or romantic relationships between teachers and their learners are problematic. While learners (girls and boys) have the power to resist and are social actors in their own right, teachers are in a position of structured institutional power and thus their charges may find it difficult to resist/fight off their sexual advances.

Gossip, rumour, scandal and innuendo form an important part of the sexual field in general, and, more specifically, in schools. Epstein and Johnson (1998) argue that they act almost as the cement that holds schools together (and, on occasion, makes them fall apart). Knowing the gossip offers cultural capital and girls’ friendships are often formed around the ability to share rumours about who fancies whom, who is having a relationship, and so on (Hey, 1997).

Sexual liaisons between students and teachers are not necessarily coercive. In the UK context, for example, Sikes (2006) has used life histories (including her own) to explore consensual romantic/sexual relationships between students and teachers, showing that these relationships could be fulfilling and long-lasting. ⁶ There is no reason to suppose that this is different in South Africa. Indeed, the fact that many of the learners are actually in their late teens or early twenties means that such liaisons may often be between consenting adults, even though the learner-teacher relationship remains one of structurally differential power.

Relationships between young women and older men, ‘sugar-daddies’, is widespread and widely accepted in sub-Saharan Africa (Luke, 2003). Leach et al. (2003) confirm that sexual/romantic liaisons between teachers and school students are not uncommon in the three African countries that they investigated. While they conceptualize this as abuse, others, working more ethnographically, place more stress on the agency of young women involved in such relationships. Many young African women living in impoverished circumstances are active agents in sexual relations with older men although not under conditions of their own choosing (Hunter, 2002; Silberschmidt and Rasch, 2001). Such relationships are not forms of sex work (prostitution) but forms of ‘transactional sex’, involving gifts both in cash and kind:

Women typically see multiple-boyfriends as a means of gaining control over their lives, rather than as simply acts of desperation – although the two of course are linked. The very vocabulary of sex, centred for women, around the verb qoma (to choose a man) – is suggestive of women’s agency . . . [Women] see themselves as having some sense of choice, although within often brutal and economically coercive relationships. (Hunter, 2002: 112)

And yet, while girls clearly do have agency, they are caught in a web of relationships and ideals that make it difficult for them safely to navigate risk. Furthermore, some, particularly those who are in resource-poor circumstances or who lack family support, are less able to assert themselves than others (Jewkes and Morrell, 2010).

In poverty-stricken communities throughout sub-Saharan Africa, daughters constitute a form of capital that families may seek to realize through the payment of lobola (bride price). Teachers in such areas, who usually have significantly more disposable income than the families of their learners and rival suitors (e.g. male learners), are thus regarded as highly suitable marriage partners and are considered gendered forms of realizable reproductive value (Guy, 1990). Liaisons between the two may thus be encouraged in the hope that they will lead to marriage or, should the young women become pregnant without marrying their sugar daddies, that the latter will pay appropriate inhlawulo (a fine or damages) for having impregnated them.

What, then, was the status of gossip as evidence in the case of Josiah? The fact of its existence is evidence to be used in developing an understanding of how gossip, rumour and scandal work in the sexual field of South African schools, but not necessarily data about his actual sexual behaviour.

Men and women teachers at all the township schools where we carried out research said that sexual relationships between teachers and learners did happen, ‘but not here’ and when asked if such relationships should be reported the universal response was a definite ‘no’. In a typical response, one teacher said:

What happens here in a primary school, all of the males are married. I’m not saying that they can’t do that, ja, ⁷ but I haven’t experienced that one, but what I can say that one grade 8 learner came to us. I don’t know if it was last year or the other year, she came complaining about the secondary educators. Those, that child was 15 if I’m not mistaken, but she was [mumble]. And the educator was proposing to her by giving her R50, writing love letters with very powerful words. And this learner came to us to rescue her . . . to go and explain to the principal that one of his educators is doing this to the grade 8 learners (our emphasis).

Here, the possibility of inappropriate sexual relationships between staff and students is attributed to the secondary school space, feeding off both the childhood innocence discourse (Bhana, 2007; Epstein et al., 2003) and the ‘not in our school’ one.

Ethical dilemmas: research choices

What should we have done when we heard the gossip? We could have gone straight to the principal and told him of it; and this is, indeed, what many people in UK universities with whom we have discussed the matter said we should have done. But given our uncertainty about the accuracy of what was being said, we felt reluctant to do this immediately. The situation was made even more complex by the need to weigh up the various roles we were playing and interests we were pursuing. Primarily we were in the school in order to conduct research. We were also committed to promoting gender equality and, in this respect, saw little gap between our own agenda and that of the school, which specifically endorsed national gender equality policy. Our ability to conduct our research rested on the cooperation and indeed the support of the school. In developing cooperation and becoming part of the school community we formed ties of friendship and trust. We did not anticipate that our interests would bring us into possible conflict with members of the school community but it became clear that the gender inequality we were researching was being lived and enacted by members of the school community and it became evident that they could be perpetrating gender inequalities and abuse against one another. Even more complicating was the possibility that our definitions of abuse did not match those of the school community.

