QUALITATIVE RESEARCH METHODS

For a number of reasons qualitative techniques are especially appropriate –maybe uniquely appropriate– to the study of student experience and other aspects of teaching and learning in Higher Education. But the nature of the data which are produced by qualitative research, the way such data must be handled, and the use to which such data can be put, are all rather readily open to confusion and misunderstanding. Confusion often lies in the failure to differentiate among several orientations to qualitative data. (i) Positivist research presupposes that there is some underlying, true, unequivocal reality, and a theory covering this is to be sought by the research. There must be evidence of validity –in the sense of a match between the data and the reality they are sup-


Abstract
For a number of reasons qualitative techniques are especially appropriate -maybe uniquely appropriate-to the study of student experience and other aspects of teaching and learning in Higher Education.But the nature of the data which are produced by qualitative research, the way such data must be handled, and the use to which such data can be put, are all rather readily open to confusion and misunderstanding.Confusion often lies in the failure to differentiate among several orientations to qualitative data.
(i) Positivist research presupposes that there is some underlying, true, unequivocal reality, and a theory covering this is to be sought by the research.There must be evidence of validity -in the sense of a match between the data and the reality they are supposed to reveal.
(ii) Non-positivist research is of a number of kinds, despite often being treated as unified.Examples are (a) descriptive ('phenomenological') research; (b) interpretative ('hermeneutic') research, and (c) discourse analysis.

*
A variant of this paper, focusing on health research, is published in two parts in the journal Nurse Education Today: "The variety of qualitative research, part one: Introduction to the problem" and "The variety of qualitative research, part two: Non-positivist approaches".Nurse Education Today, 17, 215-218;219-224 (1997).

INTRODUCTION
Very many kinds of research in Higher Education call for qualitative approaches.Indeed, it would be possible to argue that the more genuinely the topic of interest reflects either student experience or the core student-lecturer relationship, the more likely it is that qualitative methods will be found most appropriate.
However, the immediate impetus for the writing of this paper was the all-toofrequent occurrence of misunderstandings over qualitative methods in papers which I received for refereeing.Perhaps the extreme form of error is the assumption, especially found in papers reporting the evaluation of curriculum innovations, that if research employs 'qualitative methods', no rigour at all is required and analysis can merely consist of the presentation of 'quotable quotes' from conversations, interviews of various sorts, or comments sheets.
But even when attempting to use qualitative techniques such as interviewing or participant observation rigorously, colleagues and students often show confusion in deciding how to treat the data in an appropriate way and in thinking about what the methodological issues are which need confronting.For instance, they may struggle to frame a hypothesis which their work will test, when their objective is not really hypothesis-testing at all, but is to undertake an exploratory investigation of an as-yet uncharted area of student or tutor experience.Or they may anxiously seek evidence for the validity and reliability of the interviews they have undertaken, when the data was being used to look at the nature of certain attitudes and values in the talk itself and there is no call for any external criterion of adequacy, even if one were available.
It has become clear that there is a need for an account of the variety of qualitative data, so that such confusions can more easily be put to rest.In this paper, then, I intend to discuss the diverse nature of qualitative research data, making it perhaps rather more plain why, for example, issues of validity ought to be addressed in some contexts but not all, why the data sometimes must be regarded as referring to the nature of the culture rather than being the creative product of the individual person, and what is implied when the research focus is on 'experience', on 'discourse' or on 'variables'.
By way of concrete illustration in what follows, I will use material chosen from the transcripts of interviews on student cheating and plagiarism.Just to clarify the context, 1 "Guilty in whose eyes?University students' perception of cheating and plagiarism".Studies in Higher Education, 22, 187-203 (1997) In this workshop, the varieties of qualitative research will be exemplified (rather than described in technical detail) using examples from our recent work on cheating and plagiarism 1 .And there will be participation throughout.
the research was carried out as coursework towards the end of my semester-long Master's degree unit in qualitative research interviewing.Course members each interviewed one student who was not undertaking the unit, and completed a full analysis and report on that one interview.Further analysis on the set of interviews as a whole was carried out by myself and my colleague Philip Bannister.
The students had, by the time they began interviewing, received some nine hours of instruction and discussion in qualitative research interviewing.Confidentiality and anonymity were promised to the interviewees, and this was scrupulously maintained (especially important since confessions of cheating were not uncommon).
Take the following as an example of the material.One thing to register immediately (if this oft-iterated point needs to be re-emphasised) is the immense richness of qualitative research data.But what I want to do now is to show how varied can be the view of the data taken by researchers who have different purposes in mind for it.

