‘Learning to labour’ and the labour of learning: a question of research methods

ABSTRACT

It is now half a century since Joey explained to Paul Willis: ‘Vandalising […] that’s the opposite of boredom – excitement, defying the law’, one of many similar comments subsequently recorded in Learning to Labour (34). The book rapidly became an academic best-seller, and has since become an academic ‘national treasure’. But, before a round of celebratory articles and book chapters which can be expected to mark the 50th anniversary, this article gives a critique of the research methods used and argues that the influence of the book has been detrimental to the quality and rigour of the research methods demonstrated in many current ethnographic studies.

KEYWORDS:

• Paul Willis
• learning to labour
• research methods
• ethnography
• rigour

Introduction

Mills and Morton (2013, 26) state that: ‘For many ethnographers of education, Paul Willis’s Learning to Labour is the defining modern classic, and remains the most discussed and cited, partly because of Willis’s sophisticated theoretical discussion of the relationship between education and social-class formation.’ They were probably correct in their assessment in 2013 and it might be even more appropriate today as the book has now been translated into German, Swedish, Finish, Japanese, Spanish, Portuguese, Korean, and, as recently as 2013, into Chinese. The reception of the book in China is discussed by Moskowitz, She, and Chunwen (2018). Meanwhile, an American edition was first published in 1981, and the UK edition has been continually reprinted by Saxon House, Ashgate, Gower, and Routledge as smaller publishers have been incorporated within larger ones over the years. Extracts have regularly appeared in Readers and an early chapter (Willis 1976a) was commissioned specifically for a Reader for a new Open University course on Schooling and Society and published in 1976 a year before Learning to Labour was published. Along with this, has been an industry of academic papers both praising and criticising aspects of the work as well as developing the ideas and extending the work to new areas (e.g. McGrew 2011; Nolan 2018; Trondman 2018; Walker 1986). Various authors criticised it for what was seen as its glorification of laddish culture, its lack of consideration of ethnicity, its sexism, and so on. Indeed, the amount of attention to it given by critics is probably one of the reasons it became recognised as a classic study. A celebratory edited volume was published by Dolby and Dimitriadis in 2004 to record the 25th anniversary of the book’s publication in 2002, while the journal Ethnography, co-founded by Willis in 2000, published a Special Issue in 2018 to celebrate the 40th anniversary (Trondman and Lund 2018; Willis 2018). We can confidently predict a further outpouring in 2027.

There can be little doubt that Learning to Labour has ‘classic’ status within the ethnography of education and that, as Mills and Morton state, this is mainly due to its contribution to theories of resistance and reproduction within the field (Willis 2018). There have been several challenges to these theoretical claims but in this article, I wish to focus on the research methods used by Willis and argue that Learning to Labour has been damaging to the development of educational ethnography, for others have followed his ways of working. I recognise that all research involves the constant interaction between theory, method, data, and critique and that much could be said about all four in relation to Learning to Labour but my focus here is on methods used. I shall consider issues relating to site selection, generalisation, selection of informants, lack of consideration of historical context, over-reliance on interviews, and more. In the sections that follow I shall, first, give an outline of Willis’ career and early research work, and give some historic context. Subsequent sub sections will focus on the limitations and problems with the methods used in Learning to Labour.

Paul Willis’ early career

While Learning to Labour was Willis’ first book, it does not report his first ethnographic research. His less well-known book, Profane Culture, was published in 1979 but was based on his doctoral work conducted at the Center for Contemporary Cultural Studies (CCCS) at the University of Birmingham. This book was published by the far more prestigious publisher of Routledge & Kegan Paul who probably took a far greater time to move from manuscript to publication than did Saxon House, the initial publisher of Learning to Labour in 1977.

To understand the importance and limitations of Paul Willis’ work it is necessary to see it in the context of his personal career and the social, political, and economic situation at the time. Willis was born in 1945, attended Wolverhampton Grammar School then, in 1963 when he achieved higher grades in his A-levels than expected, he was given a late place at Peterhouse College, Cambridge, to read English Literature. The Cambridge course did not go well and he gained a 2 (ii) in 1966 (although there were far fewer 1st and 2(i)s at the time). However, postgraduate courses in Business Studies at Manchester Business School, and in Industrial Relations at the LSE followed, both supported by grants (Mills and Gibb 2000, 202–203). In 1968 he started a part-time PhD in the interdisciplinary Centre for contemporary Cultural Studies (CCCS), Birmingham, initially supervised by Richard Hoggart and then Stuart Hall, researching ‘youth culture, music and how young people live now’ (Mills and Gibb 2000, 203). As Connell and Hilton (2015, 288) state, CCCS has always had an acute sense of its own history and Willis has been interviewed and published about his research several times (e.g. Bracke, Davidson, and Guadeloupe 2021; Mills and Gibb 2000). While, as with all interviews, these accounts must be interpreted with care (see later), they offer information and insights about Willis’ ways of working that are not available elsewhere.

