The Devil in the Detail: Producing an Account of Self-Harm
Bogusia Temple and Jennifer Harris

The Qualitative Report, Volume 5, Numbers 1 & 2, May, 2000


In this article we discuss self-harm data from an A & E (Emergency Room) Department in an English hospital. In order to be able to examine the relationship between data collection, analysis and findings we focus on the processes we used as researchers in constructing the dataset. Doing this, we argue, is as important as just analysing findings since this process in part constructs the findings. Moreover, how people's actions are defined may impact on the way they are treated.


C. Wright Mills (1959) argues that certain kinds of sociological research are obsessed with what he calls the 'syntax' of research. That is, research is undertaken in a rigid, mechanistic manner that is concerned with the statement of rules and logic. He believes that when the social world is involved the 'semantics' of research are just as important. Discussions of meanings, relationships and principles can be used in ways that stretch the sociological imagination. There is a need, therefore, to look not only at how to do research, but also at what it is that researchers are doing. There are many textbooks concerned with the 'how' of research, that is the techniques or methods used. These are skills that social scientists can contribute to research. However, they can also make the kind of contribution advocated by Mills. That is, they can reflect on what it is they are doing and discuss the implications of using the methods they have chosen. It is in this spirit that we examine quantitative data about self-harm in an Accident and Emergency (A & E) Department in England qualitatively. We argue that it is as important to examine how data are collected and analysed as it is to discuss the findings. Indeed, we believe that the former influences the latter.

Method As Sacred Prescription

There has been much recent debate about the status of social science research (for example, Hammersley & Gomm, 1997; Humphries, 1997). We do not rehearse the arguments in detail. There are broadly speaking two camps in this debate. Subscribers to the first, here called 'positivists', believe that it is possible to be objective in research; followers of the second, 'social constructionist', camp argue that it is not possible to remove the influence of the researcher from their research. No one can see an issue from all perspectives and who we are in terms of our class, gender, age and ethnicity, amongst other social characteristics influences how we see things. We position ourselves in the second camp and in this article we discuss some of the decisions that had to be made in our research and our perspective on them. We believe that there is more than one way to set up and analyse the data and that the process of construction is important.

Following Burr's (1995) definition of a social constructionist perspective we use the term to include any theory of knowledge which contains one or more of the following assumptions. First, it involves a critical stance towards knowledge. Knowledge of the social world cannot be gathered objectively and in an unbiased way. We can only know our social world from our particular position in it and we cannot remove the influence we have on our research. Data is not detachable from the theory it is located within, since facts only make sense within a theoretical framework. Social scientists select which pieces of information will be considered as 'facts'. They interpret information as significant or not within a chosen perspective. This is not to a denial of 'reality', rather it is an acceptance that researchers are human beings who cannot see an issue from all sides at once.

Second, all knowledge is seen as historically and culturally specific. Concepts change over time and place and no one definition of a concept is any more 'true' than another. Third, social processes sustain knowledge. By this Burr (1995) is indicating that knowledge is not out there waiting for the researcher to find it, but rather it is produced in the interactions between people. In other words, research is a social encounter. So, for example, there is no 'self harm' existing independently of the medical and lay constructions of it. People attach a label to a set of symptoms for a particular purpose. This does not mean that there is no pathology. Evidence is gathered by specialists using their experience as well as their training from a range of possible symptoms to put together a diagnosis of 'self-harm'.

Fourth, knowledge and social action go together. If understanding is negotiated between researchers and subjects of research then there are many possible constructions of social life. The way concepts are defined matters for the kinds of action which follow. Burr (1995) gives the example of drunkenness. If it is defined as a crime then imprisonment may follow. If it is defined as a kind of addiction, then medical or psychological treatment may be seen as more appropriate. This point is particularly relevant for researchers examining self-harm where 'it' has been defined as an individual problem and as evidence of 'mental illness'. This has consequences for treatment (see below). Re-defining 'it' as a social act as well as an individual one has implications for how people who self-harm are seen and treated.

