The impossibility of anonymity in ethnographic research

Introduction

I recently attended an Ethnography and Education conference where one of the participants presented a paper that was centrally concerned with the issue of anonymity in research. The study dealt with a group of participants that was so small that it was impossible to disguise any of the leading individuals in the research sites, and the important disputes between different sub-groups could not be discussed at all due to this anonymity problem. The author thus felt unable to publish. This conference paper and the resulting discussion encouraged me to think again about the practice of anonymity in ethnographic research – it was well over a decade since I had written on the topic (Walford, 2005) and much had changed since that time.

By chance the conference coincided with the tenth anniversary of the introduction of the iPhone. The ever-present smart phone has changed the way most people live their lives more than any other 21st century invention so far. It is always with us - with its instantly available camera, microphone, and recording devices, and the associated ability to share photographs and videos with friends, enemies, and millions of other with whom we have no relationship at all. Most of us now live our lives in public through Facebook, Twitter, and other social media. Our lives are recorded on millions of closed circuit cameras, and our careless un-thought-through words can be shared with most of the planet. Geolocational data is constantly collected about us such that it is possible to trace our physical movements. Much has changed in the years since I first considered the problem of anonymity in research.

As part of this wider change, schools, colleges, universities, and other less formal learning sites are no longer isolated from the wider world. Indeed, many schools now base their teaching around one-to-one digital technologies and encourage students to bring their own iPads and other devices into the learning environment. Recent ethnographic studies may question the extent of the transformation of teaching through digital technology in the classroom (Selwyn et al., 2017) and report variations in the extent to which social, digital, and learning networks enable or disempower young people (Livingstone and Sefton-Green, 2016), but it is certainly clear that the digital social connections to which young people, parents, and teachers now have access dramatically change the chance of research sites and people within them remaining anonymous.

The assumption of anonymity

A simple search of research ethical guidelines makes it clear that the default assumption for all educational research is that the names of sites and individuals involved in the research should be anonymous. For example, the most recent BERA Ethical Guidelines (2011) state:

25. The confidential and anonymous treatment of participants’ data is considered the norm for the conduct of research. Researchers must recognise the participants’ entitlement to privacy and must accord them their rights to confidentiality and anonymity, unless they or their guardians or responsible others, specifically and willingly waive that right. In such circumstances it is in the researchers’ interests to have such a waiver in writing. […]

Similarly, the AERA Code of Ethics (2011) state:

12.06 Anonymity of Sources

(a) Education researchers do not disclose in their writings, lectures, websites, or other public media confidential, personally identifiable information concerning their research participants, students, individual or organizational clients, or other recipients of their service which is obtained during the course of their work, unless consent from individuals or their legally authorized representatives has been obtained.

(b) When confidential information is used in scientific, scholarly, and professional presentations, education researchers disguise the identity of research participants, students, individual or organizational clients, or other recipients of their service.

Both of these guidelines are designed for all types educational research, but they are usually interpreted as being appropriate for ethnographic research and institutional ethics committees often base their own codes of conduct for research on these or similar guidelines and codes of ethics devised by the professional associations. The situation is, of course, more complicated for ethnography than for most other forms of educational research, for the details of what is done in an ethnography unfold as the research proceeds. It is not possible to give full ‘informed consent’ to any institutional gatekeepers or to individuals involved simply because the ethnographer can never know what might be found or how the focus of the research may change over the fieldwork period. Ethnography also deals with groups of people rather than a selection of independent individuals as in many surveys or psychological experiments, which means that those in the group know who else is involved and what might have been observed by the ethnographer. However, the idea that anonymity should be the default option has been a widely accepted part of ethnographic work.