CCCS was an interdisciplinary centre started in 1964 as a result of Richard Hoggart’s appointment as Professor of English and his inaugural lecture which set out the rationale for a research programme into Literature and Contemporary Cultural Studies. Funding, of £2400 per year from 1964 to 1975, came from Penguin Books which initially paid for Stuart Hall to become the first Research Fellow (Connell and Hilton 2015, 293). Connell and Hilton (2015, 293) claim that by 1973 there were 39 students attached to the CCCS, so Willis’ doctoral period marked a rapid growth in student numbers.

This part-time doctorate was self-funded and Willis was able at that time to find a number of part-time academic jobs to support himself including a part-time lectureship at Aston University. Just after Willis started, some staff and students of the Centre became centrally involved in a sit-in during November 1968. This political protest and the departure of Hoggart to UNESCO were crucial to plans to democratise working practices (Connell and Hilton 2015, 296). Stuart Hall became effective director in 1969, but decisions were made by the collective at a weekly Centre General Meeting. Students became involved with admissions of new students as well as many other aspects of the work of the Centre. Willis states that, while he was involved with the protest, as he was a part-time, unfunded student he was not at the Centre ‘anything like 100% of the time’ and was there only when he wanted to be and to attend the compulsory ‘so-called Theory Seminar’(Mills and Gibb 2000, 204–207). Remarkably, even with a marriage and child born during the period, change of supervisor, and many disagreements within the Centre, the PhD was awarded after only four years in 1972 (Willis 1972). It is this work that led to Profane Culture (1978, 2014) and related papers (Willis 1976b). The disagreements within the centre might be illustrated by the fact that this chapter by Willis’ on drug use is directly followed in Resistance Through Ritual by another by Pearson and Twohig (1976) that strongly challenges it.

The somewhat chaotic yet politically exciting changes within CCCS, coupled with the fact that ethnography played only a minor part in the Centre’s work, and the general lack of systematic doctoral research training at the time, meant that Willis received no research methods training (Mills and Gibb 2000, 218). Instead, as he states:

I took the close-reading technique, which I’d learned at Cambridge and switched from the poem to lived forms of culture, youth culture, or, in fact, context, asking the same kinds questions: Where does meaning come from? How do the symbols work? What would happen if you altered the arrangement or sequence of symbols? What was the meaning embedded in discourses like the poem, but carried in lived culture and often without texts but which could be music or the bike itself [?]’. (Bracke, Davidson, and Guadeloppe 2021, 8)

Profane Culture presents the results of separate ethnographic studies with two very different groups of young people who he calls ‘the motor-bike boys’ and ‘the hippies’. The first group was located through contact with an unnamed motor-bike club linked to a church in an unnamed large English city. He conducted field work and interviews over a nine month period with a loose and variably numbered group which centred round one person and a core of five people with another two who were often present. The second group (the hippies) was contacted through attendance at a public house in a large unnamed industrial city. Over a period of five months Willis spent time with three different small groups of three young people and interviewed them in their homes. Both of these studies were thus small scale – which is exactly what would be expected for a doctorate – and the book documents elements of the living and creative cultures of these two small groups of young people.

An excellent example of Willis’ use of ‘close reading’ of symbolic forms can be found in his description of the cultural role and meaning of the motor-bike itself (Willis 1978, 52–61), where he describes the physical experience of high-speed riding and the centrality of the motor-bike in the culture by questioning the physical object and experience as one might examine a poetic text. This is a good example of what Willis (2000) was later to call ‘ethnographic imagination’.

But the book claims far more than is warranted. It falls into a common trap of generalising way beyond the ethnographic data can support (Walford 2007). The introduction to the book states (Willis 1978, 1):

This book presents two important cultures generated during the 1960s and still widely influential today - the motor-bike boys, sometimes known as rockers, and the hippies, sometimes known as heads or freaks. The form of the book is of two ethnographic accounts of the inner meanings, style and movement of these cultures, but the essential theme of the book is that oppressed, subordinate or minority groups can have a hand in the construction of their own vibrant cultures and are not merely dupes: the fall guys in a social system stacked overwhelmingly against them and dominated by capitalist media and commercial provision.