Dominant amongst the self-harm discourses are medical and psychiatric ones and these are often based on a positivist paradigm. Many researchers in this area also follow such a paradigm. Medical discourses label people who self harm as 'mentally ill' since any 'rational' person would not deliberately harm themselves. Doctors are taught to treat the body and to pass on 'mental' problems to other experts. The body and the mind are separated. The patient is turned into a bundle of symptoms, an individual, who can be managed and treated by medical experts, rather than as an individual within a wider social environment that should be taken into account. The manoeuvre of separating mind and body serves to individualise and medicalise acts of self-harm, which are located within social contexts. Favazza's (1992) work is interesting here in that he examines social influences on definitions of self-harm and looks at the social factors influencing such acts. He suggests that acts of self harm may blur body/mind distinctions: "Just as a significant event symbolically can be burned in one's memory, so too it literally can be burned into one's skin" (1992, p. 195).

Survivors' accounts often target such positivist approaches. For example, Babiker and Arnold (1997, p. 12) suggest that diagnoses such as Borderline Personality Disorder involve a circular argument since they consist of descriptions of the behaviour of individuals who receive the diagnosis. Survivors' accounts of self-harm locate them within social contexts such as sexual abuse and domestic violence (Babiker & Arnold, 1997) indicating that seemingly individual problems cannot be separated from social issues. Moreover, their experiences of differences in the treatment of, for example, men and women (Pembroke, 1996) and members of minority ethnic groups (Patel & Strachan, 1997), point to influences at work in the labelling of a patient under the category 'self-harm' that have nothing to do with objective science and much to do with stereotypes.

Recent research has suggested that the definition of self-harm is culturally ascribed and a societal issue (Harrison, 1994, 1995; Pembroke, 1996; Spandler, 1996). Harrison (1994) argues for a move away from blaming, or 'pathologising' women to locating their experiences within the working of patriarchal and other power structures. Babiker and Arnold's (1997) analysis of self mutilation suggests that societal factors are at work in how people see their own bodies and relationships. Certain kinds of wounding are seen as self harm, and therefore deliberate, whilst other kinds such as body piercing are viewed as culturally acceptable. Much medical and psychiatric discourse is concerned with pointing to the individual nature of the former whilst locating the latter as socially acceptable forms of behaviour.

Schwandt (1996) argues that following a positivist (objective scientific) position leads to the pursuit of the perfection of method as the only goal: methods and methods alone produce findings, if we can just remove all bias and adopt the perfect method, then we will find 'the truth'. Method is, to use Schwandt's words, used as 'sacred prescription' (1996, p. 60). For researchers who see research in the social world along the lines spelt out by Burr (1995), then looking to the perfection of method alone is not enough. There will never be one view of reality, all accounts are contextual and contestable.

We believe that all researchers make decisions about who to include and who to leave out in their research. In the field of self harm, this involves comparing different acts of self harm, judging what they mean, and sometimes merging like cases for analysis and comparing them to 'different' cases. The ways is which these boundaries are drawn affect how the data can be interpreted. Once the research process is examined in detail it becomes impossible to see the 'facts' about self harm as lying outside of the process by which they are produced. To examine this claim we look at some of the decisions we made in our research.

The Research

In collecting data on 'deliberate self-harm' a decision had been made in A & E that it was possible to decide whether an incident involves an act of 'deliberate self-harm'. In classifying someone under this category each doctor made a decision about whether a person had harmed themselves intentionally. The term 'deliberate' self-harm is used by many doctors, as it is in much research in the area (see, for example, Crawford & Wessely, 1998; Hawton & Fagg, 1992), even though people who self harm have objected to the label as implying premeditation and willfulness (Pembroke, 1996). However, deciding whether someone has harmed themselves 'deliberately' is by no means straightforward. Moreover, research has shown (Solomon & Farrand, 1996) that suicide is not the same as self-harm. The doctors in A & E made no attempt to distinguish the two, and indeed this may not be relevant or possible. The statistics therefore include people whose acts may have had very different meanings.

