Grounded Theory Definition: The intent of grounded theory is to generate or discover a theory – an abstract analytical schema of a philosophy, that relates to a particular situation. This situation could be one in which individuals interact, take actions, or engage in a process in response to a phenomenon (Creswell, 1998). Procedures Involved In Conducting A Study: In open coding, the researcher forms initial categories of information about the phenomenon being studied by segmenting information. Within each category (a category represents a unit of information composed of events, happenings and instances), the researcher finds several properties, or subcategories, and looks for data to dimensionalize, or show the extreme possibilities on a continuum of, the property. In axial coding, the researcher assembles the data in new ways after open coding. The researcher presents this using a coding paradigm or logic diagram in which he/she identifies a central phenomenon, explores causal conditions (i.e., categories of conditions that influence the phenomenon), specifies strategies (i.e., the actions or interactions that result from the central phenomenon), identifies the content and intervening conditions (i.e., the narrow and broad conditions that influence the strategies), and delineates the consequences (i.e., the outcomes of the strategies) for this phenomenon. In selective coding, the researcher identifies a “story line” and writes a story that integrates the categories in the axial coding model. In this phase, conditional propositions (or hypotheses) are typically presented. Finally, the researcher develops and visually portrays a conditional matrix that elucidates the social, historical, and economic conditions influencing the central phenomenon. This process results in a theory, written by the researchers close to a specific problem or population of people. Challenges: The researcher needs to set aside, as much as possible, theoretical ideas or notions so that the analytical, substantive theory can emerge. Despite the evolving, inductive nature of this form of a qualitative inquiry, the researcher must recognize that this is a systematic approach to research with specific steps in data analysis. The researcher needs faces the difficulty of determining when the categories are saturated or when the theory is sufficiently detailed. Ethnography Definition: An ethnography is a description and interpretation of a cultural or social group or system (Creswell, 1998). In such a study, the researcher examines the group’s observable and learned patterns of behavior, customs, and ways of life (Harris, 1968). Here, the researcher becomes a participant observer, and gets immersed in the day-to-day lives of the people or through one-on-one interviews with members of the group. The researcher focuses on the meanings of behavior, language, and inter-actions of the culture-sharing group. Procedures Involved In Conducting A Study: The research begins with the researcher looking at people in interaction in ordinary settings and attempting to discern pervasive patterns such as life cycles, events, and cultural themes. To establish patterns, the ethnographer engages in extensive work in the field (field work), gathering information through observations, interviews, and materials helpful inn developing a portrait and establishing “cultural rules” of the culture-sharing group. The researcher is sensitive to gaining assess to the field through gatekeepers. The ethnographer locates key informants, i.e., individuals who provide useful insights into the group and can steer the researcher to information and contacts. The researcher is also sensitive about reciprocity between the investigator and the subjects being studied, so that something will be returned to the subjects being studied in exchange for their information. Lastly, the researcher is also sensitive to reactivity, the impact of the researcher on the site and the people being studied. The researcher also makes every effort to make his/her intent known from the start to avoid any trace of deception. The researcher then does a detailed description of the culture-sharing group or individual, an analysis by themes or perspectives and some interpretation for meanings of social interaction and generalizations about human social life. Challenges: The researcher needs to have a grounding in cultural anthropology and the meanings of social-cultural systems as well as the concepts typically explored by ethnographers. The time to collect data is extensive, involving prolonged time in the field. The style of writing, literary (almost story telling approach), may limit audience and may be challenging for some authors who are used to traditional approaches of writing social science research. There is the possibility that the researcher would “go native” and be unable to complete the study or be compromised in the study. Case Study Definition: Creswell (1998) defines a case study as an exploration of a “bounded system” or a case (or multiple cases) over time through detailed, in-depth data collection involving multiple sources of information rich in context. Some consider “the case” as an object of study (e.g., Stake, 1995) while others consider it a methodology (e.g., Merriam, 1998). According to Creswell, the bounded system is bounded by time and place, and it is the case being studied – a program, an event, an activity, or individuals. Procedures Involved In Conducting A Study: The researcher needs to situate the case in a context or setting. The setting may be a physical, social, historical, and/or economic. The researcher needs to identify the focus of the study. It could be either on the case (intrinsic study), because of its uniqueness, or it may be on an issue or issues (instrumental study), with the case used instrumentally to illustrate the issue. A case study could involve more than one case (collective case study). In choosing what case to study, a researcher may choose a case because it shows different perspectives on the problem, process, or event of interest, or it may be just an ordinary case, accessible, or unusual. The data collection is extensive, drawing on multiple sources of information such as observations, interviews, documents, and audio-visual materials. The data analysis can be either a holistic analysis of the entire case or an embedded analysis of a specific aspect of the case. From the data collection, a detailed description of the case is done. Themes or issues are formulated and then the researcher makes an interpretation or assertions about the case. When multiple cases are chosen, a typical format is to provide a detailed description of each case and themes within the case (called within-case analysis), followed by a thematic analysis across the cases (called a cross-case analysis), as well as assertions or an interpretation of the meaning of the case. In the final stage, the researcher reports the “lessons learned” from the case (Lincoln and Guba, 1985). Challenges: The researcher needs to identify his/her case among a host of possible candidates. The researcher needs to decide whether to study a single case or multiple cases. The motivation for considering many cases is the issue of generalizability, which is not so much of a pressing issue in qualitative inquiry. Studying more than one case runs the risk of a diluted study, lacking the “depth” compared to a single case. “How many” cases becomes a challenge then. Getting enough information to get a good depth for the case is a challenge. Deciding on the boundaries in terms of time, events and processes may be challenging. Some cases have no clean beginning and ending points. The Creation of Theory: A Recent Application of the Grounded Theory Method by Naresh R. Pandit * The Qualitative Report, Volume 2, Number 4, December, 1996 (http://www.nova.edu/ssss/QR/QR2-4/pandit.html) Abstract This paper outlines a particular approach to building theory that was employed in a recent doctoral research project (Pandit, 1995). Three aspects