POSITIVIST QUALITATIVE RESEARCH
Positivism takes it as axiomatic that there is a real world which has determinate characteristics, and the purpose of science is to model this world in its theories.These theories will show how certain variables interrelate, especially how they relate to each other in a cause-and-effect fashion.Mathematical formulations of the relationships between variables are to be sought if at all possible.The purpose of research is to test hypotheses regarding relationships between variables, and to reach, by closer and closer approximation, theories which can begin to be regarded as having the status of scientific laws.Positivism is characterised in the way I have laid out by both critics and advocates of the viewpoint (e.g.Smith, Harré and van Langenhove, 1995, pp. 2-4;Polkinghorne, 1983, pp. 15-20, and Chapter 2;Rudner, 1966, p. 11).
Different writers place different emphases on the various elements in the above picture of positive science, but the key thing is the idea that there is an underlying unequivocal reality, consisting of a web of relationships among specifiable variables, and scientific theories are testable models of that reality.Now, it is often incorrectly thought that qualitative research data cannot, by definition, fall within a positivist philosophy of science.It is certainly true that positivism favours quantitative methods, since these are regarded as capable of generating data which are measures of the variables which are taken as the stuff of which reality is made.Thus, in comparing research techniques, Galtung (1967) finds that qualitative ones are less powerful as tools of science than the controlled experiment on either or both of the criteria of (a) "immediacy of data to the research hypothesis" (because they rely on verbal accounts of cause-andeffect relationships rather than direct observation of these) and (b) the degree of experimental control which can be imposed on the collection of data.Nevertheless, Galtung and other positivists do not rule qualitative techniques out of court.They are regarded as particularly valuable in situations where there is no possibility of experimentation, or the variables have not yet been specified and an open-ended process to discover the relevant factors is needed.So the qualitative interview data from the cheating study could very well be treated in a positivistic way.It only requires that the researcher holds assumptions about the nature of the data which conform to the realist and natural scientific outlook I have outlined.
Interview data on cheating could be used to identify the variables which could later be built into hypotheses concerning the likelihood of cheating.Let's do this with the snatch of interview data I have already introduced: I mean, the actual exam at the end isn't a test of...I mean, it allows me to go further -it allows me to do whatever I want to do, so I've got to do, you know, work in that direction.But I don't think I would ... not unless I really thought I was going to fail and it wasn't a subject that I was planning to specialise in.It was, you know, just something I needed to get to that level.Because, if I thought I was failing on the subjects that I wanted to continue on, then I don't know whether I'd even be able -I'd want to carry on, because I'd want to be able to do well, without an incredible amount of problem, or I'd decide that it wasn't the course for me if I couldn't do it ... The interviewee is saying here that she would differentiate between units which matter as a basis for further studies and those whose content is not relevant in this sense to her longer-term interests.It would be foolish to cheat to avoid failure in units which are important in this sense, because things would become increasingly difficult.Then she says that she would be prepared to help someone cheat (and, we may infer, cheat herself) if units are merely hurdles to be surmounted in the process of getting onto units which matter.She talks about this in the context of her brother's GCSE problems: Yeah -like I'm prepared to help [interviewee's younger brother] in his GCSEs because I see it as just in order to get to something that he could do easily, you know, and sort of shine in -the GCSEs are just a way of testing that doesn't suit him, but is necessary and, you know, is below him-but is necessary for him to get somewhere where he, you know, and he can't do this, he has a problem with this method of, so yes, yeah, I'd be prepared to help him cheat.Like, helping him with assessments.
So, in positivist mode, we can analyse the qualitative interview data in such a way that it provides evidence of a class of variables which we can tentatively hypothesise as related to the likelihood of cheating: • If the person is having difficulty with a unit, this increases the likelihood of cheating in the unit.• If the person sees the unit as unimportant to their longer-term interests but merely a hurdle to be surmounted, this increases the likelihood of cheating in the unit.• If there are apparently arbitrary features of the unit (like the mode of assessment) which artificially increase the difficulty of passing, this increases the likelihood of cheating in the unit.
In the interview as a whole, very many more factors could be discovered, including 'levels of reprehensibleness' of instances of cheating (imaginary or real) from the student.And the tentative variables discovered in one interview could be weighed up against indications from other interviews.(We would probably have to say that there were personality factors at work, or at least that different students have different outlooks on cheating.) The aim of such qualitative work, carried out within a positivist understanding of research, would be to establish a model of student behaviour which specified the variables affecting the decision to cheat, ideally to be tested later using techniques which approximated more closely to the controlled experiment.