Our dilemma was heightened because there was never any direct accusation made against particular teachers; none of the learners who talked to us about which teacher(s) was (were) potentially the babies’ fathers said she was actually involved in a relationship with a teacher herself; and no-one wished to report their suspicions to the principal directly. In telling him what was being said in the playground, therefore, we might have been conveying unsubstantiated gossip, especially since we would be unwilling to give the names of those telling us of their suspicions. We had, after all, promised the learners confidentiality and we did not feel that they should be put in the firing line of the intense questioning that would have ensued from our disclosure.

Going to the principal would certainly have breached the trust that the young women had placed in us. We had promised that we would not disclose information given to us in ways that would make them identifiable without prior discussion and then only if we felt that they themselves were endangered or a serious danger to others. Consequently, in cases of disclosure of abuse, we (and others) have discussed with the victim the need to report to a suitable third party and facilitated counselling for her (or him), but always acting with their permission. However, in the case of the young mothers who had left school, there were different layers of disclosure. We were the recipients of gossip and rumour about one or more teachers and some young women, rather than direct disclosure. We did not have the permission of either the young mothers or our informants to share this gossip with anyone else in a context in which they might be identified or have to give evidence in what might turn out to be a quasi-judicial enquiry.

We decided to approach Josiah indirectly. We reasoned that in the climate of the research and intervention project to put the allegations to Josiah directly risked terminally damaging the research relationships that bound the research team together, prejudice our own status within the school and thus jeopardize the entire project. On the other hand, simply to ignore the gossip was to turn our backs on the overarching goal of the project and to just brush aside uncomfortable complexity.

We held a follow-up round of recorded interviews with several teachers in the school, including Josiah. We used these interviews to canvass opinion on a range of issues, including teacher-student sexual liaisons. Robert Morrell was given the task of interviewing Josiah since he was on good and easy terms with him and we felt that a man rather than a woman should raise the issue using a ‘man-to-man’ approach that would not be threatening.

Morrell addressed the issue as vaguely as possible, making no explicit reference to the gossip picked up by the research team:

When Morrell pushed Josiah about the appropriate response to teacher-student relationships, the latter cautioned against taking up the issue formally, laying a complaint, mounting an investigation and so on. He felt that the best way to deal with such relationships was to approach this circumspectly and informally:

Obviously, the fact that the girl has come and reported the matter to me, it means that she does not fall for the idea [of having a sexual liaison with a teacher]. And I think what I would do, I would call up the teacher’s consent and try to bring his attention that first of all, he knows, everybody knows, it’s out of bounds. It’s a dismissible offence if you’ve found you have been in love with a school child. So I will try and bring it to his attention that such a case has been brought to my attention and if need be [he should] call the girl and apologise to her. Because as much as I would want to take the case further, there is this thing of interfering with the life of a person. Because that would mean losing the job to the person, so it’s not an easy matter to say I am going to take it to my superiors. So it’s better then to have a gentlemanly talk, so to speak.

Teachers are all very much aware that such relationships are not permitted and many, but not all, teachers hold that they are improper. Equally, Josiah, like all his colleagues, understood the importance of having a permanent job in a context where at least a quarter of the adult work force was unemployed and many more seriously under-employed. Teachers complained that they are not well paid but a permanent job was like gold and in many cases an employee was supporting many extended family-members.

The notion of ubuntu (human-ness and collective responsibility) came into its own here (Venter, 2004). A ‘gentlemanly talk’ in which the offender was warned to desist from his behaviour was more acceptable to both Josiah and the other teachers than the risk of harm. After all, as the slogan goes in South Africa, ‘An injury to one is an injury to all’ and the potential loss of work was seen as more of an injury than any harm to the young mother.