used in conjunction indicate the project's novelty: firstly, the systematic and rigorous application of the grounded theory method; secondly, the use of on-line computerised databases as a primary source of data; and, thirdly, the use of a qualitative data analysis software package to aid the process of grounded theory building. Introduction The principal objective of this paper is to reveal how the style of qualitative research known as the 'grounded theory approach' was applied in a research project (Pandit, 1995) which attempted to generate a theoretical framework of corporate turnaround. Two lesser auxiliary objectives are firstly, to assess the utility of on-line computerised databases as a primary source of data for this type of research; and secondly, to assess the extent to which computer-based qualitative data analysis (QDA) software can aid this type of research. In the next section the elements of grounded theory are recalled. This is followed by a detailed account of the procedures and rationale of grounded theory building as employed in the project. Next, the generated theory of corporate turnaround is overviewed and finally, a reflexive account of the research experience is given. The Elements Of Grounded Theory The three basic elements of grounded theory are concepts, categories and propositions. Concepts are the basic units of analysis since it is from conceptualisation of data, not the actual data per se, that theory is developed. Corbin and Strauss (1990, p. 7) state: Theories can't be built with actual incidents or activities as observed or reported; that is, from "raw data." The incidents, events, happenings are taken as, or analysed as, potential indicators of phenomena, which are thereby given conceptual labels. If a respondent says to the researcher, "Each day I spread my activities over the morning, resting between shaving and bathing," then the researcher might label this phenomenon as "pacing." As the researcher encounters other incidents, and when after comparison to the first, they appear to resemble the same phenomena, then these, too, can be labelled as "pacing." Only by comparing incidents and naming like phenomena with the same term can the theorist accumulate the basic units for theory. The second element of grounded theory, categories, are defined by Corbin and Strauss (1990, p. 7) thus: Categories are higher in level and more abstract than the concepts they represent. They are generated through the same analytic process of making comparisons to highlight similarities and differences that is used to produce lower level concepts. Categories are the "cornerstones" of developing theory. They provide the means by which the theory can be integrated. We can show how the grouping of concepts forms categories by continuing with the example presented above. In addition to the concept of "pacing," the analyst might generate the concepts of "self-medicating," "resting," and "watching one's diet." While coding, the analyst may note that, although these concepts are different in form, they seem to represent activities directed toward a similar process: keeping an illness under control. They could be grouped under a more abstract heading, the category: "Self Strategies for Controlling Illness." The third element of grounded theory are propositions which indicate generalised relationships between a category and its concepts and between discrete categories. This third element was originally termed 'hypotheses' by Glaser and Strauss (1967). It is felt that the term 'propositions' is more appropriate since, as Whetten (1989, p. 492) correctly points out, propositions involve conceptual relationships whereas hypotheses require measured relationships. Since the grounded approach produces conceptual and not measured relationships, the former term is preferred. The generation and development of concepts, categories and propositions is an iterative process. Grounded theory is not generated a priori and then subsequently tested. Rather, it is, ... inductively derived from the study of the phenomenon it represents. That is, discovered, developed, and provisionally verified through systematic data collection and analysis of data pertaining to that phenomenon. Therefore, data collection, analysis, and theory should stand in reciprocal relationship with each other. One does not begin with a theory, then prove it. Rather, one begins with an area of study and what is relevant to that area is allowed to emerge. (Strauss and Corbin, 1990, p. 23. Emphasis added.) The Process Of Grounded Theory Building Five analytic (and not strictly sequential) phases of grounded theory building were identified: research design, data collection, data ordering, data analysis and literature comparison. Within these phases, nine procedures or steps were followed. These phases and steps were evaluated against four research quality criteria: construct validity, internal validity, external validity and reliability. Briefly, construct validity is enhanced by establishing clearly specified operational procedures. Internal validity is enhanced by establishing causal relationships whereby certain conditions are shown to lead to other conditions, as distinguished from spurious relationships. In this sense, internal validity addresses the credibility or "truth value" of the study's findings. External validity requires establishing clearly the domain to which the study's findings can be generalised. Here, reference is made to analytic and not statistical generalisation and requires generalising a particular set of findings to some broader theory and not broader population. Finally, reliability requires demonstrating that the operations of a study - such as data collection procedures - can be repeated with the same results. Table 1 provides an overview of these phases, steps and tests and forms the template for the subsequent discussion which moves from a normative or prescriptive account of recommended activities to a descriptive account of how these prescriptions were applied in the study. Table 1: The Process of Building Grounded Theory PHASE ACTIVITY RATIONALE RESEARCH DESIGN PHASE Step Review of technical Definition of research Focuses 1 literature question efforts Definition of a priori Constrains irrelevant variation and constructs sharpens external validity Step Selecting cases Theoretical, not random, Focuses efforts on theoretically useful cases 2 sampling (e.g., those that test and/or extend theory) DATA COLLECTION PHASE Step Develop rigorous data Create case study Increases reliability Increases construct 3 collection protocol database validity Employ multiple Strengthens grounding of theory by data collection triangulation of evidence. Enhances methods internal validity Qualitative and Synergistic view of evidence quantitative data Step Entering the field Overlap data Speeds analysis and reveals 4 collection helpful adjustments to data collection and analysis Allows investigators to take advantage of Flexible and opportunistic emergent themes and unique case features data collection methods DATA ORDERING PHASE Step Data ordering Arraying events Facilitates easier data analysis. Allows 5 chronologically examination of processes DATA ANALYSIS PHASE Step Analysing Use open Develop concepts, categories and properties 6 data relating to coding the first case Develop connections between a category Use axial and its sub-categories coding Integrate categories to build theoretical Use selective framework coding All forms of coding enhance internal validity Step Theoretical sampling Literal and theoretical Confirms, extends, and sharpens theoretical 7 replication across cases framework (go to step 2 until theoretical saturation) Step Reaching closure Theoretical saturation Ends process when marginal improvement 8 when possible becomes small LITERATURE COMPARISON PHASE Step Compare emergent Comparisons with Improves construct definitions, and 9 theory with extant conflicting frameworks therefore internal validity literature Comparisons with similar Also