In this context, it is not surprising that it has been suggested (e.g. by Reichenbach, 1938) that science is a two-stage process: hypotheses are first of all generated, then later these hypotheses are tested.(The distinction being between the realms of discovery and of justification).Many positivist-oriented authors have indeed treated qualitative research as appropriate mainly for the first stage, regarding quantification and hypothesis testing as a follow-up which is more precise and rigorous, and which would enable the researcher to produce scientific laws.So our qualitative research interviews concerning cheating would be expected to allow the discovery of some appealing hypotheses about the nature of student behaviour: later, these hypotheses might be tested by structured questionnaires including rating scales designed to quantify the variables which have been indicated as possible factors in the decision to cheat, or by other quantitative approaches (cf.Roberts and Toombs, 1993;Franklyn-Stokes and Newstead, 1995).
In certain cases, qualitative methods, though less 'rigorous' from a positivist viewpoint, may be the best approach even to hypothesis testing.Where there is no way of operationalising (i.e.formalising as measurement operations) the variables quantitatively, the evidence of the presence or absence of the variables -and maybe some indication of their strength-may come from the analysis of interviews or other qualitative techniques.
It will be plain now why it is that, within positivist research, in which human social activity is supposed to be the lawful outcome of real, specifiable 'variables', the reliability and validity of measures is given high priority (see, for a qualitative research account, Kirk and Miller, 1986).Qualitative techniques such as participant observation and the use of the long, conversational research interview give rise to data which could be dismissed by positivists as 'subjective' -both in the sense that the researcher has a large part in their generation and therefore will have an influence on the findings, and also in the sense that the interpretation of the group's reality (by investigator and by members themselves) may be very different from what is 'really' happening.Validity refers to a number of different things, all to do with whether the measuring instrument is actually measuring the variable which it was intended to measure.Reliability refers to the stabil-ity, consistency, and precision of the measuring instrument.If we are to take the qualitative research interview or participant observation as techniques aimed at measuring variables reliably and validly, as the positivist would demand, then particular steps have been suggested for overcoming distortions and biases (see Silverman, 1994).
I cannot produce a neat schema for doing valid and reliable work within the positivist understanding of qualitative research (I have reasons for believing such a schema is impossible to lay down exhaustively), but the following matters are relevant: • It is important to try to be aware of presuppositions, and let them be hazarded by adopting an attitude which deliberately seeks out information which contradicts early expectations.• The researcher's involvement in the world of the participants in the research is twoedged.It may be a support for validity -the actual, multi-dimensional social world tends to impose its meanings on the research and counters the researcher's naive expectations.On the other hand, measures to enhance objectivity (such as triangulation of methods mentioned below) may be thought necessary.• Something like reliability can be considered in qualitative research.Colleagues can be invited to perform independent analyses of data, for instance.• Validity checks can be encouraged by making fieldnotes and interview transcripts, which justify the claims made, available to colleagues.• The persuasiveness of findings can be strengthened by showing that information obtained from various sources (corroboration) in various ways (triangulation) converges.In other words, it is often advised that increasing the diversity of research participants and using a variety of research techniques will enhance the validity and reliability of findings.• Participant agreement.In certain circumstances, members of the group being investigated are asked to comment on the researcher's findings, and this is used to provide evidence of validity.(Though there is controversy over the validatory evidence participant agreement provides.See, for full discussion, Ashworth, 1993).
Very many methodologists concerned with qualitative research and its techniques are avowedly anti-positivistic, in that they reject the idea that the human social world can be accurately described in terms of the lawful interaction of a number of discrete variables which determine all behaviour and action.Yet often, in their overarching concern to minimise bias and increase the evidence of validity, they betray the fact that, fundamentally, they share the positivists' view.They, too, see human social life as grounded in an unequivocal reality which it is the researcher's job to reveal in as clear a way as possible.(This covert neo-positivism is discussed in Ashworth, 1995.)In my view, 'grounded theory' of Glaser, Strauss and their colleagues (e.g.Strauss and Corbin, 1990) has many of the hallmarks of positivism, despite the fact that qualitative techniques are seen appropriate to the whole research process, not merely to the early 'discovery' phase.Some users of the approach (e.g.Charmaz, 1995) explicitly reject the positivism of its founders, but it is explicit in the original expositions of grounded theory (Glaser and Strauss, 1967) that the data relate to an unequivocal underlying reality.The move is very soon made from data description to the erection of concepts which link together as a theoretical structure through which the data are explained.
