Ólavur Christiansen, Ph.D.
Two “hallmarks” explain why and how Glaser’s GT method is one of a kind and meaningless for research that is conventionally pre-framed:
(I) The finding of the core variable is the first stage of the research and it is accomplished by the systematic treatment of the data. The core variable sums up and explains the main concern (and its recurrent solution) of those being studied, and most of the variation in the data. The study is then delimited to concepts related to the core variable. This is equivalent to the finding of the research problem.
(II) During the data work, the researcher suspends his/her preconceptions and prior knowledge about the area of research, and trusts in emergence of concepts from the data. Its purpose is avoidance of preconceived academic interest concerns with, instead, sole focus on the participant’s substantive interest concerns and their solutions, and the triggering of intuitive insights during states of not knowing. One way to transcend preconceptions is to study unrelated literature in order to identify latent behavior patterns and to increase the researcher’s familiarity with the full range of possible theoretical codes.
There is a general and increasing bewilderment in the literature regarding the concept of “grounded theory”, sometimes abridged as “GT” (Bryant & Charmaz 2007, Kelle 2007, Strubing 2007, Alvesson & Karreman 2007, Charmaz 2006, Douglas 2006, Clarke 2005, Suddaby 2005, Gephard & Rynes 2004, Alvesson & Skoldberg 2000:12-36, Moustakas 1994:4-7). This confusion also impinges on the understanding of other theory generating methodologies. (Alvesson op.cit., Moustakas op.cit.). There seems to be blindness to the fact that the grounded theory term is seriously misleading, because GT has become a name, not for one method, but for an array of very different research approaches, among which Glaser’s “orthodox” or “classic” GT is fundamentally different from the rest. Yet, with a few exceptions (Holton 2007, Hartman 2001, Glaser 1998:5), authors have failed to explain the
original raison d’être for the emergence of GT as a predominantly inductive (yet, also deductive and intuitive) theory generating method. To understand this original raison d’être is of course important, as well as the fact that Glaser, on this basis, is the originator of GT as a theory discovery method (Glaser 2006:3, 1998:21-34). To misrepresent a methodology because it is “very different” is inappropriate; neither is it useful. Most likely, we need methodological diversity to achieve theoretical and applied advancements in complex fields of enquiry, e.g., business and management.
This article is based on the assumption that different research methods are just different. This means that the question about better or worse regarding methods in general, or regarding ontological and epistemological positions per se, is assumed pointless. It is assumed that this question only gains relevance when specific research purposes and frameworks are taken into account. While considering these contexts, openness and tolerance with regard to methodological diversity may provide better opportunities than forced restriction into methodological uniformity.
The purpose of this article is to try to clarify why it may be justified to chose other methods or GT-versions than Glaser’s, when the choice of Glaser’s GT would be meaningless. I will try to achieve this purpose mainly by highlighting a few characteristics of Glaser’s GT. I have tried to select those characteristics that seem to be mostly misunderstood or ignored in literature that explains and compares different GT-versions. GT is not easy (Suddaby 2005). Before use, Glaser’s prescribed research procedures should be studied in Glaser’s own writings (1964, 1978, 1992, 1998, 2001, 2003, 2005, 2007, 2008).
Paul F. Lazarsfeld’s inductive quantitative methodology, as well as other ideas that Glaser received as a student of Lazarsfeld, are the roots of Glaser’s GT. These are procedures and notions such as latent pattern analysis, core variable analysis, explanation of variation, use of crude indicators of concepts, naming of stable patterns as concepts, reliance on the interchangeability (not summation) of indicators of concepts, “explication de text”, the importance of understanding the meaning behind human behavior, and the pivotal importance of the participant’s main concern. (Glaser 1998:21-34, 2006:3-6,
2008:1-18).
The use of crude indicators means a relaxation of the required accuracy in quantitative verification-testing studies. This relaxation facilitates discovery of theory. Accurate findings and final verification or testing is not an issue. Another justification for this relaxation was Lazarsfeld’s discovery that crude indexes gave the same finding as elegant, perfected indexes based on latent structure analysis. Thus, Glaser considered the latter expensive and a waste of time (Glaser 2006:4).