improves external validity by frameworks establishing the domain to which the study's findings can be generalised Research Design Phase Research design is defined by Easterby-Smith et al. (1990, p. 21) as, ... the overall configuration of a piece of research: what kind of evidence is gathered from where, and how such evidence is interpreted in order to provide good answers to the basic research question[s]. It follows logically that the first step is to define the basic research questions. These should be defined narrowly enough so that the research is focused and broad enough to allow for flexibility and serendipity. A good source of research questions in grounded theory studies is the 'technical literature' (i.e., reports of research studies and theoretical and philosophical papers characteristic of professional and disciplinary writing) on the general problem area (Strauss and Corbin, 1990, p. 52). Once basic research questions have been generated and the research is focused, the next aspect of research design and the second step is to select the first case. Cases (the principal units of data in this research) should be selected according to the principle of theoretical sampling: The process of data collection for generating theory whereby the analyst jointly collects, codes, and analyses his data and decides what data to collect next and where to find them, in order to develop his theory as it emerges. (Glaser and Strauss, 1967, p. 45.) Accordingly, Unlike the sampling done in quantitative investigations, theoretical sampling cannot be planned before embarking on a grounded theory study. The specific sampling decisions evolve during the research process itself. (Strauss and Corbin, 1990, p. 192) During initial data collection, when the main categories are emerging, a full 'deep' coverage of the data is necessary. Subsequently, theoretical sampling requires only collecting data on categories, for the development of properties and propositions. The criterion for judging when to stop theoretical sampling is the category's or theory's 'theoretical saturation'. By this term Glaser and Strauss refer to the situation in which: ... no additional data are being found whereby the (researcher) can develop properties of the category. As he sees similar instances over and over again, the researcher becomes empirically confident that a category is saturated ... when one category is saturated, nothing remains but to go on to new groups for data on other categories, and attempt to saturate these categories also. (1967, p. 65.) A qualification springs from the fact that not all categories are equally relevant, and accordingly the depth of enquiry into each one should not be the same. As a general rule, core categories, those with the greatest explanatory power, should be saturated as completely as possible. A theory is saturated when it is stable in the face of new data and rich in detail. Theoretical sampling translates in practical terms into two sampling events. An initial case is selected and, on the basis of the data analysis pertaining to that case and hence the emerging theory, additional cases are selected. The initial case (unit of data) in this study was the technical literature on the subject of corporate turnaround. Strauss and Corbin support this approach and state: The literature can be used as secondary sources of data. Research publications often include quoted materials from interviews and field notes and these quotations can be used as secondary sources of data for your own purposes. The publications may also include descriptive materials concerning events, actions, settings, and actors' perspectives, that can be used as data using the methods described. (1990, p. 52.) The grounded analysis of the first ('literature') case led to the generation of the initial theoretical framework of corporate turnaround. Additional ('empirical') cases were then selected, one at a time, to test and extend this framework. To recall, according to the principle of theoretical sampling, each additional case should serve specific purposes within the overall scope of enquiry. Three options are identified by Yin (1989, p. 53-54): (a) choose a case to fill theoretical categories, to extend the emerging theory; and/or, (b) choose a case to replicate previous case(s) to test the emerging theory; or, (c) choose a case that is a polar opposite to extend the emerging theory. Logically, this implies that each additional case must be carefully selected so that it produces similar results (a literal replication - options (a) and (b) above ); or, produces contrary results but for predictable reasons (a theoretical replication - option (c) above). The second case or unit of data, Fisons plc which experienced a turnaround during the period 1975-84, was selected for the purpose of literal replication, that is, to fill theoretical categories and to test the emerging theory. The third case, British Steel Corporation (BSC) which experienced a turnaround during the period 1975-89 was again chosen for the purposes of literal replication. After the analysis of the three cases, the marginal improvement to the theoretical framework was small. Theoretical saturation via literal replication had been approached and the decision to conclude the research was taken. This experience is corroborated by Martin and Turner (1986, p. 149) who state: By the time three or four sets of data have been analysed, the majority of useful concepts will have been discovered. Data Collection Phase The grounded approach advocates the use of multiple data sources converging on the same phenomenon and terms these 'slices of data.' Glaser and Strauss (1967, p. 65) state, In theoretical sampling, no one kind of data on a category nor technique for data collection is necessarily appropriate. Different kinds of data give the analyst different views or vantage points from which to understand a category and to develop its properties; these different views we have called slices of data. While the [researcher] may use one technique of data collection primarily, theoretical sampling for saturation of a category allows a multifaceted investigation, in which there are no limits to the techniques of data collection, the way they are used, or the types of data acquired. (Emphasis in original.) Similarly, Eisenhardt (1989, p. 538) states: ... case study research can involve qualitative data only, quantitative only, or both.. Moreover, the combination of data types can be highly synergistic. The synergy (or 'data triangulation') referred to works as follows: quantitative data can indicate directly observable relationships and corroborate the findings from qualitative data. Qualitative data can help understand the rationale of the theory and underlying relationships. The use of multiple data sources thus enhances construct validity and reliability. The latter is further enhanced through the preparation of a case study database which is a formal assembly of evidence distinct from the case study report. Yin (1989, pp. 98-99) states: Every case study project should strive to develop a formal, retrievable database, so that in principle, other investigators can review the evidence directly and not be limited to the written reports. In this manner, the database will increase markedly the reliability of an entire case study. (Emphasis added.) To summarise, the third step is to develop a rigorous data collection protocol by employing multiple data collection methods using both qualitative and quantitative data and systematically establishing a case study database. The principal data source in this study for the two 'empirical' cases (i.e., Fisons and BSC) was archival material in the form of reports in newspapers, trade journals, business journals, government publications, broker reviews, annual company documents and press releases. These data were extracted in computerised form (i.e., ASCII files) from the Reuter Textline and Predicasts PROMT (Predicasts Overview of Markets and Technology) databases. There are over 