Because of its relative clarity, grounded theory, whereby the discovery of a 'theory' governing some social phenomenon is made inductively on the basis of the painstaking analysis of data, is coming to be a favourite form of qualitative research, going beyond the boundaries of sociological research in which it arose.A step-by-step account of a process of data analysis is provided which is comforting to the researcher who is perplexed about what to do with the mass of information that even a modest qualitative project generates.However, there is a very real danger that grounded theory can be taken as a technology for doing qualitative analyses.Non-positivistic qualitative research cannot be, and ought not attempt to become, a mechanical process akin to experimentation.
The viewpoints to be discussed in the rest of this paper take a different approach, and the way they treat qualitative data is, accordingly, quite distinct to that of the positivists.

NON-POSITIVIST QUALITATIVE RESEARCH
What is the nature of the scientific product at which qualitative research is aiming?Positivists aim at the underlying lawful reality of the human social world.However, research can, alternatively, take the view that this 'reality' is a myth, and that the social world is the multi-faceted outcome of the interaction of human agents; it is a world which does not have an unequivocal reality.In fact, the very features of qualitative research techniques which positivists regard as biases and are anxious to minimise (such as the fact that the data are joint products of the researcher and the researched), are viewed instead as reflecting essential truths of the human social world.Qualitative research is taken to be a form of human social interaction, necessarily carrying with it these characteristics.This is not a weakness: rather, positivistic efforts to turn the social interaction involved in qualitative research into a tool of measurement, aimed at teasing out the pattern of variables whose lawful relationships are supposed to make up the human social world, are guilty of a series of mistakes, the central one of which is indicated by Merleau-Ponty (1962, p. viii): 'I am not the outcome or the meeting-point of numerous causal agencies which determine my bodily or psychological make-up'.
The human social world, in this non-positivist understanding, is constructed by the interaction of conscious human agents (though some versions of discourse analysis do not press this point).Now, non-positivistic qualitative research itself contains a number of different strands and orientations.Such research sometimes aims at producing a description, sometimes seeks to generate an interpretation, and sometimes to uncover cultural discourses.(Of course this sample of research aims is not exhaustive, nor does one exclude another, nor is each aim homogenous, there are, for instance, varieties of approaches to interpretation.)Description.In qualitative research which focuses on description, the data do not represent the value of a variable (a cause or an effect), and are not to be understood to be an approximation to 'objective truth'.Rather, they may usually be taken to represent the research participants' truth, describing their life-world -their situation as they see it.My preference is to use the term 'life-world' rather than experience, because the term experience tends to lead us to think of some inner realm which is revealed by introspec-tion, whereas a focus on the life-world directs attention to the whole experienced situation of the person.
Because such an orientation owes a great deal to the philosophical school Husserl, the approach is often referred to as 'phenomenological'.However, a distinction has to be made between Husserl's (1970) philosophical work -which attempted to provide indubitable foundations for basic concepts of the various disciplines by a special process of reflection -and qualitative research based on his thought.Some have attempted to deny the legitimacy of empirical work based on phenomenology (which has no such epistemological aims).A full rebuttal of this line of attack appears elsewhere (Ashworth, 1996).Empirical research based on phenomenology is a descriptive study of selected phenomena of the life-world of research participants.In phenomenologically-based research, the aim of data collection is to provide as rich a set of descriptions of relevant aspects of the life-world of the research participants as possible.It is as if the interviewee were a researcher themselves -a participant observer investigating their own experience-and the interviewer is a facilitator, assisting in the process of reflection.
Thus, in the research on cheating mentioned above, interviews were carried out in a conversational mode.The interviewers approached the research with as general a set of topics to investigate as could be.The lack of closure in the topics raised is an attempt to set aside presuppositions in order to allow the interviewee's own account of the meaning in the life-world of cheating to emerge.
An important methodological principle of any research based on the attempt to describe the life-world of another person is that the investigator must begin by bracketing, or setting aside prior assumptions about, the nature of the experience being studied (Ashworth, 1996).Thus, in order to allow the life-world to emerge clearly, certain apparently harmless practices must be bracketed.