The issues involved in deciding the intentions in an act such as self-harm are most obvious when patients' ages are analysed. The researchers found it surprising that very young children were included in the data. This has led us to question the ascription of 'deliberate' intent to injure to children, who it could be argued, are too young to know the consequences of their behaviour. For example, should a baby ever be in the data? Even if we leave them out, the problem still remains of establishing at what age someone can be said to know the consequences of their actions. There is no simple answer to this question, but researchers make assumptions such as these when they decide how old people have to be to be included in their research. For example, Scott and Powell (1993) and Hawton and Goldacre (1982) choose 12 as the minimum age of inclusion. However, personal accounts by people who self-harm suggest that they began when they were under 12 (see for example Solomon & Farrand, 1996). There is no one age at which it is 'right' to include people. However, making this decision is not simply a 'technical' issue. It involves deciding when someone understands the consequences of their actions and if they should be included together with others in a particular category.

We give one more example of the kind of decisions involved in analysing data before discussing what we have learnt from our research about the semantics of research (Wright Mills, 1959). That is, we look at one more 'variable': the site of injury. Favazza (1992) suggests that the site of injury is important and that the part of the body chosen as the site of an act of self-harm may have significance. It may be, for example, that the eye has been chosen for religious reasons:

What I say to you is: anyone who looks lustfully at a woman has already committed adultery with her in his thoughts. If your eye is your trouble, gouge it out and throw it away! Better to lose part of your body than to have it all cast into Gehenna (Matthew 5, 28-29).

We had very detailed data about site of injury and had to make a decision about how to analyse it. We finally decided to keep the statistical analysis at the broad level of 'wound' and not to use the detailed data. This was due partly to technical reasons and to time constraints. It was also based on the assumption that people who wound themselves have something in common with others who have done so. All researchers make choices about who to include and who is 'other' for the purposes at hand.


We locate ourselves firmly within the social constructionist position described above. This does not mean that we believe that there is no such thing as self harm, but that the definition used has an affect on who is included in the category and how they are treated. Patients who are suspected of having harmed themselves are sometimes seen as less deserving of sympathy and treatment than patients who are classified as having injured themselves accidentally (Babiker & Arnold, 1997). Medical/psychiatric definitions and analyses of self harm are increasingly being challenged by the growth of self help survivor groups and by treatment which aims to provide advice and understanding to survivors regardless of sometimes arbitrary definitions of intent.

We bring our own perspectives to our research. One of us (Temple) has always been an academic researcher, the other (Harris) was previously a practising social worker. We bring our differing experiences to our research. We are arguing that it is impossible for us, or for anyone, to avoid making judgements about people's behaviour. All classifications are based on experiences. For example, we don't believe that there is one correct age at which to include someone in statistics about self-harm. People begin to self-harm at different ages and any minimum age is arbitrary. It is not possible to avoid value judgements about people when analysing data and all data analysis involves processes of classification and decisions about who is like whom and who is not. That is, all researchers produce their own pictures of people they carry out research with. Researchers who use numbers are no different in this respect from those who use words, although the processes involved may be different (for useful contributions to the debate see Nau, 1995; Kanpol, 1997).

Investigating how our dataset was constructed, by whom and for what purposes has re-enforced for us the fallacy of the 'objective' researcher gathering 'facts' about the social world. The devil is in the detail for researchers who aim to construct such an account of self-harm. Construct it one way and another researcher will re-define and re-construct it another way. Documenting the tasks involved in setting up and analysing any dataset demonstrates the impossibility of making value free assumptions about who self harms and why. Immersing ourselves in the minutiae of quantification we have learnt at least two valuable lessons: it is impossible to be objective and provide one all embracing analysis of self harm; and that recognising this involves accepting that all research accounts paint pictures of research participants. However, we must point out that we are not arguing that quantitative research is a waste of time. Many of the points we have made apply equally to qualitative research (for a useful debate the on the implications of dismissing quantitative research see Oakley, 1998). We as researchers have added our own, equally contestable, definitions and interpretations to those of others who have had an input into deciding how to paint the canvas.


