You are on page 1of 9

Grounded theory and qualitative data analysis

The comments included in this document first appeared in the HRMAS Newsletters (Numbers 9, 10 and 11- distributed in August 1998). They are brought together here so the whole discussion can be access in one document. (This document was compiled on 26 August 1998).

Contents
1. Introduction: "Grounded theory" in the analysis of qualitative data. David Thomas 2. Comment 1: Dr Gillian White 3. Comment 2: Dr Laurence Chalip 4. Comment 3: Sarah Marshall 5. Comment 4: Dr Gillian White 6. Comment 5: Dr Laurence Chalip 7. Resources for grounded theory approaches

INTRODUCTION: "GROUNDED THEORY" IN THE ANALYSIS OF QUALITATIVE DATA. David Thomas


In research proposals, some researchers make reference to the use of a "grounded theory" approach in the proposed analysis of qualitative data. A critique of the "inappropriate" use of the term "grounded theory" is given below by Dr Gillian White (Nursing & Midwifery, Massey University, Albany). Gills response was in reply to the following comments. Given that you refer a "grounded theory" analysis as necessarily using the constant comparison method, what term would be used to refer to an inductive approach (that starts with the data and develops categories from the raw data) which does not specifically use the constant comparison method? They only alternative I can think of is an "inductive data analysis" I suspect this latter approach is quite common, but not usually given a distinctive label.

COMMENT 1: Dr Gillian White


I believe that we are "bastardising" research methodologies these days by not being true to either (a) the method as originally developed or (b) the adaptations as disciples of the method have encouraged the method to evolve, and acknowledging these adaptations. The constant comparative method, for example, allows for the development of categories often through theoretical sampling, which is a significant feature of grounded theory. In particular I am concerned to see researchers loosely referring to grounded theory data analysis but paying no attention at all to the theoretical and therefore philosophical underpinnings of the method. To be using grounded theory or phenomenology or any other qualitative method makes a statement that you are coming from a specific belief system. To that extent I think your comments are quite right. People do use a simple inductive approach and call it something else. As I see it : hematic analysis breaks text down into themes which are then explored and interpreted. Content analysis examines what the words (concepts) used mean. A Grounded theory analytic approach takes a line by line analysis and constantly compares the data searching for themes/categories. This usually entails further research into the meaning of the developing categories by further sampling from appropriate participants or other data sources. A phenomenological approach identifies themes which derive from the participants' stories in order to answer questions such as "how was it for you?" Perhaps it is timely to arrive at some consensus about the use and abuse of qualitative methods and methodologies. Quite honestly examining transcripts for themes or

categories and interpreting them is how most qualitative data is analysed. If we cannot make distinctions about certain methodologies then why bother to have them at all!! For me the most important aspect is to adequately address the research question by the most appropriate means and declare one's frame of reference (e.g. feminist, social interactionist etc). Dr Gillian White School of Health Sciences (Nursing and Midwifery) Massey University @ Albany G.White@massey.ac.nz

COMMENT 2: Dr Laurence Chalip (Griffith University, Gold Coast)


Comment 1 above made the assertion that researchers choosing to compare case studies or to analyse data phenomenologically are making particular assumptions about the nature of social reality. This assertion repeats a common misunderstanding of the philosophical roots of debates surrounding choice of method in the social sciences. The misunderstanding has been to confuse ontological assumptions (i.e., the presuppositions we make about the fundamental character of social life) with epistemological preferences (i.e., the methodologies one chooses to learn about social life). Although some researchers who proclaim particular ontologies also advocate particular epistemologies, the two are neither isomorphic nor necessarily codeterminative. Those who have advocated the use of various qualitative methodologies (particularly inductive methodologies, such as those associated with formulation of grounded theory) have sought to advance the use of qualitative methods by taking arms against the orthodoxies which became associated with Fisherian (i.e., statistical and, typically, deductive) approaches to social research. In so doing, advocates have typically argued that social life is based around rules, that realities are socially constructed, and/or that social actions have meanings (and, perhaps, outcomes) which must themselves be interpreted. They go on to argue that, therefore, our methods need to be capable of identifying rules, locating social constructions, or interpreting social meanings. They then suggest that qualitative (typically inductive) methods are best able to accomplish those ends. So far, no problem! All this simply says that a particular set of ontological assumptions may recommend a particular set of epistemological preferences. Notice, however, that it does not say (because it cannot) that a particular epistemological preference necessitates a particular set of ontological presuppositions. Nor does it say (because it cannot) that a particular set of ontological presuppositions can only permit a particular epistemology. In other words there can be no claim that there is a determinative link between a particular set of ontological presuppositions and an epistemological preference. This should come as no surprise to those familiar with the history of method in the social sciences. Analytic induction -- from which Glaser and Strauss's work on grounded theory derives -- was commonly used in the social sciences (particularly