Firstly, science, in the sense of a body of 'known facts', is to be set aside in order to avoid inadvertently importing theories and findings which would distort the description of the life-world.Thus, in our work on the meaning of cheating for students, we specifically avoided drawing on psychological models such as those drawn from 'attribution theory' (for a review, see Hewstone, 1989) -which could well be regarded as relevant to the scientific explanation of our research participants' behaviour and experience.Since the focus is purely on their life-world, it would only be if perceptions of their situation which echoed the scientific concepts 'naturally' arose in the data that it would be permissible to include them in the description -and then their inclusion would be quite independent of the fact that there happened to be such notions in the scientific literature.
Secondly, the researcher is debarred from querying the validity of the life-world.External validity is irrelevant.The researcher must adopt no position on the correctness or falsity of the claims which are implicitly made by the research participant in the views and judgements intrinsic to their life-world.Data gathering should be seen as a process of discovery, concentrating in the first instance on each individual as a separate case, a possibly unique world.Indeed, we cannot even assume that the meaning of the 'same' situation is similar from different people.Subsequent analysis of successive cases may reveal general features of the situation.
It is not the place here to detail the procedures which can be used to assist in the analysis of qualitative data so as to generate a description of relevant aspects of the life-world.Generally, variants on the approach articulated by Giorgi (1985; also Wertz, 1983) are used.Key points to emphasise, however, are (a) the need for the researcher to continually bear in mind that it is the informant's meanings that are being sought, and (b) the withholding until late on in the analysis of the decision about whether there are themes in the data which are common from person to person.
Perhaps the most challenging set of ideas we had to bracket in the study of cheating was our identification with academia, which occasionally came under attack in various ways.For instance: • You could say that the tutors cheat when they mark.I mean, say they've got so many people fail and you know this is going to look bad on their statistics, "We've got far too many gone down the tube here, let's lift by 5%."It's manipulating the marks, isn't it?So people who'd scored quite well scored even better when the marks were put up and maybe some people who shouldn't have got through would have got through.Defensively, the researcher -being a member of the academic community and therefore somehow implicated-may well tend to try to put such statements in context or question their objectivity.But this would be a failure in maintaining the epoché.The quest is for the meaning of cheating in the life-world of the interviewee, and bracketing the issue of objectivity is of central methodological importance in opening out this reality.
As an instance of phenomenological analysis, take the interview material introduced earlier.With other evidence from the interview, this data built up to a general description of the way in which cheating appeared in the life-world of the individual -and in the case of cheating, the meaning may be rather specific to the individual: I mean, the actual exam at the end isn't a test of...I mean, it allows me to go further -it allows me to do whatever I want to do, so I've got to do, you know, work in that direction.But I don't think I would ... not unless I really thought I was going to fail and it wasn't a subject that I was planning to specialise in.It was, you know, just something I needed to get to that level.Because, if I thought I was failing on the subjects that I wanted to continue on, then I don't know whether I'd even be able -I'd want to carry on, because I'd want to be able to do well, without an incredible amount of problem, or I'd decide that it wasn't the course for me if I couldn't do it ... Yeah -like I'm prepared to help [interviewee's younger brother] in his GCSEs because I see it as just in order to get to something that he could do easily, you know, and sort of shine in -the GCSEs are just a way of testing that doesn't suit him, but is necessary and, you know, is below him-but is necessary for him to get somewhere where he, you know, and he can't do this, he has a problem with this method of, so yes, yeah, I'd be prepared to help him cheat.Like, helping him with assessments.
What is cheating in the life-world of this person?It is within the context of a view of a student career, with definite priorities and interests, matched by irrelevancies (not necessarily annoying ones, but not central to the personal project of the student).Assessments are hurdles to be cleared, but they can also indicate the extent to which one is on target, with sufficient mastery to continue on the chosen career, or the extent to which a re-think is necessary.The wisdom of cheating is to be judged in the light of its usefulness to this scenario.To cheat is a real option (the interviewee does not seem to have a pre-emptive moral objection to cheating as such).But it would be foolish to cheat if that means passing an exam which matters (i.e. one which assesses knowledge which is pre-requisite for later units which she has as central items in the career as currently envisaged).Cheating in assessments which are mere hurdles and the content is of no further significance would be acceptable.
The outcome of descriptive qualitative research is not generalisable laws.The positivist approach (in which a population, defined by certain quantitative parameters, is sampled, and then facts concerning the sample are found which can be generalised to the whole population), is an inappropriate model for such work.However, it can be found that the themes in a set of descriptions capture essential features which would be true of cheating (or whatever the phenomenon under review is) in any life-world.