However, Lazarsfeld had missed one procedure that became fundamental in Glaser’s GT. This is the constant comparative analysis that is applied during coding (categorizing, conceptualizing) of the data. The fractured data slices or conceptual indicators are constantly compared. Consequently, the naming of the concepts is recurrently refitted to what the data purport to express about the main concern and its recurrent solution. This refitting continues until sufficient interchangeable indicators have saturated each concept, and each concept’s properties are also saturated. This procedure facilitates the finding of stable latent patterns that explain variation, especially variation in qualitative data (variation within and between categories). (Ibid., Glaser 1978:149-155, 1998:139-152).
Glaser’s first GT study was based on quantitative data (Glaser 1964). Later, he combined, refined and modified these procedures into a methodology that also could handle qualitative data. In accomplishing this, Glaser’s cooperation with Anselm Strauss was instrumental. Strauss was trained in qualitative methodology. Glaser and Strauss carried out the first GT study that used qualitative data (Glaser & Strauss 1965). Two years later, Glaser and Strauss explained the methodology they had been using (Glaser & Strauss 1967). Subsequently, Glaser has maintained and elaborated the method through numerous books (Glaser, op.cit.) and by his Grounded Theory Institute.
Thus, Glaser’s GT method is essentially a “concept-indicator model” with quantitative-methodological origin. It can use any type of data, quantitative and/or qualitative. However, because qualitative data are easier to collect, the method has so far almost exclusively been used on qualitative data. Furthermore, because much of human behavior and its variation are unquantifiable, quantitative data may sometimes be insufficient for a study of behavior. When this is the case, and when the purpose is discovery of the concealed and unanticipated (not minute accuracy as in verification/testing studies), statistical significance may be very different from theoretical significance, and irrelevant.
The theory that is generated directly from data has to be a theory about the human behavior patterns within the chosen field of research. The theory has to explain conceptually how the studied participants, by their behavior, recurrently resolve a main concern. Glaser’s method is predominantly inductive, but not inductive in a traditional sense. It is both more and less assumption-free than traditional induction (Hartman 2001:36-37,41-42).
Glaser’s GT is more assumption-free than traditional induction because the researcher, in his/her constant comparison of the data, deliberately has to suspend his/her prior theoretical knowledge and preconceptions about the chosen field of research. This suspension does not apply for knowledge about the full range of possible theoretical codes. (These will be explained later). However, assumptions that are merely inferred from extant theory do not count as assumptions. Only what the data relate will count as assumptions (Ibid.). As in solving a puzzle, the analyst transcends or minimizes speculative logical interpretation in favor of straightforward categorization (coding, etc.) of data slices according to fit of patterns.
Glaser’s GT is less assumption-free than traditional induction because the researcher has to use his/her own data-generated ideas and to base his/her continued inquiry on these ideas. Assumptions that are recurrently sustained by data indicators do count as assumptions. The researcher has to base his/her continued inquiry on these assumptions, and they may be modified by new indications that emerge during the continued use of the research procedures (Ibid.).
This means that the researcher’s perpetual closeness to his/her data is a paramount requirement. This also means that the researcher refrains from any predeterminations (preconceptions) before and during the research process – or from “knowing” before he/she is in a position to “know”. This disparate application of induction explains how Glaser’s GT is distinct from other versions of GT as well as other inductive research approaches. However, Glaser’s GT is not entirely inductive. Without some reason and intuition in addition to empiricism, it is not possible to generate a theory from data. The role of intuition and deduction will be explained later.
The units of analysis are behavior incidents within the chosen field of study – not people or organizational units. Thus, these units of analysis comprise the data. Usually, these data are generated from a sample of organizational units (e.g., businesses). This sampling is theoretical, not statistically representative. In the beginning phase of theoretical sampling, the researcher selects units with maximum differences. Then the analyzing (i.e., coding by constantly comparing, conceptual sorting) and synthesizing (e.g., memo-writing, sorting) of the data determines what unit to sample next for comparison of more incidents. These may, or may not, indicate stable latent patterns or concepts (categories), or provide indicators for the elaboration of emerging concepts.
When data is generated by interview, preconceived questions should never be asked, and pre-structured interviewer guides must never be used. The interviewee should just be encouraged to talk freely and confidentially about how they continually solve their main problems. (Taping and intensive note taking during interviews will destroy confidentiality). When stable patterns have emerged, the interviewer can ask more “empirically grounded” questions. The researcher has to acknowledge that the subjects know much better their work-procedures than the researcher does. With reference to organizational units, time, place and people, they can easily describe it. However, because of this “engrossment” in their activities, they are usually unable to conceptualize and conceptually explain their recurrent patterns of behavior. This is the profession of the researcher. This includes the task of building a theory that transcends the boundaries of time, place and people.