600 active information sources contributing to Reuter Textline, of which approximately 200 are primary sources and over 400 are associated sources provided by third party contributors. The earliest material on Reuter Textline dates from 1980. Predicasts is the largest on-line source of business information of its kind. The Predicasts family of complimentary databases contains more than 5,000,000 article abstracts, forecasts, statistical series and full text records from a broad range of business, industry and government sources. The earliest material on PROMT dates from 1975. It was whilst reading a paper by Turner (1983) that the idea of developing grounded theory based on this type of data was formed. In that research project, ... documentary sources were treated like sets of field notes. Analysis and category generation was commenced at the first paragraph of the report, and a theoretical framework generated which would handle the aspects perceived to be of interest to each paragraph. (1983, p. 342.) Case study databases were constructed within the qualitative data analysis software package ATLAS. A full discussion of the procedures followed is provided in the data analysis section below. The fourth step was thus to ensure that data was collected and analysed simultaneously and flexibility is maintained. This overlap allows adjustments to be made to the data collection process in light of the emerging findings. Eisenhardt describes such flexibility as 'controlled opportunism'. Data Ordering Phase The fifth step was data ordering. Following Yin (1989, p. 119) data for the two 'empirical' cases were ordered chronologically: The arraying of events into a chronology permits the investigator to determine causal events over time, because the basic sequence of a cause and its effect cannot be temporally inverted. However, unlike the more general time-series approaches, the chronology is likely to cover many different types of variables and not be limited to a single independent or dependent variable. Data Analysis Phase Once data were ordered, the sixth step was to analyse the data. Data analysis is central to grounded theory building research. For the study as a whole, data collection, data ordering, and data analysis were interrelated as depicted in figure 1 (the attached numbers indicate the activity's analytic sequence). Figure 1: The Interrelated Processes of Data Collection, Data Ordering, and Data Analysis to Build Grounded Theory Data Analysis (4) Theory Development (5) Data Ordering (3) Theory Saturation ? Yes Data Collection (2) No Reach Closure (6) Theoretical Sampling (1) Within this general framework, data analysis for each case involved generating concepts through the process of coding which, ... represents the operations by which data are broken down, conceptualised, and put back together in new ways. It is the central process by which theories are built from data. (Strauss and Corbin, 1990, p. 57.) There are three types of coding: open coding, axial coding, and selective coding. These are analytic types and it does not necessarily follow that the researcher moves from open through axial to selective coding in a strict, consecutive manner. Open coding refers to that part of analysis that deals with the labelling and categorising of phenomena as indicated by the data. The product of labelling and categorising are concepts - the basic building blocks in grounded theory construction. Open coding requires application of what is referred to as 'the comparative method', that is, the asking of questions and the making of comparisons. Data are initially broken down by asking simple questions such as what, where, how, when, how much, etc. Subsequently, data are compared and similar incidents are grouped together and given the same conceptual label. The process of grouping concepts at a higher, more abstract, level is termed categorising. Whereas open coding fractures the data into concepts and categories, axial coding puts those data back together in new ways by making connections between a category and its sub-categories (i.e., not between discrete categories which is done in selective coding). Thus, axial coding refers to the process of developing main categories and their sub- categories. Selective coding involves the integration of the categories that have been developed to form the initial theoretical framework. Firstly, a story line is either generated or made explicit. A story is simply a descriptive narrative about the central phenomenon of study and the story line is the conceptualisation of this story (abstracting). When analysed, the story line becomes the core category: The core category must be the sun, standing in orderly systematic relationships to its planets. (Strauss and Corbin, 1990, p. 124.) Subsidiary categories are related to the core category according to the paradigm model, the basic purpose of which is to enable the researcher to think systematically about data and relate them in complex ways. The basic idea is to propose linkages and look to the data for validation (move between asking questions, generating propositions and making comparisons). The basic features of this model are depicted in figure 2 below. Figure 2: The Paradigm Model Causal Conditions Phenomenon Context Intervening Conditions Action / Interaction Strategies Consequences The core category (i.e., the central idea, event or happening) is defined as the phenomenon. Other categories are then related to this core category according to the schema. Causal conditions are the events that lead to the development of the phenomenon. Context refers to the particular set of conditions and intervening conditions, the broader set of conditions, in which the phenomenon is couched. Action/interaction strategies refer to the actions and responses that occur as the result of the phenomenon and finally, the outcomes, both intended and unintended, of these actions and responses are referred to as consequences. An important activity during coding is the writing of memos. Corbin and Strauss (1990, p. 10) maintain that, Writing theoretical memos is an integral part of doing grounded theory. Since the analyst cannot readily keep track of all the categories, properties, hypotheses, and generative questions that evolve from the analytical process, there must be a system for doing so. The use of memos constitutes such a system. Memos are not simply "ideas." They are involved in the formulation and revision of theory during the research process. At least three types of memo may be distinguished: code memos, theoretical memos and operational memos. Code memos relate to open coding and thus focus on conceptual labelling. Theoretical memos relate to axial and selective coding and thus focus on paradigm features and indications of process. Finally, operational memos contain directions relating to the evolving research design. In the past, the tools used to aid the type of data analysis elucidated above were simply scissors, a copier and piles of blank paper. In this research project, data were analysed using the qualitative data analysis software package ATLAS which also facilitated the construction of case study databases. The use of computer programs to aid the analysis of qualitative data is a recent innovation: ... there has been considerable progress in the analysis of qualitative data using a variety of specially written computer programs ... There are at present around a dozen programs on the market or under development, each with different characteristics and facilities. (Lee and Fielding, 1991, p. 1.) The principal advantage of using a program is that it simplifies and speeds the mechanical aspects of data analysis without sacrificing flexibility thereby freeing the researcher to concentrate to a greater extent on the more creative aspects of theory building: The thinking, judging, deciding, interpreting, etc., are still done by the researcher. The computer does not make conceptual decisions, such as which words or themes are important to focus on, or which