Finally, it must be re-emphasised that phenomenological study of the life-world of the individual does not give rise to data that can be straightforwardly regarded as causal.The objective causes of cheating are not sought by this methodology, though, in interview, students themselves sometimes came up with reasons why cheating takes place.
-Going through a 'phase': When I was 14, I was going through a phase.I was angry with my parents, I was angry with the teachers, so every subject which annoyed me I didn't do any work, ... I used to come into the class with no knowledge ...And put the book on my lap and try to work.
-The inadequacy of the rules or the ease with which they can be evaded: • [In a 'seen exam'] there was a long list of rules you couldn't, of things you were supposed to do.But I mean there was so much else you could put into it that wasn't written in the rules.[The interviewee cheated by taking more than the permitted number of pages of notes into the exam.] • I suppose an exam is actually harder to cheat in -because you are being watched all the time ... whereas forms of continuous assessment where you can go away-so yes, it would be very easy to cheat, to go to somebody else, to go to an expert, to get a second opinion, to use uncredited references and that sort of thing...
-Personal inclination or 'personality': I mean sometimes I find it's quite fun, it's good to play with fire.
-The attraction of a payoff: I know somebody else who ... copied a large portion of someone else's results.He got away with it.He got his qualification and he's got no moral qualms about it at all.
-Failure to see the relevance of a particular subject: People just don't want to read about things that are uninteresting to them, I don't, you can't take it in ...You force yourself to read it ... yeah, I would say that it would probably force people to cheat more.
And so on, the instances could be multiplied.But these statements really must not be misconstrued: they must not be read as indicating 'causes' of cheating.They are the conditions of cheating within the interviewee's life-worldly experience.Possibly they could also be understood as interviewees' perceptions of people's reasons for cheating.Although some qualitative researchers can be seen to present such data as if it were evidence of causality in the positivist sense (I think 'grounded theory' is especially open to such abuse), this is wrong.Causality cannot be derived from work based on nonpositivist qualitative research.
Interpretation.Description of the life-world cannot avoid aspects of interpretation; despite trying to keep empathically in tune with the life-world of participants, the research necessarily highlights matters of 'relevance' to the interests of the research and de-emphasises links with other aspects of the individual's experience which they themselves might well have retained as being important.Also, the outcome of qualitative research is (one hopes) to present a description which is in some ways clearer and more open to understanding than the individuals would have been able to provide unassisted.
The broad tendency of thinking which goes under the name of hermeneutics regards interpretation as an aspect of the very mode of existence of the human being.So 'understanding', the effort to interpret or make sense, is a definitive feature of human existence.As the British psychologist Frederic Bartlett (1932) put it, human beings are motivated by an 'effort after meaning'; more portentously, Maurice Merleau-Ponty's (1962) view of the human condition has us 'condemned to meaning'.These claims underline the centrality of hermeneutics: the study of the basis of understanding or interpretation is effectively focused on a cardinal aspect of human existence.
A word of warning.Just as we had to draw a distinction between phenomenological philosophy and descriptive research, so it is important to say that there is a distinction necessary between the thinking of such people as Heidegger, and empirical research which adopts a hermeneutical style.In particular, empirical research does not mirror the anti-humanism of the later Heidegger.To give a flavour of the odd style of writing he adopts in trying to avoid any hint of person-centredness, he discusses the technological world, not as something people do or even as an ideology, but as a phase in the history of the self-revelation of Being.(See Heidegger, 1977).
We approach the world of our experience 'understandingly', then.We come to the research with primary 'lived' understanding of what it is we are intending to study.In interpretation, then, there are always presuppositions, which determine its direction: In interpreting, we do not, so to speak, throw a 'signification' over some naked thing which is present-at-hand, we do not stick a value on it; but when something within-the-world is encountered as such, the thing in question already has an involvement which is disclosed in our understanding of the world, and this involvement is one which gets laid out by the interpretation.(Heidegger, 1962, pp.190f) Interpretation depends on standpoint, and the meaning 'of something' has to be in terms of the relevance of the thing to the project of the interpreter, the researcher.
Building on this, Gadamer (1989) argues that the circle of understanding arises from, and returns to the 'cultural' standpoint of the interpreter.In saying this, Gadamer holds that 'prejudice' is an essential feature of understanding, since there is no approach to the material which one is attempting to address without some initial interpretative stance, what he refers to as the 'tradition'.Not only is this inevitable, but (Gadamer argues) it is thoroughly desirable, because it is only through approaching the new area with one's existing cultural standpoint and meeting it on these terms that the new area can speak to one's tradition and be genuinely informative.