Furthermore, we owed a duty of care and confidentiality to Josiah, who was also our research participant. Like the learners, all the teachers involved in the project participated willingly and had been immensely helpful in allowing us access to their school, taking time out of exceedingly onerous duties to talk to us, and going to great lengths to give us access to learners. In explaining the purpose of our project to them, we had promised anonymity. The result of reporting rumours to the principal officially would, the teachers believed, have resulted in possible suspension, a very public act that would bring him into disrepute regardless of whether the allegations turned out to be true or not. He could also have lost his job if the hearsay proved founded. Furthermore, though perhaps less important, taking action against a teacher in this way would have compromised the trust that our informants (teachers and learners) had invested in us. Beyond this rationalist assessment of our responsibility to informants was the consideration of the relationships we developed with the teachers we worked with for the duration of the project. In most cases the line between researchers and informants was blurred by the emotional connections that we made and the ties of friendship that developed. Making an ethically correct decision also involved assessing whether friendships were being betrayed. In a different context (nursing) similar ethical dilemmas are present around the choice of caring for a patient and caring for oneself. While care of the patient is a central obligation of medical practice, caring for one’s own health is also important and in some contexts care for the patient actually poses risk for one’s own health (Pfrimmer, 2009). In our case we had to choose between the ‘health’ of our research project and our own relational health status.

We faced one other dilemma and this related to our explicit commitment to gender equity. What would we be saying about our commitment to gender equity if we decided not to follow up the rumours? Would we be condoning an obvious example of gender abuse and becoming complicit in granting legitimacy to a common, gender unequal, practice? Were we ignoring our stated objective of engaging in gender transformation because it was becoming inconvenient and difficult for us to take up the issue more vigorously with our participants? Our research was strongly tied to gender equity interventions and we worried that, in this possibly most critical of cases, we were walking away from that commitment. The question was: to what extent were we justified in giving precedence to our gender activism? Could we override research ethical considerations in the interests of tackling gender inequity or abuse? Which set of ethics took precedence: commitment to anonymity or commitment to gender equity?

As we indicated at the outset, dilemmas are an unavoidable part of undertaking ethnographic research; the most one can do once confronted with a dilemma is to be conscious of the choices that one has and to weigh up the respective merits of the case. In the end, after much angst-ridden discussion, we decided to go the route of not reporting, but rather, of reframing the next stage of the research in order to investigate and raise consciousness about the complexities, difficulties and improprieties of teacher/learner relationships and work with teachers and students to promote gender equity in schools. We also decided that we needed to write this article.

Conclusion: ethical bureaucracies and ethical decisions

Conducting ethnographic research involves making choices. Choices often involve dilemmas that remind us that we cannot be neutral. We were present in school to undertake research as part of our work as academics. We were also there to contribute to a personal, national and global commitment to gender equality. Simultaneously, we made ourselves part of the school community, developing relationships and friendships, building understanding and trust. While this was necessary for the success of our research and aided our goal of promoting gender equality, it also opened up space for us to encounter dilemmas. A school is not homogeneous – its various actors do not sing with one voice; we discovered (or thought we discovered) dissonance that went to the very heart of our work and called on us to think about our ethical commitments. What we have demonstrated is how difficult and complex the processes of decision making were.

Increasingly, research proposals are subject to scrutiny by ethical bureaucracies in universities around the world. The basis of their decision making, structure and remit rests on bio-medical models of research, which have very little to do with social research (Hammersley, 2009). The key issues to which such committees generally attend are those of ‘informed consent’.

Boden et al. (2009) argue that ‘informed consent’ is a chimera, designed primarily to protect research organizations (such as universities or pharmaceutical companies). Issues arising in research, whether medical experiments or qualitative sociology, cannot be fully or accurately predicted – otherwise there would be no point in the research.

In our research we did not foresee that some of the very teachers who were most active in promoting gender equity in their schools might be implicated (albeit through rumour) in the pregnancies of their students. Knowing that there were teachers who engage in sexual relationships with learners did not mean that we expected to meet them in our particular research setting. Somehow we too bought into the idea ‘not in our school’. We personally knew the staff and were familiar with the schools and somehow the research that suggested widespread teacher-student relations did not seem relevant, despite research suggesting high levels of teacher-student sexual interaction (Jewkes et al., 2002). While we had a blind spot in this regard, so too did the teachers we worked with. There was no intimation from them of anything untoward in this regard. If we had predicted and sought to pre-empt this particular outcome, we would likely have lost their trust and, consequently, access to their schools.

In tackling questions of harm, the judgements to be made are not clear cut. Many local community members and teachers said that reporting alleged misconduct would be the most harmful course of action we could have taken. While not wishing to fall into complete relativism, or abrogate our own duty to make moral judgements, we found this particular call very difficult. Was it more ethical to adhere to the collectivism of ubuntu as expressed by the teachers we interviewed and not report? Or would it have been more ethical to ignore these local views and report our suspicions? We are not convinced that we made the right choices, but neither would we accept that we made the wrong ones. Because ethical dilemmas always call for judgement, there is always the possibility of getting it wrong. But committees and bureaucracies, by their nature, look for definite answers. Things are either ethical or unethical. We would suggest, in contrast, that there are many shades of what one might call ‘ethicality’ between these two binary opposites and that ethics committees are not in a position to judge them.