analytical step to take next. These analytical tasks are still left entirely to the researcher. (Tesch, 1991, pp. 25-26.) Lee and Fielding summarise: It is likely that computers will bring real benefits to qualitative researchers, making their work easier, more productive and potentially more thorough. (1991, p. 6.) There are two modes of data analysis within ATLAS: firstly, the 'textual level' which focuses on the raw data and includes activities such as text segmentation and, coding and memo writing; and secondly, the 'conceptual level' which focuses on framework building activities such as interrelating codes, concepts and categories to form theoretical networks. In general, we found the procedures within ATLAS to be both efficient and firmly based on the principles of grounded theory generation. Once a theoretical framework relating to the first case has been generated, the next and seventh step in theory building case research is to test and develop this framework by selecting additional cases according to the principle of theoretical sampling, that is, with the aim to extend and/or sharpen the emerging theory by filling in categories that may need further refinement and/or development. The eighth step, reaching closure, is taken according to the principle of theoretical saturation, that is, when the marginal value of the new data is minimal. Literature Comparison Phase The ninth and final step is to compare the emerged theory with the extant literature and examine what is similar, what is different, and why. Eisenhardt (1989, p. 545) states: Overall, tying the emergent theory to existing literature enhances the internal validity, generalisability, and theoretical level of the theory building from case study research ... because the findings often rest on a very limited number of cases. The emergent theory of corporate turnaround was compared with the extant theories in the broader field of strategic management. This revealed the discovered theory to resemble in many ways Pettigrew's "content-context-process" model of strategic change (1987). An Overview of a Grounded Theory of Corporate Turnaround Through the process of open and axial coding in ATLAS/ti a number of concepts and categories were generated and developed. During selective coding (i.e., the integration of categories) the core category was defined and labelled 'recovery strategy content'. The other major categories were then related to this category. The content of appropriate recovery strategies were found to be contingent upon six sets of contextual factors: the causes of decline; the severity of the crisis; the attitude of stakeholders; industry characteristics; changes in the macroeconomic environment; and, the firm's historical strategy. The content of recovery strategies was usefully decomposed into operational level actions (management change, improved controls, reduction in production costs, investment in plant and machinery, decentralisation, improved marketing, and restructuring finances) and strategic level actions (asset reduction/divestiture and product/market reorientation). An implementation or process dimension was also discovered. Successful actions to effect recovery fall into four distinct (but overlapping) stages (the management change stage, the retrenchment stage, the stabilisation stage, and the growth stage). A diagrammatical depiction of this framework is given in figure 3 (see Pandit, 1995 for a fuller discussion). Figure 3: A Theoretical Framework of Corporate Turnaround CONTEXTUAL FACTORS The causes of decline The severity of the crisis The attitude of the stakeholders Industry characteristics Changes in the macroeconomic environment The firm's historical strategy IMPLEMENTATION / PROCESS OF RECOVERY ACTIONS RECOVERY STRATEGY Management change stage CONTENT Retrenchment stage Operational level: Stabilisation stage Management change Growth stage Improved controls Reduction in production costs Investment in plant and machinery Decentralization Improved marketing Restructuring finances Strategic level: Asset reduction/divestiture Product/market reorientation 53 propositions linking the concepts and categories within the framework were generated and tested. Table 2 lists a sample of five (see Pandit, 1995, pp. 277-278 for the full list). Table 2: A Sample of Propositions Generated by the Literature Case and Supported by the Cases of Fisons and BSC Fisons BSC Not Not Proposition Generated by Explicitly Implicitly Explicitly Implicitly Referred Referred the Literature Case Supported Supported Supported Supported To To A sustained deterioration in performance is the result of X X both internal and external causes. Successful turnaround firms are more severely affected in terms of financial X X performance in the downturn phase than unsuccessful recoveries. If the causes of decline are primarily internal in origin, actions that improve X X efficiency at the operational level should be emphasised to effect successful recovery. If the causes of decline are primarily external in origin, strategic level actions should X X be emphasised to effect successful recovery. Appropriate recovery actions vary according to X X industry stage. Reflections With respect to the two lesser auxiliary objectives of this study I found firstly, that the data available from the on-line databases Reuters Textline and Predicasts PROMT to be extremely appropriate for this type of research. The hundreds of articles extracted provided a rich and diverse source of information for the two 'empirical' cases. Events were easily traced over time, differing viewpoints provided much intellectual stimulation, and reported interviews with key people at the time rather than retrospectively provided valuable insights and served as an efficient and effective substitute for conducting similar interviews ourselves. Ultimately, my assessment of the quality of the data is a tribute to the quality of business journalism in the UK, the USA and continental Europe (where most of the reports we analysed originated). My second auxiliary objective was to assess the utility of computer-based qualitative data analysis software packages when used in conjunction with on-line data in grounded theory research. In general, I found the packages to be of limited use (rather than easing the process they tend to overcomplicate it) with much development required before they can make a significant impact on the conduct and quality of qualitative research. However, I found the package that I chose (ATLAS) to be very much the exception to the rule. A number of attributes distinguished it from the alternatives. Firstly, it is very 'user- friendly' and operates in a similar manner to the more widely used Windows package developed by Microsoft. Secondly, it is powerful. Given the immense volume of data to be analysed, problems were expected but thankfully never materialised. Finally, it is thoroughly based on the principles of grounded theory generation and therefore few compromises had to be made. Five problems were encountered in this study. Four relate to the research process and one, more fundamentally, to the research approach. Firstly, the process of grounded theory research is extremely time-consuming. The sheer volume and complexity of data generated for this study was quite daunting, although the use of ATLAS aided matters considerably. Secondly, grounded theory research involves long periods of uncertainty. Without a priori hypotheses to test and established protocol to follow, much of the first half of the study period required a good measure of faith and hope. Thankfully, there did come a time after much patience, persistence, and perspiration when things become clearer. Thirdly, the data extracted from the two on-line databases was sometimes found to be incomplete. Often, and particularly with long articles, only summaries and not the full text was available. This was particularly disappointing given that longer articles are usually the most informative and, therefore, potentially of most use. I estimate