The aim of becoming immersed in the world of experience that one is attempting to interpret is not merely difficult but undesirable.If it were possible, it would mean that the new area would not address our own original stance, enabling it to be affected by our discoveries within the others' world.
... [U]nderstanding is not based on transposing oneself into another person, on one person's immediate participation with another.To understand what a person says is ... to come to an understanding about the subject matter, not to get inside another person and relive his experiences ... (Gadamer, 1989, p. 383) Thus, a central feature of understanding is setting out from one's cultural standpoint and to interpret -to read, to learn -the world of the cultural other, returning to one's starting point enriched.The conclusion is clear: for the theory of interpretation the notion of a culture-free 'objectivity' is nonsense.The interpreter, both must and should start from the viewpoint of their own culture and tradition, and reach some kind of 'fusion of horizons' with the area under investigation, thereby enriching the initial understanding.It is not that objectivity is a hard quality to achieve, but that it is impossible -not least, because there is no unequivocal, culture-free reality to be sought.
This approach to qualitative research leads to a rather different view of 'data' than descriptive, phenomenological work.We have noted that description is not devoid of interpretation; in the view of hermeneutics it cannot be.Description is already interpretative.The grasp that the person has of their own experience is mediated by language and culture, for instance.No-one can address their experience directly, untrammelled by the categories by which, in fact, the world is lived and known.
This insight does not intend that we should think in terms of two separate things: raw experience and interpretation.No, experience is already shot through with interpretation.We live interpretatively; it is part of the makeup of human beings.
For qualitative research, then, the hermeneutic approach provides a new view of the meaning of data.Interview data, for example, (a) is the record of a joint process by which two individuals have negotiated a 'fusion of horizons', or each came to an appreciation of a way in which their initial fore-understanding was transformed by bringing it up against the others' conversation, and (b) is to be understood as an account of the way certain situations are interpreted or understood (without any notion that, behind such understandings, fluid and of varying certainty though they may be, there is any 'true' reality which could be uncovered with appropriate questioning).
To return to the interview material then, we find that it is open to a variety of interpretations.
I mean, the actual exam at the end isn't a test of...I mean, it allows me to go further -it allows me to do whatever I want to do, so I've got to do, you know, work in that direction.But I don't think I would ... not unless I really thought I was going to fail and it wasn't a subject that I was planning to specialise in.It was, you know, just something I needed to get to that level.Because, if I thought I was failing on the subjects that I wanted to continue on, then I don't know whether I'd even be able -I'd want to carry on, because I'd want to be able to do well, without an incredible amount of problem, or I'd decide that it wasn't the course for me if I couldn't do it ... Yeah -like I'm prepared to help [interviewee's younger brother] in his GCSEs because I see it as just in order to get to something that he could do easily, you know, and sort of shine in -the GCSEs are just a way of testing that doesn't suit him, but is necessary and, you know, is below him-but is necessary for him to get somewhere where he, you know, and he can't do this, he has a problem with this method of, so yes, yeah, I'd be prepared to help him cheat.Like, helping him with assessments.
In reading the transcript of this interview, one interpreter may be struck by the individualism of the speaker (the passage does not hint at the idea of other examinees or the notion of 'fairness').Another interpreter might notice the motivational thrust ('because I'd want to do well'; 'something he could do easily and ... shine'; 'necessary for him to get somewhere').Both these interpretations are justifiable.And it is worth mentioning the obvious fact that the interviewees are each interpreters of cheating, and come to different conclusions.For example, the morality of cheating was mentioned in very different terms: • Although cheating can get people into trouble it can help some people, it makes them feel they can do something.
• It is inexcusable.
• I could take an ethical stance here -it is wrong to lie... Gadamerian hermeneutics leads to a very liberal view of the meaning of qualitative data: its meaning for the interpreter.This is not to say that any meaning will do and that the whole thing is entirely subjective (Fay, 1996).Rather, the statements of interviewees are immensely rich and afford a wide range of justifiable interpretations.
Discourse analysis.Discourse analysis can be seen as an extension of hermeneutics.Here, data are interpreted as expressing one or more culturally-available discourses, for which the research participants are the channel through which the discourses flow, rather than the originators of personal constructions of the world (though the issue of agency is controversial).In discourse analysis (Hollway, 1989;Potter and Wetherell, 1987), social and mental life is taken to be embedded in the local cultural and linguistic forms available to participants.It is the intention to discover in the participants' accounts the implicit or explicit structures of meaning, or latent themes, which are understood to be both constitutive of and constituted by the wider social reality.