The book contains the hallmarks of Willis’ work, in that claims are initially made on data drawn data from very limited case studies, and these claims are then amplified into claims about entire sub-cultures. The selection of participants is opportunistic and there is little attempt to test theories developed from these ethnographic case studies against other forms of data or other groups. Within Profane Culture there is also a characteristic over-reliance on interviews with little consideration of the situation in which the data were constructed or the author’s own influence on that production process. It also shows underplaying of any numerical data both within descriptions of the methods used and within the ethnographies themselves.

In many ways, it is unfair to make such criticisms of the results of a doctorate undertaken at that time. A doctorate is always a learning process and a way of coming to an understanding of the practice of academic labour. Willis was labouring to learn about academic research and writing in a situation where most of the support given by CCCS was about theory rather than methods of research. While Willis has long argued the necessity for getting into the field and understanding what is going on through detailed ethnography rather than statistical surveys, this was a significant challenge to orthodoxy during the late 1960s and early 1970s. Developing a broad theory about the way that groups generated their own cultures was used in a process of justifying conducting case-study ethnographic work.

Learning to labour

The development of educational ethnography in the UK is often presented as drawing from anthropology of education and sociology (e.g. Delamont and Atkinson 1995; Mills and Morton 2013; Walford 2008). But Willis was always somewhat outside this framework, and developed his own way of doing cultural studies of education within a humanities-focused research centre with, at that time, few links to developments within sociology or education or anthropology. Rigourous research methods of selecting particular sites and people, generating and analysing data, and making this process clear to readers were seen as unnecessary and part of a positivistic research agenda which was to be avoided.

In many ways, it is again thus profoundly unfair to criticise Learning to Labour for its methods. The book certainly deserves its reputation as a classic, however, it is because of this classic reputation and the fact that it is still so widely read and cited that it is important to study its limitations in research methods as an example of ethnography. Looking back at the book nearly 50 years on, critique is not about what was done ‘wrong’ or ‘correctly’, but about how a similar study might now be conducted given what is broadly now accept as good ethnographic practice (see, for example, Walford 2009, 2018a). It is worth doing this with Learning to Labour simply because of its classic designation. Many other books and papers of the period exhibit similar elements that a present-day ethnographer might wish to critique.

Lack of full descriptions and ambiguities of the research methods

In the 1970s, publishers had a great deal power within academic publishing. In this case, Willis states that ‘I am strictly forbidden, due to shortage of space, to discuss methodology and its relation to theory and practice at any length’ (Willis 1977, 194). His publisher probably thought, with other publishers, that methodology was too abstract and dull to be included in a book and urged writers to cut such ‘details’. The book thus does not provide a full systematic account of the methods used, and readers are referred to other general references and a now-unobtainable stencilled final report to the SSRC for some more information (The Main Reality, SSRC Final Report). The publishers were also probably responsible for the blurb which states that the research was conducted in a ‘tough midlands comprehensive school’ when, according to Willis (1977, 4), it was a ‘boys only, non-selective secondary modern school’ which had a ‘good’ reputation. The difference is important but Willis’ description, which must be assumed to be correct, does not aid understanding. It was non-selective, but only in the sense that boys who had passed the 11+ examination and were entitled to attend a grammar schools, could choose (or their parents could choose) to attend the secondary modern instead. From other information given, the school was part of a wider selective system where grammar schools selected an unknown proportion which might range from 10 to 30% of the most academically able children. The research school was thus designed for boys who had already experienced academic failure in their 11+ examination. The lack of clarity, and desire to generalise, even fooled Aronowitz (1981, ix) who wrote an introduction to the USA edition (which changed the title (only) to Learning to Labor) where he states that the research was conducted in an ‘English comprehensive school’ and then treats the findings as automatically applicable to US High Schools.

Learning to Labour was based on a research project supported by the UK government-funded Social Science Research Council. In order to obtain such funding it was necessary for someone to apply and set out the nature of the planned research, how it was to be conducted, and why it was worth doing. Such research applications were likely to propose over-ambitious plans in order to obtain funding and the research project that funded Willis was no exception. What was planned was extensive, in part no doubt, to deal with possible objections to funding of small scale case studies and to introduce some more ‘positivistic’ comparison between groups. While the book is largely based on an ethnographic study of ‘a group of twelve non-academic working-class lads’ (Willis 1977, 4) from a single boys’ secondary modern school that served an inner-city, working-class housing estate, the wider study actually included ‘comparative case studies’ of five further groups of boys (4–5) These were a group of conformist boys in the same school, a group of working-class conformist boys in a nearby mixed-sex secondary modern school, another group of non-conformist lads the local single-sex grammar school, another group in a comprehensive in the larger conurbation, and further mixed-class male non-conformist group in a high status grammar school in the same wider conurbation. The plan was for a multi-sited case study, but the details of the nature and extent of actual involvement are unstated. Assuming 12 in each group, even a single interview with each of the boys would have required 60 interviews (probably enough for many complete research projects at that time), but the results from these comparative case studies feature only lightly in the book.