that about 10 per cent of the data extracted was shortened in this way. Also, any graphs related to an article are not reproduced in computerised form within the databases. Once again, valuable information is lost. Fourthly, collecting data from on-line databases is expensive. Fortunately, my access to Reuters Textline was at a preferential rate (used for promotional purposes) and my access to Predicasts PROMT was free of charge owing to the fact that it was on a limited trial period. Without these two contingencies our runs (which amounted to well over 100 hours of on-line time) would have cost many thousands of pounds sterling. Let us now turn to the fifth and more fundamental problem with the overall approach of this study. Grounded theory research requires certain qualities of the researcher. In particular, confidence, creativity and experience (both of doing research and of the context(s) being researched) are of great benefit. Accordingly, the approach does not favour the novice researcher who may be just beginning to develop these qualities. This is not to say that novice researchers should not embark upon grounded theory studies; rather, I imply that (a) they are likely to find the approach more difficult than more conventional methodologies; and, (b) the more experienced (probably postdoctoral) researcher is likely to produce better theory. Overview Grounded theory begins with a research situation. Within that situation, your task as researcher is to understand what is happening there, and how the players manage their roles. You will mostly do this through observation, conversation and interview. After each bout of data collection you note down the key issues: this I have labelled "note-taking". Constant comparison is the heart of the process. At first you compare interview (or other data) to interview (or other data). Theory emerges quickly. When it has begun to emerge you compare data to theory. The results of this comparison are written in the margin of the note-taking as coding. Your task is to identify categories (roughly equivalent to themes or variables) and their properties (in effect their sub-categories). If this is all a bit abstract, some examples later will help. As you code, certain theoretical propositions will occur to you. These may be about links between categories, or about a core category: a category which appears central to the study. As the categories and properties emerge, they and their links to the core category provide the theory. You write yourself notes about it -- memoing. As the data collection and coding proceeds the codes and the memos accumulate. You add to your sample through theoretical sampling. This is purposive sampling which increases the diversity of your sample, searching for different properties. If your core category and its linked categories saturate; you no longer add to them or their properties. This is a sign that it is time to move to sorting. You group your memos, like with like, and sequence them in whatever order will make your theory clearest. The literature is accessed as it becomes relevant. It is not given special treatment. Glaser makes the point that most research including qualitative research is hypothesis-testing. The order of your sorted memos provides you with the skeleton, and many of the words, of your thesis. You begin writing. To summarise graphically ... Over time, a grounded theory study works through the following mostly- overlapping phases. In short, data collection, note-taking, coding and memoing occur simultaneously from the beginning. Sorting occurs when all categories are saturated -- this is explained in more detail later, as are the elements of this diagram. Writing occurs after sorting. For ease of explanation, what follows may seem a bit prescriptive. Feel free to experiment with it until you find something that works for you. The theory is emergent -- discovered in the data, Glaser will say. The methods can be emergent too. This is an important issue, worth more attention. [ index ] Hypothesis testing versus emergence What most differentiates grounded theory from much other research is that it is explicitly emergent. It does not test a hypothesis. It sets out to find what theory accounts for the research situation as it is. In this respect it is like action research: the aim is to understand the research situation. The aim, as Glaser in particular states it, is to discover the theory implicit in the data. This distinction between "emergence and forcing", as Glaser frames it, is fundamental to understanding the methodology. Most of you, whatever your discipline, will have been exposed more to hypothesis-testing research than to emergent research. The research processes you have learned and the thesis structures you have internalised are those of hypothesis testing, not of emergence. Doing grounded theory well is partly a matter of unlearning some of what you have been taught or have acquired through your reading. If you judge grounded theory by the criteria you have learned to use for hypothesis testing research you will likely misjudge it, perhaps badly. In particular, the place of literature is quite different. So is the way in which both methodology and theory develop gradually as data and interpretations accumulate. In particular, judgments about the rigour of research are often based on narrow criteria: criteria which make sense only for the methodology for which they were developed. Grounded theory has its own sources of rigour. It is responsive to the situation in which the research is done. There is a continuing search for evidence which disconfirms the emerging theory. It is driven by the data in such a way that the final shape of the theory is likely to provide a good fit to the situation. In fact, Glaser suggests two main criteria for judging the adequacy of the emerging theory: that it fits the situation; and that it works -- that it helps the people in the situation to make sense of their experience and to manage the situation better. Elsewhere, I’ve offered similar arguments in favour of action research. In particular, I draw your attention to my 1999 paper to the AQR conference, and the recent paper on data driven research. Now, the elements in more detail ... [ index ] Data collection You will of course keep your eyes open. There is a lot to be learned just by observing, some of it evident within minutes of entering a situation. Interviews are frequently the main source of the information you will develop your theory from. But any data collection methods can be used. Focus groups are not uncommon in other qualitative research, and are suited to grounded theory. So is informal conversation, group feedback analysis, or any other individual or group activity which yields data. I’ve included some references in the bibliography. For interviewing I like Minichiello, Aroni, Timewell and Alexander (1990) or Kvale (1996), and of course my own Convergent interviewing. For focus groups you might try some of the recent work: Bader and Rossi (1998) I like, and also Barbour and Kitzinger (1999). For both interviews and focus groups I’ve also listed some others. For group feedback analysis, try Heller and Brown (1995). I won’t go into further detail here. I will say two further things. First, I assume that you will touch base with the literature on your chosen method. Second, I presume you’ll continue to fine tune it as you develop more experience in its use. I encourage you to experiment. [ index ] Note-taking Glaser recommends against recording or taking notes during an interview of other data collection session. Speaking for myself, I agree with his avoidance of tape recordings and word-by-word transcripts. I think you’ll get more understanding from the extra interviews you could do in the time it would take you to listen to and transcribe a tape recording. However, I think he is vulnerable on that point, and especially for thesis purposes. My suggestion is that you take key-word notes during the interviews and convert them to themes