The purpose is to suggest a set of interrelated conceptual hypotheses about the participant’s main concern and its recurrent solution. This suggestion is based on their “grounding in data” and not on any final verification. The task to verify deduced, preconceived, or grounded hypotheses by testing them is left to others that use other methods. Instead of testing, the method facilitates recurrent modification of hypotheses in order to come closer to what actually goes on within the chosen field of study with regard to the participant’s recurrent solving of their main concern.
There is no ultimate finality about such a generated theory. No claim is made that the generated theory will reach or transcend any ultimate truth line. However, due to its grounding, it can be trusted to approach what actually goes on. The theory goes as far as it goes in explaining behavior, and further research can bring it closer to the truth line.
A researcher may want continue from where another researcher has ended in generating a theory, or to use the theory in practice, e.g., to bring about some changes. Without some trust in the credibility of the earlier work, such a continuation or practical use would hardly have any purpose. In this case, it is not possible to start “empty” or “non-preconceived”. Nevertheless, the second researcher has to “try out” the fit of the earlier researcher’s concepts to his/her own data. These data may, or may not, originate from the same substantive area. E.g., refitting of some conceptual names may become relevant, and they may, or may not, amount to a substantial modification of the theory. In the later stages of the research, the ordinary GT procedures should apply. This means that the later researcher minimizes his/her preconceptions while he/she refines, verifies and/or modifies the earlier researcher’s theory beyond its original limits. (Glaser & Holton, 2005: 17-18).
Later, Strauss and Corbin (1990, 1998) have prescribed a different set of GT procedures. The roots of this different GT version are qualitative methodologies, and it has an inseparable attachment to the theoretical perspective of symbolic interactionism. Charmaz (2006) has also prescribed her own version of GT. This GT version has an inseparable attachment to the theoretical perspective of social constructivism. Clarke (2005) has also prescribed her own version of GT that is based on post-modernist theoretical perspectives and situational and discourse analysis. These other versions of GT are, due to their apriori attachment to and favoring of a specific theoretical perspective, incompatible with Glaser’s GT. The procedures of these other GT versions are by naming similar to Glaser’s, but they are used very differently. To mix procedures of these different versions of GT with Glaser’s GT procedures implies abandonment of Glaser’s GT. Glaser’s GT has to be considered as a fundamentally different methodology. (Glaser 1992, Martin 2006b). The following subsections will make this clearer.
Pre-framed professional interest concerns of a researcher or a research community are, e.g., shared ontological and epistemological assumptions and the favoring of some particular theoretical perspectives or some paradigm (Kuhn 1996) or “research program” (Lakatos 1970). Preconceived professional interest concerns unavoidably mean that the researcher preconceives the significance of particular concepts, hypotheses as well as of data. He/she has to view his/her data through this particular “lens”.
It is hardly a controversial statement that this kind of pre-framing overrules research. Such overruling applies for quantitative hypothetical-deductive studies and for qualitative studies (phenomenology, ethno-methodology, hermeneutics, postmodernism, social construction, etc.), as well as for Strauss-Corbin’s, Charmaz’ and Clarke’s versions of GT. These favor apriori a particular theoretical perspective. Glaser’s GT is very different. It entirely rejects these kinds of preconceived professional interest concerns in favor of “the substantive interest concern of those being studied” – or the explanation of the main concern and its recurrent solution of those being studied. The researcher’s professional interest concerns have to merge with this latter.
When Glaser’s GT is used, the first and most pivotal stage is to find empirically, as a concept, the main concern of those being studied, i.e., the most important and problematic for these people, and that drives and directs their behavior. This is the concept that best sums up and explains the essence of what is going on in the data – what the data relate about of the essence of relevance for those being studied. This concept then becomes the core variable (or core category) of the emergent theory. The remaining theory generation is delimited to concepts that are most related to the core variable. During the remaining theory generation, the researcher has to view his/her data through this “lens” – through the “lens” of the core variable.
This fundamentally different kind of overruling of the research is the original raison d’être for the emergence of Glaser’s GT as a theory generating methodology (Glaser 1998:5). This overriding concern explains Glaser’s prescription of research procedures. The conceptualization of the core variable is carried out by an analytic data-fracturing and constant comparison and re-synthesizing of the fractured data-slices in order to find latent patterns in the data and relationships between these patterns. These patterns are categorized and named (coded) as concepts. The main criterion for identifying the core variable is that it is the concept (pattern) that is most related to the other concepts (patterns), or is a super-pattern of the other patterns, and that it explains most of the variation in the data (the behavior).