All of this information can be found within the early pages of Learning to Labour, but no justification is given for the existence of these comparative studies or of the selection of particular schools. No information is given on how the samples of informants were selected in these schools, or on the nature and extent of Willis’ involvement with any of them.

Details of how the main group of 12 boys was selected and who did the selection are also missing. We are told that the selection was done on friendship links, but it is unlikely that it was a single pre-existing group. The extent of involvement with the main group of 12 lads is also lacking, but we are told that involvement included ‘observation and participant observation; informal interviews and diaries’ (5), attendance at lessons, ‘taped long conversations with all the parents of the main group, and with all senior masters of the school, main junior teachers in contact with the group, and with careers officers coming into the school’ (5). All this occurred over five terms (or perhaps four, see below) in the students’ penultimate and ultimate years in school, and then all twelve (plus three others from the comparative schools) were followed into the first months of their working lives for short periods of participant observation and taped interviews with the individual, and with selected interviews with foremen, managers and shop stewards. No reason is given for the unusually long period of engagement and nothing is made in the description or analysis of any differences between data drawn from the penultimate and final years in school. One would expect considerable changes over this period, but these are not discussed systematically.

It is actually unclear whether the research was conducted over a three or four year period in 1972–1975. (In Willis (1976a) an exact period for the SSRC funding is stated as April 1973 to June 1975 where it is also stated that the 12 lads were followed for their final four terms rather than five; in Willis (1979) the period is stated as 1973–1975.) It was most probably done over three years but, even if it was four, the amount of work to be done by a single researcher was phenomenal. Just arranging to visit this number of schools and places of work, fixing times for interviews, and even listening (where it was taped) to all the data generated (there is no indication that much was transcribed) would have been a very full-time job. Yet, within this period there was also the need to analyse the data generated, read the academic work of others, write, and simply think.

The lack of detail on the number and nature of interactions and interviews conducted with ‘the lads’ is important for it denies the reader sufficient information to be able to assess the validity of the data. We do not know how many visits Willis made to the school – it might have been four days each week, once a week, or once a month. We have little idea of the amount of data that was generated or how these data were analysed. Indeed, we do not know if these data were actually analysed systematically or quotations, anecdotes, and observations simply selected to support a pre-existing or developing theory. We do not know if the examples selected for publication were unusual, extreme cases, or examples of regular occurrences.

In a more recent ethnography, readers would certainly expect to find much more information on all of these issues. While it is difficult to include everything desirable in an academic article where space is limited, there is little reason why book-length ethnographic studies should not include a clear account of what was done and why certain methodological decisions were made. Willis is certainly not alone in lacking full information, but it means that Learning to Labour does not provide an ideal example for present-day ethnographers to follow.

Problems of generalisation and site/sample selection

All research is conducted within a particular social, economic, and political environment. The time when Willis was researching for Learning to Labour was one of rapid change and instability within schooling in England and Wales. In particular, it was a time of change from a selective system of grammar schools and secondary modern schools to comprehensive schools, and a time of change in examinations. In 1970, there were 1145 comprehensives in England and this more than doubled to 2596 by 1975. In terms of students the increase was from 937 K to 2460 K in comprehensive schools. The Certificate of Secondary Education (CSE) was introduced in 1965 for those who stayed on until 16 in secondary modern and comprehensive schools, and the top grades were officially considered equivalent to a General Certificate of Education (GCE) O-level taken mainly by grammar school children. The CSEs covered both academic and more vocational subjects, but the possession of CSEs marked the owner as a ‘failure’ in the 11+. Selecting a secondary modern school for the main research site was selecting a school where most students were likely to already consider themselves as academic failure – and they were so defined by others.

A third major change was the Raising of the School Leaving Age (RSLA) from 15 to 16. This change had been provided for in the 1944 Education Act, the Newsom Report of 1963 recommended action, and money had been provided for new school buildings, but it was only on 1st February 1972, when about two-thirds of young people overall were already deciding to stay an extra year, that Margaret Thatcher as Secretary of State for Education laid an Order in Council before Parliament to make the change – from 1st September 1972 all children were forced to continue in school for an extra year, leaving at either the Easter or Summer after their 16th birthday rather than their 15th birthday. The boys in Willis’s study were thus the first year to be forced to stay until 16 and probably learnt that it would certainly apply to them in February of the same year in which it was implemented. This was thus a very unusual year to choose to study the transition from school to work of non-academic boys. Indeed, this might explain why the boys were studied for five (or four) terms as the research may have started before it was known for certain that these boys would be forced to stay an extra year. Conducting an ethnography for two terms within school and about six months into work might have been a far more appropriate research design than following them for five terms and then into work.