      Babiker, G., & Arnold, L. (1997). The language of injury: Comprehending self-mutilation. Leicester: The British Psychological Society.

      Burr, V. (1995). An introduction to social constructionism. London: Routledge.

      Crawford, M. J., & Wessely, S. (1998). Does initial management affect the rate of repetition of deliberate self-harm? Cohort study. British Medical Journal, 317, 985.

      Favazza, A. R. (1992). Bodies under siege: Self mutilation and body modification in culture and psychiatry. Baltimore: John Hopkins University Press.

      Hammersley, M., & Gomm, R. (1997). Bias in Social Research. Sociological Research Online, 2(1). Available:

      Harrison, D. (1994). Understanding self harm. London: MIND.

      Harrison, D. (1995). Vicious circles: An Exploration of women and self-harm in society. London: GPMH Publications.

      Hawton, K., & Fagg, J. (1992). Trends in deliberate self poisoning and self injury in Oxford, 1976-90. British Medical Journal, 304, 1409-1411.

      Hawton, K., & Goldacre, M. (1982). Hospital admissions for adverse effects of medicinal agents (mainly self poisoning) among adolescents in the Oxford region. British Journal of Psychiatry, 141, 166-170.

      Humphries, B. (1997). From critical thought to emancipatory action: Contradictory research goals? Sociological Research Online, 2(1). Available:

      Kanpol, B. (1997). Reflective critical inquiry on critical inquiry: A critical ethnographic dilemma continued. The Qualitative Report, 3(4). Available:

      Nau, D. (1995). Mixing methodologies: Can bimodal research be a viable post-positivist tool? The Qualitative Report, 2(3). Available:

      Oakley, A. (1998). Gender, methodology and people's ways of knowing: Some problems with feminism and the paradigm debate in social science. Sociology, 32(4), 707-731.

      Patel K., & Strachan R. (1997). Anti-Racist Practice or Heuristics. Practice, 9(3), 15-26.

      Pembroke, L. R. (Ed.). (1996). Self-harm: Perspectives from personal experience. London: Survivors Speak Out.

      Schwandt, T. (1996). Farewell to criteriology. Qualitative Inquiry, 2, 58-72.

      Scott, J., House, R., Yates, M., & Harrington, J. (1997). Individual risk factors for early repetition of deliberate self-harm. British Journal of Medical Psychology, 70, 387-393. [Not cited in the text]

      Scott, S. W., & Powell, J. C. (1993). Adolescent self-mutilation in a rural area. Journal of Adolescence, 16, 101-105.

      Solomon, Y., & Farrand, J. (1996). "Why don't you do it properly?" Young women who self-injure. Journal of Adolescence, 19, 111-119.

      Spandler, H. (1996). Who's hurting who? Young people, self-harm and suicide. Manchester: 42nd Street.

      Wright Mills, C. (1959). The sociological imagination. London: Oxford University Press.

Author Note

+Dr. Bogusia Temple is Research Development Manager in the Institute for Health Research at The University of Salford, England. She is interested in general methodological issues and has published theoretical articles on both quantitative and qualitative methodology, as well as articles specifically concerned with ethnicity and gender.

Dr. Jennifer Harris is Senior Research Fellow in the Social Policy Research Unit at the University of York, England. Her main research interests are in the ethics, implementation and design of research. She has published in the fields of qualitative methodology and disability studies.

The authors can be contacted through Dr Harris at the Social Policy Research Unit, University of York, York YO1O 5DD, England or by e-mail at:

Article Citation

      Temple, B., & Harris, J. (2000, May). The devil in the detail: Producing an account of self-harm. [20 paragraphs]. The Qualitative Report [On-line serial], 5(1/2). Available:

Bogusia Temple and Jennifer Harris
2000 copyright

Return to the top of the paper.

Return to the Table of Contents.