This means that Glaser’s GT is only a methodological paradigm or methodological “research program”. However, can such a paradigm exist without ontological and epistemological assumptions, as well as other theoretical assumptions? The answer is that its basic assumptions are limited to this: Man is a meaning-making creature. Consequently, social life is patterned and empirically integrated, and the core variable organizes and explains most of the variation in how the main concern is continually resolved. It is only a question of applying a rigorous and systematic method for discovering and explaining these patterns. No preoccupation is necessary regarding questions of ontology and epistemology or interpretation: Generation of concepts that are abstract of time, place and people, will transcends these issues. Thus, just do it. (Glaser 2001:199, 2004).
It is implied in these assumptions that the study of behavior, which is the basic task of all social science disciplines, is very different from the study of natural science phenomena. As one distinction, the element of “meaning-making” is mentioned – “meanings” link “subject” and “object”. Another distinction is the significance of the “main concern” of those being studied.
This raison d’être makes Glaser’s GT methodology one of its kind in its contribution to methodological diversity. As mentioned, no claim is made that it is better – or more relevant. Professional interest relevance is relevance – and substantive interest relevance is relevance. Glaser’s GT just gives researchers a different option with a different perspective on relevance.
What has been taught as well as researched in, for example, business schools has sometimes been criticized by insiders for being irrelevant “and so far removed from what businesses do that even our MBA students sell back their textbooks” (Jennings 1994). Glaser’s methodological paradigm provides an entirely new perspective on this debate. This perspective is entirely absent in the current debate (Pfeffer 2007) about the relevancy of management research.
Consequently, some provocative questions and answers can be put forward about relevancy of business and management research. What is the main basis upon which academics conceptualize and build theory? One obvious answer is pre-made concepts from extant theory and preconceived professional interest concerns of their research communities. To keep their work within these pre-framed borders becomes their main concern – the most important and problematic for academics, and this drives and directs their behavior. Their research relates more about their own main concern than about the main concern of those being researched. Furthermore, they only address minor issues of minor importance for practitioners, and not their main concerns. This answer is, of course, just a perspective, and it may not make full justice to the traditional academic concerns.
The main issue is conceptualization. Concepts are the building blocks of theory. The core variable is, of course, the pivotal building block. A new concept is a new idea. The concepts that have been generated by users of Glaser’s GT are typically new concepts, or concepts generated by “putting together of ideas into new connections” (Glaser 2006:3)
Glaser makes a distinction between two main types of concepts. These are substantive concepts ( patterns) and theoretical codes (models). Substantive concepts are stable latent patterns (the underlying meaning, uniformity and/or pattern within a set of descriptive incidents) that image the area being researched, and summarize the empirical substance of the data (Glaser 1992:38, 1998: 163). That the patterns are stable means that they are saturated by many interchangeable indicators and conceptual properties in the data (i.e., grounded).
Theoretical codes, on the other hand, model the types of relationships or connections between substantive concepts. For example, within basic economics, “price” and “quantity” are substantive concepts while a “demand curve” or “supply curve” as theoretical codes model their relationships. The researcher cannot know beforehand what kinds of theoretical codes will fit as models. Glaser recommends that the researcher should be open to the emergence of any kind of theoretical code and be knowledgeable about the widest possible range of theoretical codes in order to be able to recognize fitting codes when they are indicated in the data. Fitting codes may originate from very different theoretical perspectives. Consequently, Glaser recommends the researcher to read extensively in literature that is unrelated to the researched area during the research process. During this reading, the researcher should look for patterns and especially theoretical codes that link these patterns.
Glaser has provides a listing of 40 theoretical coding families (Glaser 1978:73-82, 1998:170-175, 2005:7-30), and he admits that the list is far from exhaustive. The recommenders of other versions of GT have a more restricted understanding of theoretical codes, which they mostly understand as “causality” (see next section). Thus, Glaser explains by other theoretical codes what recommenders of other versions of GT understand as simple causality.