afterwards. I also suggest that you tape-record the interviews and check your notes against the tape recording. This won’t be as time consuming (or alternatively as costly) as full transcripts and in my experience it will do the job well. If it’s not for thesis purposes I think you can make your own choices. I neither take notes during interviews nor use a tape recorder. I find rapport develops more rapidly and effectively if I don’t. However, I do have a memory system which allows me to memorise up to 20 distinct themes (more if it’s necessary) and recall them in order. The coding (which follows) will be much easier if you do it adjacent to the interview notes. You can leave wide margins (as much as a third of the page, perhaps) for that purpose. [ index ] Coding So -- in reality or in imagination -- you have in front of you a set of interview notes. They are written in the left hand two-thirds of the page, let’s say. You’ve identified any important bio-data about the person interviewed at the head of the notes (this may later help to identify properties). Have some other pieces of paper, or preferably cards, for memoing. The benefits of that will become evident soon. You begin to code. You take a sentence at a time and examine it. [ index ] o Constant comparison For the first interview you are merely asking yourself: What is going on here? What is the situation? How is the person managing that situation? Therefore, what categories (plural) are suggested by that sentence? Code the second interview with the first interview in mind. Code subsequent interviews (or data from other sources) with the emerging theory in mind. That’s constant comparison: initially comparing data set to data set; later comparing data set to theory. o For instance, suppose you were to ask the postgraduates in the coursework higher degrees at Griffith University about the course, as I did recently. The first two people might mention (as they did) having to organise time or organise work. You may tentatively code these sentences as "organising" (perhaps among other codes). As you do this, be aware of any theoretical ideas that come to mind. If any do, note them down immediately. For easier sorting later, I use 125 mm x 75 mm systems cards. They fit in my pocket and are very convenient. I carry a pocketful around with me most of the time. [ index ] o Categories and properties In effect, a category is a theme or variable which makes sense of what your informant has said. It is interpreted in the light of the situation you’re studying, and other interviews, and the emerging theory. o In the two sentences considered above, I’ve already mentioned "organising" as a tentative category. What is different between the two sentences is this: one is about organising time, one about organising work. Perhaps this will be a property, a sub-category, of organising. [ index ] o Core category After a time one category (occasionally more) will be found to emerge with high frequency of mention, and to be connected to many of the other categories which are emerging. This is your core category. It is hazardous to choose a core category too early in the data collection. However, when it is clear that one category is mentioned with high frequency and is well connected to other categories, it is safe to adopt this as the core category. (If more than one core category emerges, Glaser recommends focussing at one time on one only. You can recode for the second of them later, if you wish.) o All five of the postgraduates I talked to in my miniature example mentioned the use they were making or would make of what they were learning. If became evident before long that this one category (titled "application") fitted the two criteria of frequency of mention and high connectedness. When a core category has been identified, you cease coding any sentences which do not relate to it. You will find that in most instances your coding rapidly becomes more efficient as the study progresses. You now code for the core category, other connected categories, and properties of both. You record any identified connections between categories in memos. You continue doing this, adding to your sample as necessary (see sampling, below), until you achieve saturation. [ index ] o Saturation In collecting and interpreting data about a particular category, in time you reach a point of diminishing returns. Eventually your interviews add nothing to what you already know about a category, its properties, and its relationship to the core category. When this occurs you cease coding for that category. [ index ] Sampling Your initial sample is likely to be defined by your choice of research situation. If there are many people associated with the situation, you might begin by putting together as diverse a sample as you are able. (I don’t recall anywhere that Glaser offers a clear description of the beginning sample, though I may be mistaken there.) As categories emerge from your data, you then seek to add to your sample in such a way that you further increase diversity in useful ways. Your purpose is to strengthen the emerging theory by defining the properties of the categories, and how those mediate the relationship of category to category. Glaser and Strauss refer to this as theoretical sampling. The sample is emergent, as is the theory and the method generally. o The small group of postgraduates I talked to were either studying part-time and working, or had worked at some stage. One might expect a category such as "application" to be influenced by work experience. I could therefore have added usefully to the sample by identifying and talking to people from the program who had never worked. [ index ] Memoing I have mentioned already that memoing continues in parallel with data collection, note-taking and coding. In effect, a memo is a note to yourself about some hypothesis you have about a category or property, and particularly about relationships between categories. Glaser makes the point, and I agree, that memoing is given high priority. As an idea occurs to you, pause in what you are doing and write a memo to yourself. I carry a pocket full of 125 mm x 75 mm system cards in my pocket most of the time, for jotting down memos. In time your core category and the categories related to it will have saturated. By the time this happens you will have accumulated a large number of memos. Between them they will capture the different aspects of the theory which has emerged from your data. o In the example, early memos might record hypotheses that "organisation" and "application" were categories. Another memo might question if "present application to work" and "future application to work" might be properties of application. A further memo might hypothesise that application is a core category. Another memo might query if organisation is important at least in part because it may lead to better application. In short, in using grounded theory methodology you assume that the theory is concealed in your data for you to discover. Coding makes visible some of its components. Memoing adds the relationships which link the categories to each other. The next task is to decide how you will structure the report to communicate your theory to others. That is the purpose of sorting. [ index ] Sorting My reason for using cards for memoing is twofold. They are easier to carry, so I can jot down ideas whenever they occur to me. They are easier to sort. For the actual sorting I work on a large table or on the floor. First I group them on the basis of the similar categories or properties they address. I then arrange the groups to reflect on the sorting surface their relationship. The intention is that their layout in two-dimensional space will capture the structure of the eventual report or thesis. I then gather the cards in the sequence which will allow the structure to be described. This provides the basis for the writing up, which follows. [ index ] Writing up Having done all this -- coding, memoing, sorting -- the writing is less a chore than it might