3.3.1 Overview
Grounded Theory, as presented by Glaser and Strauss in 1967, and subsequently developed by Glaser and by Strauss and Corbin (1998), came out of research from the 1960s, concerning a series of studies carried out into the dying process as experienced by hospital patients and the nurses who worked in that environment. Prior to grounded theory, research methods in social sciences tended to focus mainly on how to verify theories. GT offers a middle ground between the extreme empiricism of the positivism and complete relativism by offering a middle ground where systematic data collection is used to develop theories that explain the interpretive realities of those actors in their organisations’ who have to make sense of the construct of CSR
Glaser and Strauss (1967) wanted to encourage researchers to use their intellectual imagination and creativity to develop theories relating to their areas of inquiry, to suggest methods for doing so, to offer criteria to evaluate and work on discovered theory and to focus on generation rather than justification (Locke 2001). Grounded theory rejects a priori theorizing; rather it focuses on research and discovery through direct contact with both the subject and the environment being studied. This focus leads to knowledge being an emergent phenomenon free from the constraints that trying to fit around a priori theory. This does not mean that a researcher should embark on the process of grounded theory without some sort of orienting theoretical perspective; however these should not obstruct the development of theories by coming between the researcher and the knowledge (Partington 1998, Suddarby 2006).
Page 116 Indeed, GT when rigorously applied requires the researcher to have a detailed understanding of the key issues in the subject area lest the focus be on irrelevant pieces of data from the interview set. Glaser and Strauss, whilst agreeing that testing theory is an important part of research and that research generating theory goes hand in hand with verification of it, were concerned that in the social sciences there had been a concentration on testing either existing theories or theories that researchers had barely begun to generate (Glaser and 1967).
The purpose of Grounded Theory (GT) is theory construction rather than description or application of existing theories and in this thesis GT will be used to build and support the case studies that will be discussed in more detail in section 3.10 where the research design of this specific study will be the focus. Grounded theorists engage in data collection and analysis simultaneously in an iterative process that uses comparative methods (Charmaz 2006). This method analyses actions and processes rather than themes and topics. A defining strategy is theoretical sampling i.e.
sampling for developing the properties of a tentative category and not for ensuring representation of a sample of people with a specific demographic characteristic. – again this will be examined in more detail in this chapter. Grounded Theory provides a very structured method to focus on contemporary issues in qualitative research and whilst it is a clear method (Strauss 1998, Glasser 1967) it can also be used as a broad strategy for analysis (Saunders et al 2003). This use of GT as a strategy has been identified as problematic as a stringent process must be followed in the coding process so to use it simply as a guiding strategy (as opposed to a method) is likely to lead to a study that lacks rigour.
(Saunders, Lewis and Thornhill, 2003 p 398) .
Grounded theory rejects a priori theorizing; rather it focuses on research and discovery through direct contact with the social world being studied.
This focus leads to knowledge being an emergent phenomenon free from the constraints that trying to fit around a priori theorizing brings (Strauss 1998). Glaser and Strauss, whilst agreeing that testing theory is an
Page 117 important part of research and that research generating theory goes hand in hand with verification of it, were concerned that in the social sciences there had been a concentration on testing either existing theories or theories that researchers had barely begun to generate (Glaser 1999).
Grounded Theory focuses on the difference between substantive and formal theory. Whilst formal theory is the ultimate goal, it must be developed from a substantive grounding in concrete social situations, substantive theory being that developed for a substantive or empirical area of inquiry, and formal being developed for more formal or conceptual areas of inquiry, suggesting that substantive theory comes before formal.
In the study of management many of the theories are substantive e.g.
leadership or decision making, and into this category might fall CSR.
More formal theories mean according to Locke (2001) those inquiries that operate at a high level of generality (e.g. agency theory).