sociology) prior to the era of Fisherian orthodoxy (and is still used today). Analytic induction is also the epistemological root for much of Yin's work on comparison of cases, as well as Campbell's work on finding "degrees of freedom" in case studies. Analytic induction is typically qualitative; it makes use of comparisons (typically of cases); it often makes use of techniques which share some affinity with phenomenology and hermeneutics (though researchers using analytic induction don't use those labels). The important point here is that researchers who used analytic induction typically worked in the positivist tradition insomuch as they assumed that the social scientist could find an Archimedean point from which to base an objective analysis. In other words, there is a recent history of extensive inductive social research which has made use of qualitative methods and case comparisons while never wedding itself to the ontologies now commonly associated with phenomenology, hermeneutics, post-modernism, or critical theory (to name the most obvious modern advocates for similar techniques). The point is that the choice of method does not mandate a particular set of ontological assumptions. Similarly, the fact that one takes the view that social reality may be socially constructed (or that one wants to work inductively) does not rule quantitative procedures out of court. There is an increasing volume of work in that works dialectically between quantitative data and theory, as well as work that seeks to incorporate quantitative data and techniques into the task of building theory interpretively and inductively. For example, much recent work into subcultures of consumption began using standard qualitative, inductive procedures. However, elements of that work now routinely incorporate various quantitative analyses to further explorations of subculture, including its variations and its impacts. Nevertheless, that work is self-consciously rooted in social constructionist views of subcultural life, and has borrowed unabashedly from hermeneutics and critical theory. This is not to say that one should not be concerned about the assumptions one is making about the world or about knowledge. It is quite clearly useful to be clear about one's ontological assumptions and to understand how those may frame one's epistemology. That is simply a matter of doing good social science. However, when we shift from being clear about assumptions to any argument claiming that our ontological assumptions dictate our epistemologies (or vice versa), then we are creating the kind of orthodoxy which straightjackets social science, and impedes its progress because it establishes unnecessary incommensurabilities among the many communities of social science discourse. Science does not advance through methodological proscription. It advances as a consequence of meaningful questions, empirical work to answer those questions, and a thoughtful analysis of the assumptive underpinnings of our work. In other words, doing robust social science requires that a disputatious community of social scientists attack its questions analytically, applying any and all methods that may be useful to find answers. When we make any claim for an essential and necessary link between a particular ontology and a particular epistemology, we disallow the very basis for doing social science. Yes, let's be clear about our assumptions. But, no, let's not confound methodology with paradigm. Laurence Chalip School of Marketing and Management Griffith University PMB 50 Gold Coast Mail Centre Queensland 9726 AUSTRALIA Email: L.Chalip@gu.edu.au

COMMENT 3: From Sarah Marshall, Nursing, Massey University.


Because Grounded Theory methodology is underpinned by a certain perspective from the field of sociology (Symbolic Interactionism) it's comes complete with it's own jargon. The constant comparative method is only the constant comparative method when it is used to generate a Grounded Theory. In practice the line by line coding may be exactly the same as a general inductive approach to data analysis. However, in a true Grounded Theory the data analysis would be validated as the constant comparative method if the yield of the induction was a theoretical explanation of behaviour. The "codes" and "categories" must weave together conceptually to tell a story about human behaviour, rather than remaining suspended as a description of an event or feeling. The constant comparative method that does not produce a theory, is in fact an anomaly. Gill introduces the debate on method versus methodology. It seems to me that proponents of method seem to approach the data with an aim to address the question, as Gill highlighted. Those who prefer methodology seem to see the question within a certain perspective and seek to resolve it accordingly. Well, that's my hypothesis, though I would hesitate to statistically analyse it! I found the book commentaries on "Resources for Grounded Theory Approaches" by Mary Pat Mann (1995) entertaining. Scholars of Grounded Theory will know that following the foundational works on Grounded Theory (Discovery etc), Glaser and Strauss differed in their approaches to the methodology. In this way it may be wise for new readers to select one author or the other, rather than attempting to learn the basics from both of them. Undoubtedly, Strauss offers to most detailed "how to" instructions. Personally I have gritted my teeth and stuck with Glaser because under Strauss' instruction I found I was deferring to description rather than conceptual analysis. Having said that, it is very difficult to learn grounded theory through books alone. Generally, you rely on wisdom from your supervisor! I am interested in exploring how Grounded Theory methodology might be used for evaluation purposes. I would appreciate some information on the fundamentals of evaluation research. Sarah Marshall Master of Philosophy (Nursing) candidate, Massey University.