As it is, Willis plays down these dramatic changes in the educational system and writes about the culture he found in the groups of 12 lads largely as if they were irrelevant. He claims, without offering any evidence, that RSLA had a ‘peculiar function exposing the oppositional dimension of what had been a more submerged counter-school culture before’ (Willis 1977, 115, note 16). But, in many ways, these lads had every right to feel they were being imprisoned within school. One would expect them to be angry that, as they saw it, another year of their lives was going to be wasted in school when they wanted to be out in the world earning money. They were underachieving boys who wanted to leave school and get into one of the many easily-available manual jobs which would give them some money and independence. Or rather, full-time work that would give them more money, as Willis (1977, 39) suggests that all of ‘the lads’ were already in some form of part-time work (presumably in their final school year). They were working in ‘small businesses, shops, on milk rounds, as cleaners, ice-cream salesmen, as stackers in supermarkets. […] Over ten hours a week is not uncommon.’ Willis claims that one boy earned over £20 per week and that the average for the others was ‘something under five pounds’. It is sad that no comparisons are drawn between these figures for ‘the lads’ and those for the ‘ear’oles’ and other groups in the study school, for this might or might not set them aside as already having experience of work that allowed a ‘partial penetration’ of working-class life in employment. What is interpreted as a feature of the lads’ culture might well have been common to many in the school.

In the previous sub-section, I indicated that there was insufficient information on how the 12 lads were selected and on what basis. To say that they were working-class, non-conformist boys is insufficient, as conformity is a matter of degree and would have been continually negotiated by most of the boys in the school. But what is clear is that the 12 were picked-out four or five terms before they left school, as were the boys in other schools. The sample was fixed and did not change as the boys grew older and as theories developed. This strategy has something in common with more positivist statistical work, and does not take into account the idea of ‘grounded theory’ which was becoming known in the USA and UK during the 1960s (Glaser and Strauss 1965, 1967; Strauss and Corbin 1994). While the idea was being greatly debated while Willis was conducting his research, he does not adopt this process but sticks rigidly to the initial plan. While there is considerable debate about the detail of grounded theory and it has developed over time, the basic idea of selecting participants as the research proceeds, attempting to get theoretical saturation to categories, and testing developing theory against data drawn from participants specifically selected because they might challenge that theory, had become widely accepted. In contrast, Willis’ plan for the research identifies comparative groups before the theory is developed but Willis does not seem to use these comparative groups to test his theory. Strangely, the planned research with its various comparative cases was structured in a standard ‘positivistic’ way suitable for survey research, which may have been seen to have been necessary to gain funding. As Willis moved directly from part-time doctoral student to researcher, the initial grant application might well have been planned and written by someone other than or as well as himself. As ethnographic work has now become much more acceptable, there should be less need to propose comparative studies in present-day application.

I would argue that the success of Learning to Labour has much to do with over-generalisation – both by the author and by readers. Further, over-generalisation of the experiences of some non-conformist lads has led to a lack of understanding (or even problematization) of why, at that time, the bulk of working class boys actually did get working class jobs. This concern is embodied within the subtitle of the book: ‘How working class kids get working class jobs’. For the book has been taken, and is still taken by many ethnographers, as showing exactly this – how working class young people (at best, restricted to boys and at that particular time in history) creatively constructed a culture that echoed in many ways the culture of the shop floor jobs that they eventually occupied. The cultural form that ‘the lads’ created was seen to have within it a ‘partial penetration’ of the conditions of existence of its members and of their position within a class-based society. Their behaviour and attitudes towards school were seen as having a form of logic and rationality, yet, there is a moment ‘in working class culture when the manual giving of labour power represents both a freedom, election and transcendence, and a precise insertion into a system of exploitation and oppression of working class people’ (120). Willis’ focus was always class inequalities and the ways in which social inequalities reproduced themselves. Willis chose the 12 lads specifically because they were most likely to produce data that supported his commitment to the positive, creative, and artistic production of lived cultures. Ethnography simply cannot answer a research question that is as broad as ‘How do working class kids get working class job?’