While the very nature of ethnography makes it impossible to give complete anonymity to participants as they all know who the other participants are and (nearly always) know that the researcher is conducting research, it was previously relatively easy to adhere to the spirit of the idea of anonymity of the site. When Stephen Ball published Beachside Comprehensive in 1981 two reporters tried to locate and name the school for a local radio and a national newspaper report. They judged that, as Ball was located in Sussex, telephoning comprehensive schools on the South-East coast and asking whether the school was Beachside, would cause one of them to admit that it was, indeed, the research school. Unfortunately for the reporters, the word ‘Beachside’ was a red herring as the school was not located near any beach. The reporters did not locate the school (Ball, 1984).

Far from all ethnographic research is perceived as worth news reporters’ notice. In my own research involving two major boys’ private boarding schools, at an official level, the only person in the first school who knew the identity of the second school was the headmaster as he had been contacted to provide a ‘reference’ for me by telephone by the headmaster of the second school. Many teachers at the second school asked me which other school I had researched, but I never told them, arguing that I would also not say that I had researched their own school once I had completed the research. Both schools were part of either the Rugby or Eton Groups of schools which then numbered 29 schools in total, but I had taken care that both were ‘minor’ major schools – I did not attempt to get into Eton or Winchester as their high profile meant that the chance of maintaining anonymity would have been much reduced. Of course, some leakage of information between the two research schools may have occurred as there were regular meetings and some interchange of staff within the groups but, to my knowledge, the names of the schools have never been made public – something that I was particularly concerned about as some material published was critical and some might have been traceable to the few female teachers involved in the schools. Looking back, I am surprised that it seems that no attempts were made to identify my two schools. I had written that they were both part of a relatively small group of schools, and it would not have taken long for someone to telephone them all in the same way as Stephen Ball’s reporter. It seems that the content of the book was not sufficiently controversial or potentially damaging for anyone to bother.

The coming of the internet meant that names of schools and other research sites can often be identified through a Google search. Even back in 2005 I wrote about the possibility of identifying schools where researchers had included part of an Ofsted report in their general description of the school (Vulliamy, 2004), and I was able to make a very good guess at Peshkin’s (2001) elite research school from the information given in his book. I then entered that school’s on-line library catalogue to find that the librarian had helpfully explicitly identified the school as being the site where the research was conducted. This simple difficulty in maintaining anonymity still seems to be under-recognised. For example, in her new ethnography of an English Academy Christy Kulz (2017:18) states that ‘the name of the school, borough and all participants are pseudonyms in order to protect their anonymity’, yet I was able to identify the name of the school simply by selecting just three words from the blurb and putting them into Google!

In the interconnected world in which we now live, there is practically no chance of maintaining anonymity of research sites. At the conference that re-kindled my interest in anonymity a participant reported that at one point the leader of the group she was researching had used Periscope to show the class live – including pictures of her observing. Now schools welcome guests on notice boards at the school’s entrance and may put names of guests on their website. Students take photographs and post them on Facebook; teachers inform their students of events through email; Twitter discussions occur between students and others; any of the participants may write about their experiences on a public blog. On publication of the research, an open request from a reporter for information on the identity of the site would probably bring many replies from disgruntled students or teachers – the headteacher might even wish to break the anonymity if he or she sees the results as positive. Amazon now provides books to academics and the public across the world with reader reviews being easy to post by anyone. Just one person wishing to be identified with the book can break any claims to anonymity given by the researcher.

It is somewhat easier to give some form of anonymity to the individual participants than to the site in the published results of the study. Unless the focus of the research is on a very small number of students, it is usually possible to separate verbal comments or observations of behaviour from identifiable people. Even if we dissemble our descriptions of individuals such that gender, class and ethnicity may be changed, these involved may still be able to identify themselves or others. There is thus still the chance of harm through on-line or other bullying, but most other readers will not be able to identify them. For teachers, on the other hand, it has always been almost impossible to give them true anonymity. The headteacher will know which teachers took part in the research and, if they are still in the school on publication, there may be negative consequences. Perhaps one of the worst cases where anonymity was broken for those within the school community itself is Woods’ (1979: 217ff) description of the headteacher in his research school who is made a figure of ridicule. The discussion is within the context of staffroom humour, but it is unlikely that any of the parties involved were content with the published account. I believe that Woods regretted this description, for his many subsequent ethnographies are much more positive towards teachers. However, if the book were published now the headteacher would surely be identified externally as well as internally.