COMMENT 4: Dr Gillian White, School of Health Sciences (Nursing and Midwifery), Massey University @ Albany
First it was very gratifying that the previous newsletter provoked so many responses. I too have received a couple of thank yous. Dr Chalip's thoughtful arguments have served to open up the topic for debate which is a wonderful use of the HRMAS newsletter. I totally agree with his statement that "science does not advance through methodological proscription" but that it advances through "meaningful questions" and "empirical work to answer those questions." Indeed that was my message. Dr Chalip's challenge, however, that a basic misunderstanding is confusing ontological assumptions (described as presuppositions about the fundamental character of social life) with epistemological preferences (described as the methodologies one chooses to learn about social life) actually serves to muddy the waters. Ontology is the study of being (i.e. those deep underlying structures while epistemology is that branch of philosophy that considers the history of knowledge related to the origin, nature, method and limitations of knowledge development). In asserting, therefore, that the use of specific qualitative methods makes a statement that one is coming from a certain belief system I am not saying that researchers are making an assumption about the nature of social reality. But I am saying that the method selected derives from a philosophical perspective and that researchers, if using a specific method, need to be aware that they are the agents of the data collection and should be locating themselves within the research, particularly if qualitative in design. For example, Dr Chalip, himself, does a very good job about locating his own position. For myself, being a very eclectic researcher, I will be guided by the question or area of interest to a research design that will help me answer the question or advance knowledge about the area of interest and seek an appropriate method to do so. If the question or area of interest derives from a phenomenological 'need to know' e.g. how is it for you? then I would choose to research from a phenomenological perspective. Even then I would have to declare which School of phenomenological thought I am favouring in order to provide my audience with a clear audit trail about how I arrived at my interpretations (i.e. what the ontological assumptions are). It is possible that the findings of this initial research would raise further questions which are better addressed through utilising a different research design (method). If that was through statistical analysis the I would now be declaring that there is value in a deductive approach and I would need to remain true to the assumptions underlying the statistical tests that I used. I would like to ask Dr Chalip why he believes that a particular epistemological preference cannot necessitate a particular set of ontological presuppositions. If one is considering epistemology as the history of knowledge development then surely one is considering the ontological presuppositions. If one is simply using a research method

without consideration of the assumptions underlying the method then surely this raises an issue of trustworthiness (in the sense of reliability and validity). It is easier to accept his argument that a particular set of ontological presuppositions cannot permit only one particular epistemology. If, for example, the presupposition is one of "oppression" then the method one chooses may very well have a feminist or cultural perspective but the research method could be one of several. Which brings me back to the notion that the selected method should be able to address the question or topic of interest but that the assumptions underpinning the method should be made explicit. The link is not determinative or isomorphic on the part of the researcher but on the part of the method. I think Dr Chalip has misread the initial comment as he appears to be substituting case comparisons (which were not mentioned) with constant comparative analysis. Glaser's work on grounded theory utilises a process in which any kind of data is permissible and is thus not necessarily "typically qualitative." It is therefore my argument that if researchers use analytic induction (without the labels phenomenology or hermeneutics (to quote Dr Chalip's examples), then they are using analytic induction. If, however, they are using analytic induction (for example, using a phenomenological or hermeneutic approach) then they are making a statement about the philosophical underpinnnings of their analysis. On taking the time to read and reread Dr Chalip's comments I have the impression that we are very nearly agreeing. It is timely that these issues are discussed. To be very practical, an issue for NZ researchers (and I am sure for colleagues elsewhere) is in being able to articulate, for members of granting bodies and ethics committees, exactly what we are doing when deviating from the double-blind randomised controlled 'gold standard' of the scientific method. Thank you Dr Chalip for raising issues for debate. Dr Gillian White RM MTD BEd Dip.Soc.Sci (Psych) MA (Hons) PhD Senior Lecturer/Albany Group Co-ordinator School of Health Sciences (Nursing and Midwifery) Massey University @ Albany Private Bag 102904, North Shore MSC Auckland, New Zealand G.White@massey.ac.nz