I have argued elsewhere (Walford 2020) that many ethnographies would benefit from greater use of numerical data. We are given few numbers in the text of Learning to Labour, but it is clear that the vast majority of boys in the school were not part of this particular group and had not developed the form of oppositional culture described by ‘the lads’. The ‘ear’oles’ were probably a large group within the school, but there would have been others who had a more pragmatic relationship with learning, and others who had different attitudes to individual subjects of study. In all probability, all of these boys will also have ended in variety of manual and non-manual working-class jobs. But considering the ways in which the ‘ear’oles’ developed their own cultural form and cultural resistance did not fit with Willis’ purposes. It is only by showcasing his 12 lads at the expense of the others that these data can be used to illustrate his theory.

Of course, once a book is published the author has little control over how it is interpreted. Many of the ways in which Learning to Labour has been used to bolster arguments, find little support in the actual text, and the book is far more nuanced than some readings suggest. However, the pattern started in Profane Culture is repeated in Learning to Labour. The main case study school was at the heart of an inter-war housing estate which had ‘substantial West Indian and Asian minorities’ (4) who are not studied. Willis (1977, 4) states ‘I wanted to be as certain as possible that the group selected was typical of the working class in an industrial area’, but he also states that the 12 boys themselves were selected on the basis of friendship links and membership of some sort of oppositional culture within the school. These ‘lads’ were, in fact, deliberately selected to be not representative of the wider body of working class boys in the school or of working class boys in general – they were selected because they were perceived to be part of a particular oppositional school culture.

The problem with doing so is that it is simply inappropriate to generalise from the experiences of a small non-randomly selected sample of boys in one particular school. The move from the particular culture of these twelve lads to the much more general ‘working class culture’ is made continually within the book, when all that can be claimed realistically is that this particular cultural form is one of the many possible working class cultural forms that could be generated – indeed, only one of the many possible non-conformist forms that could be generated within the study school. While the book does include some discussion of alternative cultures that some other working class students develop in school, the reader is led to understand that it is only this particular form ‘created’ by ‘the lads’ that has continuities with working class jobs involving manual labour. The experiences and cultural forms created by the other boys are presented as the other pole of a simplistic bipolar ‘lads’ and ‘ear’oles’ dichotomy, even though Joey (15) talks of a third intermediate group, and there were probably a variety of different groups with their own cultural forms within even the one school.

Over-reliance on interviews

At a time when nearly everyone carries in their pocket or bag a smart phone which allows the almost limitless recording of both sound and pictures, it is difficult to think back to a situation before the internet, before mobile phones, before cassette recorders, and even before easily portable tape recorders. Referring to his doctoral work, Willis (Mills and Gibbs 2000, 218) talks about the arrival of a new tape recorder at CCCS:

The Centre had just spent a small fortune on a massive, top quality reel to reel tape-recorder which was like a suitcase, and I carried this huge suitcase around the hippy pads, stuck it on the table, unpacking great spools of tape. […] I saw it at the time as “For god’s sake get some data, get some data!”

With such ‘new technology’ it was possible for the first time for researchers to record directly the every word spoken in an interview. It could record not only the words, but the intonation, emphasis, pacing, and timing of what was said. Moreover, the tape recorder could be seen as providing ‘hard’ data that gave a spurious scientific validity to ethnographic research. It was a breakthrough, which almost every ethnographer gradually adopted as the technology became more compact and less intrusive. Willis was one of the ethnographers of education who led the way to an over-reliance on interview transcripts in ethnographic writing.

I have argued elsewhere that interviews are very strange situations (Walford 2018b). In its most structured form, the socially accepted rules of conversation and reciprocity between people are suspended. One person takes the lead and asks a series of questions of the other. The other has agreed that this is to be a special form of verbal interaction and is prepared for his or her views to be continuously questioned without the usual ability to be able to return the question. There is no natural ‘turn-taking’ and ability for both to control the flow of topics. Instead, the topics to be covered are under the control of the ‘interviewer,’ and the ‘interviewee’ is expected to have opinions or information about each of the questions asked. Moreover, and most strangely, what the interviewee says in this particularly odd situation is usually taken to have lasting importance – it is recorded for future analysis. This is not a transitory conversation but one that is invested with future significance. If interviews between two people are strange situations, group interviews are like the Wild West. Individuals make presentations to one another as well as the interviewer and ‘group think’ can escalate disclosed views beyond anything that an individual would hold. Given that the interview is such an extraordinary situation, it is remarkable that so many researchers take the data generated by them as reliable and valid. While Willis and other ethnographers are likely to use much more open-ended interview procedures, the basic format remains the same – topics and limits are determined by the interviewer and words recorded as ‘data’ for later analysis.