Glasser and Strauss did deviate in their perspectives and approaches to the use of Grounded Theory leading to subtle but important differences in the process and application of the method. Easterby-Smith, Thorpe and Jackson (2008, p. 101-102) stated that Glaser took a more realist approach and would remain independent vis-à-vis the already existing data. Strauss & Corbin (1990) on the other hand adopted a continuous interrogative analysis of the data with the establishment of a coding process and would take a flexible approach against arising data and be open-minded (Easterby-Smith, Thorpe and Jackson, 2008, p. 101-102).
This slightly more constructivist approach to GT offers a more accessible form of GT. Unlike the earlier and more Glassarian view of GT the Straussian approach acknowledges the influence of the researcher, accepts the notion that multiple realities may exist and rejects the assumptions that the researcher should (or even could) set aside their prior knowledge in developing new theories. Regardless of the debates between the Glaser and Strauss perspectives on GT, both agreed that a GT approach builds up a theoretical framework which demonstrates mutual links (Parker and Roffey, 1997)
Page 118 While the differences between the two approaches have been positioned as significant and overall a more Straussian approach has been adhered to through the interpretation of the data gathered there is still much of the original Glassarian structures and perspectives underpinning the application of the method in this thesis.
Bailey, White and Pain (1999, p. 174) wrote that the openness and continuous evaluation of data is central key to a GT as well as critical thinking. The scholars reiterated that a GT approach necessitates a continuous interpretative approach, and that it is a process (Bailey, White and Pain, 1999, p. 176). This interpretative process is the dominant approach throughout the research, starting from the literature review, the analysis of the collected data through its overall discussion.
Suddaby (2006) elaborated on the GT methodology giving a warning to researchers of what a GT is and is not. Suddaby (2006) revealed a series of misconceptions around a GT. First, not considering existing literature and prior knowledge is a mistake (Suddaby, 2006, p.634). Suddaby suggest that there is a misconception that GT involves the researcher being a ‘blank sheet devoid of experience or knowledge’. He further suggests that that ignoring relevant and pre-existing literature is likely to lead to a ‘mass of descriptive material waiting for a theory or a fire’
(Coase 1988 p 230). Suddaby suggest that the GT method is not simply a random collection of raw data but focuses on the casual effects between actors (Suddaby, 2006, p. 635). Third, GT is not a simple theory for the sake of being a theory; it is rather an interpretive approach of a series of different contexts (Suddaby, 2006, p. 637).
3.3.2 The Process
Glaser and Strauss (1967) suggest a constant comparative methodology consisting of a 4-stage approach to generating theoretical explanations from qualitative data. Due to the access that can be gained to actors in
Page 119 this industry this would appear to be a most relevant method of examining the data generated by semi structured interviews to generate theory.
3.3.3 Comparing Incidents Applicable to Each Other
The aim is to assign multiple observations a common meaning that is captured or composed in a conceptual category. As a start point the data incidents that have been collected are used with a view to understanding a particular substantive problem. Grounded Theory insists that any preconceived notions, theories or expectations are suspended (this can be done through a critical examination of values and biases discussed later). At this stage the researcher must try to develop abstract meaning for these data incidents by articulating what they believe is happening or being expressed. This involves studying an incident in the data set and giving that incident a name that represents an interpretation of what is happening with regard to that incident. The naming of this incident is not restricted to a single name; it can be names in as many ways as there are interpretations of the incident. In tandem with the naming process comes comparing. This activity helps to develop a common name or category for multiple observations and helps to clarify what we perceive. An example of how this works is given by Locke (2001) that although using examples straight from Glaser and Strauss’ original work and not from the perspective of CSR, give an excellent overview of how the technique is applied. The transferability of this technique to interviews concerning CSR, sense making and benefits should not pose any difficulties. The example given by Locke concerns Nurses reactions in Glaser and Strauss’ study of their reaction to patients dying