A move towards openness

I have argued that modern interconnected life does not allow ethnographers honestly to give anonymity to research sites or individuals within them. In the past, the form of anonymity offered has always been limited as there are going to be many who are involved in the research and who may choose to break anonymity on purpose or by accident. Most importantly, the people who have direct power over those being researched have usually been able to break any claims to anonymity that a researcher might give. It is not accidental that some have seen ethnographers as spies, and claims to give anonymity have often neglected the fact that those who might do most potential harm to the teachers or students involved are exactly the people who are most easily able to break any attempts at anonymity.

However, it is worth remembering that ethnographic researchers generally do not seek to find incriminating evidence against individual people – their objectives are concerned to uncover the cultures of schooling and learning. In a less-interconnected world, there were reasonable expectations that limited anonymity could be maintained. This is no longer true and we have to move to openness.

There are both disadvantages and advantages in openness. From a researcher’s viewpoint, the main disadvantage in naming research sites (but not necessarily the participants) is frequently seen as that of access. The claim that it is becoming more difficult to obtain access to schools and other places of learning is not new (Troman, 1996), and promising anonymity is widely believed to increase the chance of gaining access. Perhaps there is some truth in this, or perhaps we do not recognise that it is likely to get harder to obtain access once we are older and in positions of more influence. Or perhaps we do not work hard enough to obtain access. It is worth noting that radio, television, and other media still manage regularly to produce documentaries about schools and other educational institutions. Here not only are the schools named, but cameras allow the wider world to see interactions within the classrooms, staffrooms, and corridors, and usually with very little control over the final product.

I have argued elsewhere (Walford, 2008) that researchers need to see the access process as a sales campaign where it is vital that information is gained about those who have the power to grant access before any approach is made. We need to conduct preliminary research on the targeted gatekeepers and the institution, and see the process as one of selling ourselves and our research to those who have power to grant access. In sales jargon, we need to work through the classic four-fold process of researching who to approach and how, interesting the gatekeepers in the project, building a desire to be involved, and finally selling the project to them. We need to see access as a process where we are continually negotiating within the research and affirming the benefits that the research offers. We need to recognise that the institutions pay a cost to admit us, that we have something to offer in return, and that we need be prepared to strike bargains.

Research access has never been easy. In the case of my work on the first City Technology College (Walford and Miller, 1991) the site had to be named as it was opened two years before any other similar schools. These were a new type of school which later became a model for Academies and thus highly controversial and political. I was responsible for the ethnographic part of the book and the process of gaining access was messy and one that I would not recommend to others (Walford, 1991). The final bargain that I negotiated was that I would have complete control over most of the planned book, and the Principal would have up to 15,000 words at the end where she could put forward her own views. Eventually, she did not do write anything for the book, presumably because she was broadly happy that she could trust me to write something appropriate, and her pressures of work were too great. I rather regretted this, as I think that the book would have had greater impact if her views had been included. Of note is that I would never have negotiated away my right to write what I believed to be true. I was happy for the Principal to read it before it was published and to make suggestions that I might or might not follow, but she agreed that she should not have any control over what I finally wrote as the main report of the study had to be seen as independent of the school if it was to have any value.

Reich (2015) has argued that the rise of online communications and social networking has made research access even more difficult as potential participants can now easily check-up on the researcher before agreeing to take part. Further, she argues that there has been a shift in power and reduction in boundaries between the researcher and the researched such that there are changes to the way ethnographers recruit participants, generate data, leave the field and disseminate results. Participants are easily able to maintain contact with the research and researcher before and after publication, and make their own views know to a wide audience. While she does not propose abandoning attempts to anonymize sites and people, she does see some advantages and the necessity of acknowledging the pervasiveness of social media in research processes.