Concern over the validity of interviews is far from new. How interviewees respond to questions may well depend upon the time of day or year, the weather, and external events. Crucially, the appearance, gender, ethnicity, clothing, accent, and other variables associated with the interviewer can also influence what is said. We know that interviewers and interviewees co-construct the interview and that the replies to questions are produced for that particular occasion and circumstance. Interviewers and interviewees take part in a performative and rhetorical interchange and will select their words with care and will construct what they have to say to the particular circumstances. At best, interviewees will give only what they are prepared to reveal about their subjective perceptions of events and opinions. These perceptions and opinions will change over time, and according to circumstance. They may be at some considerable distance from any ‘reality’ as others might see it.

Douglas (1976, 57) argued that researchers should assume that people and groups are in conflict with one another, and that their aims and objectives often clash. He describes the problems that we encounter in interviews in terms of misinformation, evasion, lies, and fronts, and follows this with a detailed exposition of the problems of taken-for-granted meanings, problematic meanings, and self-deception. What people tell us in interviews is often not to be taken at face value, and this is particularly true when the interviewer asks about aspects of experience that are of special importance to the person being interviewed – for example, issues of sex, money, and power.

But misinformation, evasion, lies, and fronts are only part of the problem. We do not have full control over our lives – they are not inherently stable and coherent. The unexpected happens; the expected does not. In this uncertainty we all try to make sense of our own worlds, and the interview is one occasion when we try to do so in a semi-public forum. We try to present a reasonably rational image of our own uncertainty. People construct accounts about themselves, their activities, and beliefs that are far more coherent than the lived reality. In group interviews, in particular, as Wenk (2017, 10) argues, ‘familiar memories are recalled, retold, embellished (intentionally or not), and then restored as the newly edited storyline’. People’s brains construct memories such that they fit with how they wish to present themselves to themselves and to others. Just as many continually consciously ‘curate their lives’ on Instogram, we all unconsciously curate our live in our brains.

Yet, in interviews, and in particular in the transcribed versions of interviews, a coherent and constructed account gives the impression of permanence to something that is inherently transitory. It becomes a text that can be edited, copied, and recontextualised. In short, the transcribed interview encourages the possibility of the spoken word being taken too seriously. The phrase that someone happened to have used on a hot Monday afternoon following a double mathematics class gets wrenched out of its context and presented as if it represented the ‘truth’ about one person’s views or understandings.

In the case of Learning to Labour, some of the details of how the interviews were conducted are scattered throughout the book. It is only in the Appendix, where some of ‘the lads’ now in full-time work discuss their views of the book, that we discover that ‘the lads’ were taken out of lessons to be interviewed either individually or in groups. This appendix was very unusual and brave at the time, and even now, for it challenges the data production process. While not, of course, taking what is recorded here at face value for it was still produced in a group interview at a particular juncture, Bill, for example, says about the sessions with Willis, ‘[…] I enjoyed it ‘cos it was a skive, y’know, get out of lessons, I really wasn’t interested in it at first y’know. I could get out of lessons, have a smoke for an hour or so’ (Willis 1977, 196). Perc’s suggests that in a group ‘when yer with yer mates yer say a lot of things yer know that don’t really happen’, but is cut down by Bill and Joey who claim that none of it was made up. However, Joey then admits that ‘even if the individual acts were exaggerated the point’s still there’ and John (referring to ‘the lads’ themselves) replies that ‘They seem in the book a lot tougher than they actually were’.

Allowing a group to get out of lessons and to talk about themselves, where they know that co-operation with the researcher will be rewarded with further ‘skiving’ is likely to produce animated descriptions of laddish behaviour. We do not know how many individual or group interviews were conducted or even whether the group interviews were as a group of twelve or smaller sub-groups. But in this case of multiple gatherings and interviews, it is a good guess that the Hawthorne effect occurred and that the process of doing research influenced behaviour. It is also easy to imagine that the group solidified over the period of research and that their anti-school behaviour was, in part, designed to provide new anecdotes to recount to Willis in the group sessions. Continued stories of how they ‘had a laf’ ensured further ‘skiving’. An adult treating their accounts as descriptions of acceptable and meaningful activity is likely to reinforce such activity. Their growing self-identities as ‘the lads’ did indeed give some meaning to the lives of these individual boys who were unlikely to leave with any qualifications and who were defined by practically everyone else as failures. In the book there is little consideration of the ways in which Willis might have influenced ‘the lads’ even though he later writes a great deal about reflexivity (Willis 1980, 2000) in relation to theory in particular. Again this is an example of Willis labouring to learn the best ways of conducting ethnography.