otherwise be. The sort structure is the report structure. It is often just a matter of preparing a first draft by typing up the cards in sequence and integrating them into a coherent argument. [ index ] The place of literature There are two important points to be made about the literature. The first is that, in an emergent study, you probably won’t know at the beginning which literature will later turn out to be relevant. This has implications both for the place of reading in your own research process and for your report. The second is that the literature is not given a position of privilege when compared to the data. It is treated as data, with the same status as other data. [ index ] o Literature as emergent Most people embarking on a research project will first examine the relevant literature. Thesis candidates often do not begin data collection until well into their candidature. In an emergent study you can begin collecting data as soon as you have a research situation. You can then access literature as it becomes relevant. Glaser (especially 1978) makes much of the prior background reading which provides the models to help make sense of the data. He recommends reading widely while avoiding the literature most closely related to what you are researching. His fear, which I share, is that your reading may otherwise constrain your coding and memoing. At the same time, I think this approach may leave you vulnerable to criticism from examiners or referees or colleagues. The defence is to take special pains to be responsive to the data, to seek disconfirming evidence assiduously, and to defend by careful argument your decision to do so. Reading later is less an issue for Glaser. My own view is that it makes sense to access relevant literature as it becomes relevant. Most examiners and colleagues will expect you to locate your study within the relevant fields of literature. You can also reach a wider sample, in effect, by refining your findings in the light of the literature in slightly different but related fields. In short, a progressive accessing and reading of relevant literature can become a part of your data collection procedures. [ index ] o Literature as data Constant comparison remains your core process. Your aim as you read is to compare literature to the emerging theory in the same way that you compare data to the emerging theory. For instance you might follow the same procedure of data-collection (in this instance reading) overlapping with note- taking, coding and memoing. Whether or not you do precisely this, the key issue is how you treat apparent disagreement between your emerging theory and the literature. You don’t assume that your theory must be wrong. After all, you have been concerned throughout with its fit to the data and its ability to make sense of actual experience. You seek to extend the theory so that it makes sense of both the data from your study and the data from the literature. This issue -- of treating disagreement appropriately -- has been a focus of some of my own work on rigour in action research. It is in fact possible to substitute some action research procedures for some of all of data-collection, note-taking, coding, memoing and sorting. [ index ] A variation based on action research I research my own practice as educator, facilitator and consultant. The methods I use were developed until recently entirely independently of grounded theory. I wasn’t familiar with its literature. When I did eventually start to read that literature I was pleased at the obvious parallels between the two approaches. Let me illustrate this by describing how I approach ... let’s say, organisational diagnosis, using interviewing. I’ll do this in such a way that the parallels are evident. I think you will find that the parallels are such that you can substitute parts of one for parts of the other. [ index ] o Convergent interviewing In diagnostic interviewing (see my 1990 monograph) I begin in a very open- ended way. For instance I may often say "Tell me about this organisation" or whatever it is. I then keep the person talking for somewhere about 45 minutes without asking specific questions. This increases the likelihood that the data come from the informant's experience, not from the questions I ask. I memorise the themes they mention (some of my colleagues instead take key-word notes, which serve the same purpose). I prefer to work with a colleague who at the same time interviews a different informant. After each pair of interviews we compare notes. We identify those themes which both informants mention. Sometimes those themes are mentioned in the same way by both informants. Sometimes they mention the same theme, but with disagreement. o I was evaluating an action learning program with Karyn Healy, a colleague. Many informants mentioned that they weren’t provided resources which allowed them to pay someone to do their work, to free them up for their action learning. An example of agreement might be two informants saying words to the effect that their action learning was done in their own time, which they both resented. A disagreement would be when both mentioned doing it in their own time, but one of them mentioned this with satisfaction, not dissatisfaction. When we identify an agreement we devise probe questions to seek exceptions to the agreement. o For example, we might ask if there were people who didn’t resent the intrusions on their own time. We might ask if there were advantages to being able to devote their own time to the action learning projects. When we identify a disagreement we devise probe questions which seek explanations for the difference. o For instance we might say something like this ... "Many people have mentioned taking part in the action learning in their own time. To what extent was this your experience? How did you feel about that? Some have mentioned this with substantial resentment. Others seem not to mind. Can you help us understand how this difference might arise?" As with grounded theory the explanations emerge gradually from the data as the study proceeds. All interviews begin open-ended. In the later interviews there are more probe questions. And more of those probes are specific. The theory emerges from the data, from the informants. In the early stages it consists primarily of themes. These become more elaborated as the study develops. This is depicted diagrammatically below. I suggest that, in deciding your own methods, you choose those which will be easiest for you to defend to examiners, readers or colleagues. What you do is probably less important than how well you justify it. [ index ] Qualitative psychology This is for those attempting to use grounded theory in psychology. Many of you will barely have heard of qualitative methodology, let alone learned how to use them. However, there is a growing, if small, band of people exploring the application of qualitative methods in psychology. I’ve included some of them in the bibliography below. It may help you with your supervisors, committee or examiners if you can demonstrate that there is an alternative tradition to which your studies are a contribution. [ index ] Contribution to knowledge I have no doubt, by the way, that you will have a contribution to knowledge at the end of it all. The theory will arise more quickly than you imagine. You’ll enjoy doing it (Glaser calls it the "drugless trip"). There is a good chance it will be an addition to the literature because most psychological research builds on what has gone before. You, on the other hand, are going to be responsive to the research situation as it is. You are going to find out what is really happening there. By some chance you may discover a theory only to find someone else has come up with the same theory using more traditional methods. You still have a contribution to knowledge, and a valuable one. You have cross-validated, using a very different methodology, the theory previously offered.
Pages to are hidden for
"Grounded+Theory"Please download to view full document