Researchers scrutinize a data fragment in their field notes that represents the nurses response to the death of a patient and identify the nurses comment ‘what a loss - he wanted to be a teacher’ (1) Locke (as do Glaser and Strauss) asks how this might be coded and how can we name what is being expressed? Locke suggests ‘death as a loss’ (A), ‘unfulfilled ambitions’ (B) or ‘society denied a contribution’ (C), we then seek other
Page 120 incidents where nurses responses are captured e.g. ‘well at the age of 85 with 7 grandchildren he had had a full life’ (2) ‘she had 4 little children, what will happen to them now that she is gone’ (3) and we compare them to our labels A, B, C. Locke suggests that at first glance all appear to be accommodated under A, 1& 3 perhaps suggest an element of someone or a group being denied, so perhaps (C). Thus by comparing incidents with each other and with initial labels, an examination of data incidents may well reinforce that nurses are assigning value related meaning to patient deaths. As more comparison is done, so the meaning might be refined. Glaser and Strauss refined the meaning to ‘Social Loss’ (D). In parallel, the data incidents are compared for what might be different – e.g. (2) might be seen as less important. This points to various properties of the category e.g. whilst all may have (D) in common, there may be different representations of it e.g. the young mother being a more significant loss than the 85 year old. This helps us to define theoretical properties of the category e.g. how staff arrive at a determination of loss, whether it is high or low etc. The data might suggest age as a factor thus a search might be made to discover other expressions of age and to understand and articulate the role that age might play in determining social loss. The conceptual categories must thus earn their way into the framework by being persistent and recurring. During this process (and indeed subsequent to it) notes or memos must be kept as to why categories are chosen, what constitutes a category, why particular labels are used etc. This informs reasoning at later stages as well as giving validity to the process.
It is possible to see how this process might be utilised in interviews concerning CSR, how the conceptual labels might be arrived at from particular data incidents and the types of categories that might arise, although until interviews are underway pre-emptive categorisation might be seen as amounting to a priori theory which is contrary to the Grounded Theory approach. Validity becomes an issue when using grounded theory. To ensure both internal and external validity it becomes important to ensure that both the transcription and interpretation are accurate. As
Page 121 such interview transcripts and interpreted meaning need to be confirmed by the interviewee.
Glaser and Strauss give further guidance to assist with the coding process suggesting that the coder asks a series of questions
(a) What is happening, (b) What is the basic problem being faced by the actors here, (c) What category or aspect of as category does this incident suggest and (d) What does this incident suggest that this is a theory of?
The sampling technique used will, in line with grounded theory, be purposive sampling. Whilst it could be argued that due to the close geographic proximity the amount of access to managers and operatives in the Lincolnshire Co-operative Society (LCS) amounts to convenience sampling, as it is likely that those interviewed will have some connection with programmes being run at the University of Lincoln’s Business School however such are the numbers attending that purposive sampling is still possible and access will be negotiated through LCS directly and not via registered students. The fact that access is relatively easy to employees at all levels of the LCS means that the sample can be constructed to take into account all of the various stakeholder groups and can also be constructed to give the best insight into the phenomenon under consideration.
As we have noted, Grounded Theory has been used in the past to investigate CSR (Morimoto 2004). Glaser and Strauss (1967) stressed that Grounded Theory is a particularly useful tool for providing explanations when researchers are confronted with substantive issues where they have no real theories. This does not mean that the researcher can somehow ‘fumble around in the dark’ in the hope that Grounded Theory will miraculously provide explanations, theories and insight (Coase 1988). Glaser and Strauss assume that the researcher has already a clear purpose to the study and the issues that it is hoped to illuminate as well as the practices that it might influence. Grounded
Page 122 Theory assumes that the research question has been arrived at and that a strategy is in place for gathering information.