Looking back on his early work Willis shows himself to be well aware of the potential deleterious effects of taped interviews (Mills and Gibb 2000, 219–220). He talks about the need for self-security during the research process and of researchers generating a store of data that can be used later. He admits: ‘I think I’m partly responsible for cultural studies’ over-reliance on verbal responses, and on a cultural studies notion of a quick, dirty raid which is discursively based.’ He is correct, although Learning to Labour also draws from considerable observational work, but the over-reliance and uncritical use of recorded interviews also applies to other ethnographic work, and he is far from being the only researcher who should take some of the blame.

Willis’ later academci career

Between 1972 and 1981, Willis was a Senior Research Fellow at CCCS working on various studies on youth culture projects, the first of which was funded by the Social Science Research Council from 1972 to 1975, which led to Learning to Labour. He reports in the Preface to Learning to Labour that a second research project had been funded by SSRC to focus on ‘the young workers and shop floor culture’ which was to provide material for a sequel to the book, but this book did not materialise, although one extended interview was used in a much later book chapter (Willis 2001). Willis lost his research fellowship at CCCS in the early 1980s due to cut-backs in university funding, and was unemployed for about a year (Mills and Gibb 2000, 212).

Willis’ academic career is far from the traditionally conventional for he spent the years from 1981 as a policy advisor to Wolverhampton Borough Council and was soon asked in to conduct a Youth Review to understand youth activities and attitudes in a very changed occupational environment to a decade earlier with ‘the lads’. There was increased concern about growing unemployment and one third of these were under 25 in Wolverhampton. This project was mainly a questionnaire-based survey conducted by interviewers who were recent sociology graduates from Wolverhampton Polytechnic. Some of these then worked full-time on the project and various advisors were also directly involved. It also involved a review of all the main local agency departments relating to youth, and various group sessions with young people, interviews with interested parties, and the publication of the initial review and a later book (Willis et al. 1985, 1988). This research marked a distinct step away from ethnography, but probably not from choice. In 1988, he was describing himself as a ‘Visiting Professor at Wolverhampton Polytechnic and a freelance writer and researcher on youth and youth culture’ (Willis et al. 1988: back cover). But from 1988 he directed a research programme funded by the Gulbenkian Foundation into the cultural activities of young people, which led to two further books (Willis 1990; Willis et al. 1990). It is a common irony that once ethnographers have begun to learn the labour of conducting ethnographic research they frequently get pulled away from conducting their own work as they move into administrative and directorial roles. In this case the ‘spine’ of the project was an ethnographic and interview project conducted again in Wolverhapton, but mainly by Joyce Canaan, and the two books are a synthesis of at least 15 related projects.

From 1987 onwards Willis was a part-time Equal Opportunities and research methods advisor at Wolverhampton, and from 1992 to 1997 he was half-time Reader in Media and Cultural Studies at the newly designated Wolverhampton University, but paid by the University of Växjö, Sweden. He became Head of Media Studies in 1997, and was a Professor at Wolverhampton from 1999 to 2002 during which he published The Ethnographic Imagination (2000) and co-founded the journal Ethnography (Bracke, Davidson, and Guadeloupe 2021). He moved to Keele University in 2003 first as Professor of Social/Cultural Ethnography, then as Head of the Department of Management Studies. He officially retired in 2010, but went on to invited professorships at Princeton (2010–2014) and Beijing Normal University (2014–2017). The latter resulted in a book on China (Willis 2020).

Conclusion

In this paper, I have made some major criticisms of the research methods used in a classic ethnographic research study, Learning to Labour. In many ways this criticism is unfair as Willis was one of the first to use ethnography to try to understand the relationships between school and employment and to examine the micro-level activities and processes that might lead to class reproduction. Willis, without any specific methods training, was learning and labouring to become an ethnographer at a time and situation where ethnography was not highly valued in his local research environment or in the wider growing disciplines of cultural studies or sociology of education.

However, it is important to examine the research methods used simply because Learning to Labour is still widely read, cited, and used as an example of ethnographic work. I argue that Learning to Labour has been damaging to the development of educational ethnography because the methods used and ways of writing have been followed by many others learning how to do ethnography.

I have discussed these problems as (1) lack of full descriptions and ambiguities of the research methods, (2) problems of generalisation and site/sample selection, and (3) over-reliance on interviews. More recent ethnographic work is more likely to give more space to research methods in publications, but details are still often missing. Problems of generalisation and site/sample selection still abound. All too often researchers make wild generalisations from a small sample not even specifically selected to match the research questions. They select sites more on the basis of convenience than research purposes, and keep rigidly to initial plans rather than adapt their methods to suit increasing knowledge of the issues being researched. Finally, the over-reliance on interviews has increased such that, in the USA in particular, many researchers happily give the title ‘ethnographic’ to interview-only studies.

Disclosure statement

No potential conflict of interest was reported by the author(s).