Here, again, the meaning of data is transformed by the approach which is adopted to qualitative research.In the interview material on which we have been relying for illustration, there are certainly discursive themes worthy of note.The most obvious one is a moral discourse.Individual variants on moral appraisal of cheating have already been discussed.In general, it seems that student experience is informed by a morality of fellow-feeling (which may contrast with the teacher's more academic values or organisational orientation).This being the case, cheating is seen as quite justifiable -ethically justifiable-if it seems to reduce the misery of a fellow student in dire circumstances; issues of the validity of assessment and the meaning of the award are not part of this discourse.
Another culturally-available discourse which appeared in the interviews surprisingly was to do with working life.For instance: • In future life when you are looking for a job, if you've got a false degree then you go for a job and you can't do it, then you're not just fooling yourself because you'll make yourself embarrassed because you're not going to be able to do the work ... It's not going to prove any benefit to the person who is cheating in the long run, is it?
• My personal attitudes to cheating are that it shouldn't be done.Within nursing for example, the end result is whether or not I am fit, and able, to practice.If I have cheated then I would have a feeling that I am not able to practice, so potentially I'm leaving myself open to litigation, potentially I'm doing myself out of a job.
It may be that the emphasis on achievement and progress noted in our earlier interview material is also to be regarded as part of the 'working life' discourse.
In the context of the analysis of discourse, it is not clear that concepts such as reliability and validity have any application.Rather, the data themselves must be presented, together with the interpretation, and the reader must be left to judge the cogency of the claim that we have here an instance of a particular discursive tendency.
In what has been said so far, discourse analysis has been presented in terms of the interpretation of the content of talk in terms of the cultural resources which are drawn on.Potter and Wetherell (1995) make it clear that they understand the approach to entail much more than this -not just uncovering the resources which are available, but also the way these resources are utilised-discourse practices.Entailed in these practices is the construction by people of descriptions of the world which are then taken to have some solidity or pragmatic truth.Interpersonal processes are involved in all this, of course, and discourses are employed rhetorically in argumentation and negotiation, and ways of associating or dissociating people from discourses -and thereby constructing moral identities-are employed.But in all this, the 'material' used is discursive and the products of discourse practices are themselves discursive.
Discourse analysis moves qualitative research firmly away from concern with the individual as such to the view that data (though emanating from a particular person) is to be seen in the context of the culture as a whole.Discourse analysis is in this way opposed to any dualism of individual and society.

CONCLUSION
It has been the aim of this paper to emphasise the way in which 'qualitative research' ramifies into a number of forms of investigation, each with characteristic ways of viewing the meaning of the data, and ways of handling it.If the data are seen as pointing to causal variables which would account for the phenomenon in question, the issues to be addressed are very different to those which arise when data is taken to be descriptive of the life-world experience of the research participant.Similarly the postmodern interpretive freedom which a hermeneutic understanding of the data allows is seriously modified if the source of interview material is understood to be 'socially available discourse' rather than the conscious awareness of the person who utters that material.
Without an awareness of these matters, researchers who believe that they are using qualitative techniques find themselves in difficulty -making causal claims where none are justified; anxiously seeking for criteria which would validate their findings when there could be no such criteria and it is misunderstanding of the nature of the data to search for them.
. Peter Ashworth, Philip Bannister and Pauline Thorne, with Student members of MA Course Unit, Qualitative Research Methods, Autumn 1995 (Sheffield Hallam University).
I mean, the actual exam at the end isn't a test of...I mean, it allows me to go further -it allows me to do whatever I want to do, so I've got to do, you know, work in that direction.But I don't think I would ... not unless I really thought I was going to fail and it wasn't a subject that I was planning to specialise in.It was, you know, just something I needed to get to that level.Because, if I thought I was failing on the subjects that I wanted to continue on, then I don't know whether I'd even be able -I'd want to carry on, because I'd want to be able to do well, without an incredible amount of problem, or I'd decide that it wasn't the course for me if I couldn't do it ... Yeah -like I'm prepared to help [interviewee's younger brother] in his GCSEs because I see it as just in order to get to something that he could do easily, you know, and sort of shine inthe GCSEs are just a way of testing that doesn't suit him, but is necessary and, you know, is below him-but is necessary for him to get somewhere where he, you know, and he can't do this, he has a problem with this method of, so yes, yeah, I'd be prepared to help him cheat.Like, helping him with assessments.