It is hardly controversial that due to the element of “meaning” in social science, social science causality is usually very different from natural science or “classical physics” causality (Sorokin 1956, 1964). Thus, the connections between substantive concepts in natural science on the one hand, and social science on the other hand, may be very different. This possibly explains Glaser’s long list of theoretical coding families that go far beyond “classical-physics causality”. Nevertheless, a selection of recommended theoretical texts for the identification of the widest possible range of theoretical codes would be helpful for users of Glaser’s GT.
The substantive building blocks of the theory and their connectedness (theoretical codes) have to “earn their way” into the ensuing theory by repeated and stable indications in the data of conceptual fit, as they sum up and explain what drives and directs the participants’ behavior and what is most relevant and important for those being studied. Fit corresponds to positivist validity, but it is fit in action and usage and not by testing, and quantification is not an issue. The criteria for evaluating the theory are the fit of its concepts in action or usage, their workability in explaining, their relevancy or usability for practitioners, and their modifiability by new data (Glaser 1998:18-19). Experienced people who know the studied area will be able assess the theory’s credibility and usability by comparison to their own experiences.
The distinction between conceptual levels (abstractness and generality) of concepts is determined by high or low abstractness with regard to time, place and people. Theoretical codes are as models on a very high conceptual level and they may transcend borders of disciplines. The substantive concepts of a theory will be on different conceptual levels. The core variable will be the substantive concept of the theory that has the highest conceptual level. As descriptions bounded by the specificity of time, place and people, data is on the lowest conceptual level. A property is yet another type of concept; it is a conceptual characteristic of a substantive concept or theoretical code, or a concept of a concept. It has lesser conceptual level that the concept that it refers to. While data represents stories or contextual narratives that soon become superseded, concepts may last forever and easily transcend contextual boundaries (Glaser 1998:135-139). Consequently, description (contextual or conceptual) is the opposite of conceptualization.
The terminology of Glaser’s GT is only apparently similar to the terminology of Strauss-Corbin’s GT. It is in fact very different, because behind similar or identical terms, there is a very different meaning. This easily leads to misunderstandings. Open coding and selective coding in these two versions of GT do not mean the same. The significance of the core variable is fundamentally different the same applies for theoretical codes. Charmaz (2006) and Clarke (2005) have no place for the core variable. Strauss-Corbin’s axial coding is not used in Glaser’s GT. The axial coding paradigm is pre-programmed to take into account only the theoretical codes of causality (6 C’s: cause, consequence, contingency, context, condition, covariance) and the stimulus-organism-response model that is associated to the symbolic interactionist perspective (Strauss & Corbin 1998:126-137).
The researcher only chooses a general and loosely formulated research topic, and the researcher has to discover the research problem by the systematic treatment of his/her data. The first stage of this discovery of the research problem is the finding and conceptualization of the participants’ main concern and the next stage is the conceptualization and conceptual explanation of how the participants recurrently resolve their main concern. Thus, the finding and solving of the research problem amount to the entire research process, and the researcher should never preconceive the research problem.
If the researcher wants to preconceive the research problem, he/she should choose another research method. The researcher may preconceive the research problem by defining it in accordance with what he/she thinks is most relevant, or what the literature claims to be most relevant, or by spotting gaps in the literature in order to identify untested hypotheses. To study the literature as the first stage of the research with the deliberate purpose to define the research problem is a common pre-framing solution. If this were the case, the choice of Glaser’s GT method would be a meaningless choice. The researcher may be very familiar with the literature. However, this is no problem as long as the literature is not used as his/her source of substantive concepts.
If a researcher has decided to use Glaser’s GT, a preliminary study the literature in order to derive the research problem would be waste of time. The research problem, when empirically discovered from behavioral data, may be very different from what the extant or originally identified literature assumes it to be. (Glaser 1998:67-80). However, studies that have applied Glaser’s GT in closely related fields of enquiry could give some clues. Reading of them is recommended, and coding of data for emergent fit could be an option. As mentioned, the systematic reading (i.e., “explication-de-text-reading”) of unrelated literature in order to obtain a general training in the discovery of behavior patterns and theoretical codes is also recommended.
This restriction with regard to preliminary literature studies does not prevent the researcher from carrying out literature studies in order to find a loosely defined research topic that fits to his/her interests. However, if the researcher believes either that he/she can derive the participant’s “main concern and its recurrent solution” from this literature, or that he/she can ignore the empirical discovery of this “main concern” as the first stage of research, the choice of Glaser’s method would be meaningless.