The aim is to assign multiple observations a common meaning that is captured or composed in a conceptual category. As a start point the data incidents that have been collected are used with a view to understanding a particular substantive problem. At this stage the researcher must try to develop abstract meaning for these data incidents by articulating what they believe is happening or being expressed. This involves studying an incident in the data set and giving that incident a name that represents an interpretation of what is happening with regard to that incident. The naming of this incident is not restricted to a single name; it can be names in as many ways as there are interpretations of the incident. In tandem with the naming process comes comparing. This activity helps to develop a common name or category for multiple observations and helps to clarify what we perceive. Once the interviews had been conducted, a detailed analysis of the content was carried out and the data was coded so that it could be immediately compared with what had gone previously to establish themes and connections. Thus point A1.1 was compared to A1.2 and A1.3 was compared to A1.1 and A1.2. Appendix 1 gives an indication of the open codes utilised from one interview and are summarised below
A1.1: Culture
A1.2: Strategic Pillars
A1.3 Define and Develop Values
This means that from the first interview began to develop themes of culture and values. This process was repeated up to point A1.34 in the first interview and then the process was repeated for interview A2, through to A52
A2.1: Traditional Values A2.2: Generating power
A2.3 Efficiency of the generating process A2.4: Embedding the culture
Page 123 and similarly through B,C,and D. Whilst the range of data labels becomes so large that any meaningful appendix containing all of them is not possible, to give an overview, an abridged range of categories from one of the interviews has been included in the appendix 1
3.3.4 Integrating Categories and their Properties
This stage aims to develop and provide organisation for the conceptual categories in order that they can account for similarities and differences in the data incidents. This stage also allows progress in the formulation of our conceptual scheme e.g. the perception of one initiative can be compared with the impact of a different initiative. And any links or underlying value not yet identified can be established. In order to arrange our categories that they might add up to a theoretical framework, the arrangement of the conceptual elements relative to each other becomes important – thus allowing clarification of the relationships between the categories and their properties.
These were compared in the same way that the categories in A1 were done and then compared across A1 and A2. This same process continued up to A41 and generated between 10 and 40 categories per interview. Once the entire A categories had been exhausted the same process was repeated with B, C, D and E. Each one of these headline categorisations representing different organisations. The process generated many hundreds of key points from which concepts began to emerge. The first concepts that began to emerge came from the individual interviews but these became themes across all interviews.
Concept 1: Understanding of values and culture A1.1, A1.2, A1.3, A2.1, A2.4 …D5.1, D5.3
Concept 2: Making sense
Concept 3: What constitutes important initiatives?
Concept 4: Paradigm Concept 5: Benefits
Page 124 From these began to emerge the categories of concepts that shared similar characteristics and the language that was associated with them.
There began to appear broad categories of strategic importance of initiatives and interventions. Had the process simply continued in this way then some interesting findings would have been generated and connections between categories would have been established.
Once the initial open coding had been completed the analysis then began looking for the relationships between the categories of data that had emerged from the open coding process. The essence of this stage is to explore and explain the phenomenon being investigated by an examination of what is happening and why. This stage considers the environmental factors that impact the process and what the outcomes of these interactions might be. Once these relationships have been identified it is possible to cross check them again against the data that had been collected (Strauss and Corbin 1998, Corbin and Strauss 2008).
Having competed this process the final element is that of selective coding. Selective coding identifies the principal categories that are core to the study (Corbin and Strauss 2008)
The analysis of the data revealed 8 Selective or Final Codes from the data. These codes were:
1. Creation of Meaning and Sense making 2. Definition
3. Activities and Focus 4. Beliefs
5. Paradigm and Shared Value 6. Structure
7. Business Case and Measures 8. Culture and Leadership
The codes were arrived as through the GT process of constant comparison of the data as it was uncovered. As GT has not previously been used to examine CSR via multiple case studies there was no
Page 125 template for the coding process. This proved not to be problematic because GT is premised on the detailed and constant comparison of the data to allow the development of firstly the open codes and then the axial codes. Whilst the axial codes varied from case to case and were seen to be more subject to the external influences present at the different times that the research was carried out this was not evident when the selective
Page 125 template for the coding process. This proved not to be problematic because GT is premised on the detailed and constant comparison of the data to allow the development of firstly the open codes and then the axial codes. Whilst the axial codes varied from case to case and were seen to be more subject to the external influences present at the different times that the research was carried out this was not evident when the selective