COMMENT 5: Dr Laurence Chalip, Griffith University, Gold Coast


I agree with Dr White that we are not very far apart in the fundamentals of our argument. She asks why I am arguing that a particular epistemological decision can't necessitate a particular ontological position. She points to epistemology as being concerned with the history of knowledge when formulating his query. However, epistemology is not only about the history of knowledge. It is also about the nature, the extent and the veracity of knowledge. Among its fundamental concerns are questions like: How is knowledge formulated? What are the differences among concepts? What part does reason play in knowledge? How necessary is sense experience for knowledge? These become implicated as an issue in the matter of methodology. It is certainly the case that a particular ontology may yield particular

epistemological preferences. This simply says that someone who believes certain things to be true about the world also believes that there are particular procedures which are best for learning about that world. However, when someone chooses to use a particular method -- even one that has been advocated by persons holding to a particular set of ontological presuppositions -- he or she may not necessarily have bought into those presuppositions. One reason is that the data thus generated must also be interpreted. It is perfectly feasible to interpret data obtained via particular methods using lenses (i.e., ontological presuppositions) that are dissimilar from those who advocate (or even invented) those methods. Indeed, that is a pivotal basis for scholarly disputation. There are several obvious reasons that such disputation is at the core of social science, regardless of paradigm. However, it is sometimes less frequently appreciated that there is a more fundamental reason to foster methodological diversity, even within paradigm. The primary advantage, as Campbell and Fiske demonstrated long ago is that findings using any one methodology may incorporate some artifact that is method dependent. Although Campbell and Fiske chose to reference forms of measurement, Wimsatt has since shown that the principle applies more generally as a requirement for what he calls "robust" knowledge in the social sciences. Laurence Chalip School of Marketing and Management Griffith University PMB 50 Gold Coast Mail Centre Queensland 9726 AUSTRALIA L.Chalip@gu.edu.au

RESOURCES FOR GROUNDED THEORY APPROACHES


(Commentaries by Mary Pat Mann, 26 Jan 1995) <45000bl14@mmail.oucom.ohiou.edu > Anselm Strauss & Juliet Corbin (1990). Basics of Qualitative Research. Newbury Park: Sage. This book written explicitly as a text on grounded theory methods for graduate level or beginning researchers. It s clearly written and explains the jargon, but reads like a textbook - a little dry. The preface says this books spells out the procedures and techniques in greatest detail and in the step-by-step fashion that is useful for learning qualitative analysis. Anselm Strauss (1987). Qualitative Analysis for Social Scientists. Cambridge University Press. Strauss considers this a more advanced text than the 1990 book described above. It does have much more detail (and its much longer), but the extended examples make reading it much more enjoyable. Barney Glaser, (1978). Theoretical Sensitivity. Sociology Press Glaser, in the forward, said it was written to extend the original work, Discovery of Grounded Theory, by focusing explicitly on techniques to increase theoretical sensitivity. It certainly reflects the times in which it was written: Glaser describes the excitement of generating theory in extended analysis sessions as a drugless trip. The now-dated jargon gets in the way a bit, but I found some of the suggestions helpful and its fun to read.

Barney Glaser and Anselm Strauss (1967). Discovery of Grounded Theory. Aldine. This is the classic - in every sense of the term. With all the interest in qualitative research lately, you might think something written 25 years ago would be hopelessly dated. In fact, all of the issues and techniques are there, just as if the authors had read the latest articles before they sat down to write. Since this book introduced grounded theory, there is more exposition and explanation. Its a good read, as long as you keep in mind that it was written for sociologists.

You might also like