Conceptual comparison to the existing literature within the field of enquiry is a very important part of the research task. However, relevant literature for conceptual comparison cannot be identified before stable behavioral patterns have emerged. This means that the literature comparisons have to be carried out at later stages of the research process, and especially towards the end. (Suddaby 2005, McCallin 2006, Andrews 2006, Nathaniel 2006, Thulesius 2006, Ekstrom 2006, Martin 2006a). These conceptual comparisons to the literature may substantiate the theory, but they may also lead to corrections or modifications of the theory or of the extant literature.
The theory should explain behavior and its variation by interrelated conceptual hypotheses (i.e., substantive concepts connected by theoretical codes), and with scope (i.e., as much as possible), and with conceptual parsimony (i.e., with as few concepts as possible). The theory should not be descriptive, i.e., just “tell a story” that is restricted to the specificity of organizational unit, time, place and people, or merely identify some themes within the data, or merely describe some concepts. Instead, the theory should explain by concepts that transcend the contextual specificity from whence the data came. Thus, “the story” (the data) is turned into concepts that are abstract of time, place and people, and the behavior and its variation is explained by the concepts and the relationships between the concepts, which both have to fit to the data. (Glaser 2001, 2003).
The distinctive characteristics of Glaser’s GT can be summed up as its two “hallmarks”. The first hallmark is first to find the core variable by the systematic treatment of the data, and to delimit the study to concepts that are most related to the core variable. This sequence eliminates the potential problem of “many equally justifiable interpretations of the same data”.
The second hallmark is suspension of preconceptions and prior knowledge about the area of research during the use of the research procedures, and to trust in “emergence of concepts from the data”. These two hallmarks are related. The second hallmark is necessary for finding the core variable and for fulfilling the first hallmark. These two hallmarks do not apply for other versions of GT. They explain the main difference between Glaser’s GT and other versions of GT. As mentioned, the point is not better or worse methods, but that the other GT methods violate Glaser’s original raison d’être for doing grounded theory.
GT literature frequently rewrites and distorts Glaser’s GT. Authors like Goulding (2002) and Douglas (2006) claim to have used Glaser’s GT, but they use axial coding or description. Clarke (2005), Charmaz (2006), Locke (2001) and many other authors (e.g., in Bryant & Charmaz 2007) challenge Glaser’s method, but without explaining its essential points. They label Glaser’s GT as “positivist”, or “objectivist” or “natural-science-based”. Charmaz justifies this labeling by Glaser’s emphasis on “conceptual explanation” and not just plain “understanding”, which qualitative research emphasizes in general (2006:126). Another objection is that “emergence of concepts from the data” is impossible, and that data and concepts are merely “interpretations” or “social constructions” (in the mind of the researcher) and not “objective realities”.
These objections seem to be based on misrepresentations and misunderstandings of Glaser’s GT method and especially of its second hallmark. Firstly, a user of Glaser’s GT has to take into account the element of “meaning” that links “subjects” and “objects”. This, as well as the first hallmark of Glaser’s GT, means that the method is deliberately created for social science research. Thus, there is no essential difference between “conceptual explanation” and “understanding”. Yet, there may be a huge difference between contextual storytelling and conceptual description that provides understanding, and conceptual explanation that also provides understanding. Even so, both can contribute.
Secondly, the researcher “surrenders to” the second hallmark of Glaser’s GT in order to submit the control of the research process to the data and to what the data relate about the participant’s main concern and its recurrent solution. This surrendering is made in order to prevent “preconceived professional interest concerns” from taking control of the research process.
Thirdly, the question about “social constructions or interpretations” versus “objective realities” is mainly a semantic issue. It is quite justified to label an obvious behavioral pattern like “going to the bathroom” as an “objective reality”, yet it is also a “construction” in the mind of the observer. The same applies for any concealed and unanticipated behavior pattern, which is uncovered in a study, and which relates to the most important and problematic for those being studied. There is no either/or, both apply. There may not be just one “reality”. However, ungrounded speculative “interpretations” or “constructions” are kept in check by keeping close to the data and by viewing the data through the “lens” of the core variable. What matters is the credibility of the theory with regard to what actually goes on in the recurrent solving of the main concern of those being studied. This credibility is also maintained by upholding the theory as a modifiable theory – it can only approach what is going on.