Advantages of openness

While there may be added difficulties if we do not offer anonymity, there are several advantages as well. The first of these is that researchers are forced to think more carefully about their choice of sites. My view is that research sites should be chosen for a theoretically appropriate reason (Walford, 2001; 2008) but, in truth, research sites are very often chosen more for convenience than for any theoretical reason related to the aims of the research. This can be particularly true in comparative and international research (Walford, 2001) where it is often the case that educational institutions drawn from two different countries are compared with little justification for the choice of countries apart from the fact that the authors happen to live there. Notice that it would be thought ridiculous to try to conceal the names of the countries, but usually there are attempts to give anonymity to the individual institutions and local geographical regions involved. It is very rare for any theoretical reason being given for national or school choices. Convenience of location is clearly important if a long period of time is to be spent at the site, but making convenience the sole criterion challenges the quality of what we can say about the findings, and also makes it more likely that the research site can be located and named if pseudonyms are used. In contrast, if we select our research sites with some theory in mind, this may sometimes make access easier. A school that is selected because it is seen as leading edge in some curriculum area or educational practice, might be more willing to be part of a study and to have its name made public.

Openness could lead to higher quality research and writing in several ways. Because ethnographers attempt to offer some degree of anonymity, they are forced to use vague descriptions of the local area and describe the social and ethnic mix of the institution in very broad terms. Thus research schools might be described as being ‘in a northern industrial city’, ‘in London’, or ‘in the south-east of England’. But the exact geographical location often matters, and the nature of events occurring within a school cannot be fully understood without knowledge of the local environment in terms of social class, ethnicity, age profile, and similar variables. Naming research sites would enable researchers to choose research sites for specific theoretical purposes, and to discuss the nature of the schools and its location fully, and thus improve the quality of ethnographic analysis and writing.

Naming schools would also potentially improve the quality of the relationship between claims and evidence in ethnographic writing. One side-effect of using pseudonyms is that ethnographers can be less careful about their analysis of data and what they write. Where they ‘feel’ they are correct, they are able to base arguments on evidence that may not be as strong as one might wish. Naming schools and possibly thus identifying leading individuals in the site, means that ethnographers have to be much more careful about the claims that they make. At the extreme, they might be sued for libel where sites are named – something that is very much more difficult to do if pseudonyms are used.

Naming schools would also help to reduce the fundamental problem of ‘generalisation from a sample of one’ (Walford, 2007). Unnamed sites allow a false sense of generalizability of ethnographic research both on the part of authors and readers. The critical reader may have noticed that in the discussion above I referenced Livingstone and Sefton-Green (2016) and Selwyn et al. (2017) to bolster my claim of the pervasiveness of digital technology in the classroom. I gave no idea of the nature of the schools under study or even which country they are in, and tried to lead the reader to believe that my claims were supported by research. Both studies are excellent ethnographies, but the first is based in ‘an ordinary urban secondary school in London’ (2016: 3) with no indication as to what this can possibly mean or why it should be a suitable research site for the study. The second also claims that the three schools involved ‘might be considered ordinary (rather than exceptional) sites of technology use’ (2017: 290) and they were chosen to provide contrasting institutional contexts. All three schools are in Victoria, Australia, and some information is given about their size, nature of intake, local area, and so on, but there is no justification for the actual choice of schools. In both of these ethnographies the authors take care to limit the generalisations that they make from their small samples of one and three respectively, and acknowledge the limitations of single sites, but the conclusions inevitably stray beyond the particular to the general and reader is seduced into believing that what they find is typical of other ‘ordinary’ schools.

Jan Nespor (2000) takes this further and argues that anonymization is a semi-deliberate representational strategy which has both ontological and political implications. He argues that anonymization normalises the decoupling of events from their historically and geographically specific locations, such that the results are given a spurious generalizabilty. The school can be seen as just a more ‘general’ place – a school that ‘could be like any other school’ – an ‘ordinary’ school – one just like many that we know. The use of anonymity thus invites readers to see the findings as being applicable to other situations. While most academic readers are able to intellectually accept the lack of generalizability, we are seduced by the absence of specific details about the site and situation such that the findings from one study expand to fill our general understandings of the issues discussed.