Fourthly, in addition to empiricism, deductive reasoning and the highly subjective element of intuition are also parts of Glaser’s GT. The researcher uses deduction when he/she decides which unit to sample next for data collection. This deduction is based on what the data so far have indicated regarding to conceptual elaboration. Furthermore, it is not possible to study theoretical codes in the literature without openness to and “participation with” deductive reasoning in this literature. If “classical physics” causality were the only theoretical coding family, such literature studied would be unnecessary. Yet, theoretical codes are only included in the theory when they are indicated in the data.
The second hallmark of Glaser’s GT has yet another purpose. Its “practicing” and “the surrendering of control to the data” usually leads to prolonged periods of standstill and confusion and “not knowing” while nothing seems to “emerge from the data”. This is uncomfortable for novice researchers, who are very vulnerable in such a situation. For experienced users of Glaser’s GT, this is a good sign. Before stages of “knowing” and “understanding” there have to be stages of “not knowing” and “confusion”. Such periods usually lead to subconscious processing during which new ideas can emerge at any time. The researcher needs to have pen and paper ready day and night to make notes. Comparison of these ideas to the data may lead to new discoveries. Thus, the second hallmark of Glaser’s GT, which Clarke (2005:17) and Charmaz (op.cit.) only understand as “positivist” or “objectivist”, actually works as a procedure to trigger intuition. Their labeling of Glaser’s GT as “positivist” does not explain researcher behavior and is therefore irrelevant.
The choice of other versions of GT may be justified when the choice of Glaser’s GT is meaningless. For example, some researchers may want to generate a theory about the stages of buyer behavior with regard to some specific product or service. The researchers may want study the literature as the first stage of their research. On this basis, they may decide to delimit or pre-frame their theory generation to a theory that resembles the traditional multistage models of consumer decision-making behavior, or to suggest an adaptation of these models. In this case, the choice of, for example, the Strauss-Corbin method may be justified. Such a choice would mean a considerable simplification. It would mean that the researchers only had to take into account the theoretical codes of the 6 C’s and the stimulus-organism-response model (the axial coding paradigm) that is associated with the symbolic interactionist perspective. On this basis, an adaptation of the model could be simple and useful.
Olavur Christiansen, Ph.D.
University of the Faroe Islands
Email: OlavurC@setur.fo
Alvesson, Mats & Dan Karreman. 2007. Constructing Mystery: empirical matters in theory development. Academy of Management review, 32.4: 1265-1281
Alvesson, Mats & Kaj Skoldberg. 2000. Reflexive Methodology, New Vistas for Qualitative Research. London: Sage Publications.
Andrews, Tom .2006. The literature Review in Grounded Theory: A response to McCallin. The Grounded Theory Review: An International Journal, 5.2/3: 29-33
Bryant, Anthony & Kathy Charmaz (eds.). 2007. The Sage Handbook of Grounded Theory. London: Sage Publications
Charmaz, Kathy. 2006. Constructing Grounded Theory – A Practical Guide Through Qualitative Analysis. London: Sage Publications
Clarke, Adele E. 2005. Situational Analysis – Grounded Theory After the postmodern Turn. Thousand Oaks, California: Sage Publications
Douglas, David. 2006. Intransitivities of Managerial decisions: A grounded theory case, Management Decision, 44.2:259-275
Ekstrom, Helene. 2006. Aspects on McCallinn’s paper, “Grappling with the literature in a grounded theory study”? The Grounded Theory Review: An International Journal, 5.2/3:45-46
Gephart, Robert & Sara Rynes. (2004). From the editors: Qualitative Research and the Academy of Management Journal. Academy of Management Journal, 47.4: 454-462
Glaser, B.G.1964. Organizational Scientists: Their Professional Careers. Indianapolis: Bobbs-Merrill
Glaser, B.G.1978. Advances in the Methodology of Grounded Theory: Theoretical Sensitivity. Mill Valley, CA: The Sociology Press.
Glaser, B.G.1992. Emergence vs Forcing: Basics of Grounded Theory Analysis. Mill Valley, CA: Sociology Press.
Glaser, B.G.1998. Doing Grounded Theory: Issues and Discussions. Mill Valley, CA: Sociology Press.
Glaser, B.G.2001. The Grounded Theory Perspective: Conceptualization Contrasted with Description. Mill Valley, CA: Sociology Press.
Glaser, B.G.2003. The Grounded Theory Perspective II: Description’s Remodeling of Grounded Theory Methodology. Mill Valley, CA: Sociology Press.
Glaser, B.G. 2004. Glaser’s explanations on the significance of ontology and epistemology in GT on a seminar in London, April 2004.
Glaser, B.G.2005. Grounded Theory Perspective III: Theoretical coding. Mill Valley, CA: Sociology Press.
Glaser, B.G.2006. The Roots of Grounded Theory, The Grounded Theory Review: An International Journal, 5.2/3: 1-10
Glaser, B.G.2007. Doing Formal Grounded Theory: A proposal. Mill Valley, CA: Sociology Press.
Glaser, B.G.2008. Doing Quantitative Grounded Theory. Mill Valley, CA: Sociology Press
Glaser, Barney & Judith Holton. 2005. Basic Social Processes. The Grounded Theory Review: An International Journal, 4.3: 1-27
Glaser, B.G.& Anselm L. Strauss. 1965. Awareness of Dying. New York: Aldine Publishing Company.
Glaser, B.G.& Anselm L. Strauss. 1967. The Discovery of Grounded Theory: Strategies for qualitative research. New York: Aldine De Gruyter.
Goulding, Christina .2002. Grounded Theory: A Practical Guide for Management, Business and Marketing researchers. London: Sage Publications
Hartman, Jan. 2001. Grundad teori, teorigenerering på empirisk grund. Lund: Studentlitteratur.
Holton, Judith A. 2007. The Coding process and its Challenges. Pp. 265-290 in Bryant, Anthony & Kathy Charmaz (ed.) (2007): The Sage Handbook of Grounded Theory, London: Sage Publications
Jennings, Marianne M. 1994. “Business Schools’ Formula for Irrelevance”. Wall Street Journal, Manager’s Journal, November 28, 1994.
Kelle, Udo. 2007. “Emergence” vs. “Forcing” of Empirical Data? A Crucial Problem of “Grounded Theory” reconsidered. Historical Social research/Historische Sozialforschung, no. supp 19:133-156
Kuhn, Thomas S. 1996. The Structure of Scientific Revolutions, Third Edition. Chicago: The University of Chicago Press
Lakatos, Imre. 1970. Falsification and the Methodology of Scientific Research Programmes. Pp. 91-196 in Lakatos & Musgrave (eds.). 1970. Criticism and Growth of Knowledge. Cambridge, UK: Cambridge University Press.
Locke, Karen. 2001. Grounded Theory in Management Research. London: Sage Publications.
Martin, Vivian. 2006a. The relationship between an emerging grounded theory and the existing literature: Four phases for consideration. The Grounded Theory Review: An International Journal, 5.2/3: 47-50
Martin, Vivian. 2006b. The postmodern turn: Shall Classic Grounded Theory take that detour? A review essay. The Grounded Theory Review: An International Journal, 5.2/3: 119-128
McCallin, Antoinette. 2006. Grappling with the literature in a grounded theory study. The Grounded Theory Review: An International Journal, 5.2/3: 11-28
Moustakas, Clark. 1994. Phenomenological Research methods. London: Sage Publications.
Nataliel, Alvita. 2006. Thoughts on the literature review and GT. The Grounded Theory Review: An International Journal, 5.2/3: 35-41
Pfeffer, Jeffrey. 2007. A modest proposal: How we might change the process and product of managerial research. Academy of Management Journal, 50.6: 1334-1345
Sorokin, P.A. 1956. Fads and Foibles in Modern Sociology and Related Sciences. Chicago: Henry Regnery Company
Sorokin, Pitirim A. 1964. Sociocultural Causality, Space, Time – A Study of referential Principles in Sociology and Social Science. New York: Russell & Russell
Strauss, Anselm & Juliet Corbin. 1990. Basics of Qualitative Research. Grounded Theory Procedures and Techniques. Newbury Park, California: Sage Publications.
Strauss, Anselm & Juliet Corbin. 1998. Basics of Qualitative Research, Techniques and Procedures for Developing Grounded Theory. California, Thousand Oaks: Sage Publications
Strubing, Jorg (2007): Glaser vs. Strauss? On the Methodological & Methodical Substance of the Difference of Two Separate Grounded Theory Variants. Historical Social research/ Historische Sozialforschung, no. supp 19: 157-174
Suddaby, Roy. 2005. From the editors: What grounded theory is not. Academy of Management Journal, 2005, 49.4: 633-642
Thulesius, Hans. 2006. New ways of using literature in GT. The Grounded Theory Review: An International Journal, 5.2/3: 43-44