My early book on private schooling was called Life in Public Schools (1986) and reported the study of two schools whilst implying that there was generalisability beyond the two studies to other ‘leading’ private schools. I soon saw this as a mistake, and a later study was deliberately called ‘City Technology College’ (1991) in the singular. Although the book was based on a single named school, the publisher’s marketing team wanted it to be ‘Colleges’, and I had to fight to retain the singular noun. By that time I was aware of the problems of generalisation from a sample of one and refused to heed their claims that a plural noun would lead to greater sales. They were probably correct. Paul Willis’ (1977) famous Learning to Labour would probably have received much less attention had the sub-title been ‘How twelve selected non-academic working-class boys get working-class jobs’.

Conclusions

I have argued that there are advantages to more openness in naming our research sites. If sites and some of the people centrally involved are to be identifiable, then ethnographers need to develop new ways of working that have to be more collaborative. Ethnographers will have to come to agreements with those involved in the site so that their voices can be heard as well as those of the researcher. Here we can take advantage of new communications technology, not necessarily within the academic articles and books themselves, but possibly in the form of linked on-line materials or forums which are referenced in the publications. It is to be hoped that a more collaborative mode of working would include the sharing of publications before actually publication, such that any factual errors and misrepresentations could be eliminated, but we cannot expect complete agreement between all involved. Giving those involved specific and linked spaces on social media might allow disagreements to be acknowledged.

The section of the BERA Ethical Guidelines (2011) quoted at the start of this article continues with the words:

Conversely, researchers must also recognise participants’ rights to be identified with any publication of their original work or other inputs, if they so wish. In some contexts it will be the expectation of participants to be so identified.

This addition was designed to deal with cases of action research and close collaboration between teachers and researchers in implementing and monitoring change. It was not designed for ethnography, but it does underline the fact that others may have made distinct contributions to our work (Perry, 2007). A recent ethnography of capoeira teaching and learning in the UK by Sara Delamont, Neil Stephens and Claudio Campos (2017) has a Brazilian mestre in capoeira as the third author who was both one of the subjects in the study and who provided a great deal of historical and background information. Such a situation is uncommon in ethnography as it is the researcher who has invested a great deal of time and systematic work in the research, and who thus usually becomes more knowledgeable than any of the participants in the specific topic under study. It is a partnership, but usually not an equal partnership, for the ethnographer brings a variety of research and writing skills, as well as knowledge of other studies and theoretical frameworks that the other participants do not have. The books and publications would not be produced without the researcher’s work, so it is important that any agreement made between the parties involved makes it clear that the ethnographer’s voice is given precedence over those of anyone else. Others cannot be allowed to silence the findings of long-term systematic research just because they do not like the findings, but it may be that they also need to be given a space where they can voice their views. Digital media now allow this possibility by providing a site where all those involved in the research are able to debate the issues raised.

Any agreement also has to recognise that the researcher has responsibilities and obligations beyond those to the participants which extend to any external sponsors and to the wider community. Others have a right to know about the results of systematic research even if those involved would prefer it otherwise. Any agreement must recognize that there are sometimes competing demands between the many parties involved in the research process - not just the researcher and the researched. It is likely that the agreement will also clarify that the researcher will have the leading role in seeking out potential publishers and guiding submission to academic journals - it is a partnership where all involved have their say, but it is not an equal partnership. In the end, the researcher is offering the possibility of publication to someone who would otherwise not publish in this way.

This will not be easy, but it is no longer possible to pretend that promised anonymity within ethnographic research can act to protect sites and participants from any repercussions from the research. We are now all connected with one another and the quality of research and discussion would benefit if we develop our ways of working to deal with the problems and opportunities that this interconnectedness brings.