Research Methodology
4.5 Data Collection and Analysis
4.5.5 Data Analysis
Following each interview, the researcher summarised the interview findings and arranged a second interview to confirm or to clarify the points he had recorded. In addition, after completing all interviewees for a single case, in cases in which there were contradicting results, either between two different interviews or between an interviewee and other sources of data, the researcher sought additional interviews or advanced secondary sources to ensure that the information was accurate. The researcher also summarised the interview outcomes and sent them back to the interviewee to confirm his understanding, to ensure accuracy of written data and to enable a better collection and use of evidence, as well as to enhance the level of reliability (Kirk and Miller, 1986; Ritchie and Lewis, 2010; Creswell, 2008).
Once all the data had been collected, the researcher transcribed all of the data gathered for each case from interviews, observations and document study. He organised the words and phrases and made notes on associated non-verbal behaviours, and began the process of coding the data, line by line, seeking patterns. This was followed by the division of the findings into themes or trends and transferral to tables or matrices using Miles and Huberman’s (1994) model. The researcher iterated over the data
several times to ensure the accuracy of the themes and to enhance reliability, as suggested by Creswell (2008). The revisiting and rechecking data was also intended to minimise the personal interpretation bias, which could be a significant limitation generally, and in IS qualitative research (Walsham, 1995). A process referred to as bracketing was employed, which involved the researcher separating his views and beliefs from those of the participant; this process was supported by the researcher not making quick judgements based on the first transcript and hence keeping a more open mind about the findings (Bernard and Ryan, 2009). The matrices assisted the researcher in the comparison of the four cases in addition to providing a diagrammatic representation of each one.
The Ranking Approach
For a better presentation of the comparative analysis of the implemented ERP projects within the four cases, a ranking approach will be used. Using this approach, the results of the qualitative analysis of the eight implemented projects are classified as ‘excellent, very good, good, fair and poor’. These five values have been given a numeric value ’80-100, 60-80, 40-60, 20-40 and 0-20’ respectively. This ranking approach allows us to present presenting the qualitative data in a graphical form so as to suggest various relationships between comparative elements. These potential relationships can be further tested in a quantitative manner to improve the theory in managing roles and responsibilities in ERP implementation context.
Secondary data was also sourced from prior academic research studies, in order to supplement the findings of this study, to provide a basis for comparison, and to enhance the validity of the findings through triangulation. This was particularly effective in the stakeholder context, providing deeper explanations, facilitating comparison and assisting in the analysis of success and failure elements.
Limitations of Data Collection and Analysis
Studying stakeholders of ERP throughout implementation phases required choosing the longitudinal case study approach, in order to capture organisational experience from initiation to the post implementation phase, including the close monitoring of organisational behaviour during this latter phase. However, time constraints did
emerge that inhibited this study from capturing more details and suggesting more detailed and solid recommendations (Johnson et al. 2009)
The small number of cases that were employed to investigate the manner in which the 'soft' issues of implementing ERP effectively, which centred on the roles and responsibilities of the stakeholders involved, and the fact that the experiences of different organisations varied meant that generalisability towards all organisations was limited.
The scope of the study was somewhat limited by the number of people-related issues that could be uncovered and explained during the field and desk research. This reflected the greatest limitation, which was the time that the researcher had to plan and complete the thesis, as well as the restrictions on organisational documentation that was made available by each firm involved. The time that participants in the study were able to devote to providing their views and understanding of the ERP implementations process, was also a limitation on the accomplishing a truly representative view, even within the confines of the four organisations. The participants' experiences were restricted to the period in which they had been employed by the firm; some employees had left during the implementation phase so that experiences could not be accessed and others had joined the organisation at some stage during implementation and could therefore provide time-limited data.
The openness of some participants was likely to have been limited by fear of revealing certain feelings and opinions that might be detrimental to the organisation in which they were employed, or with which they were associated for the completion of the project, as well as to their position within it. In other cases participants may have provided restricted input in terms of what they felt the researcher wanted them to say or what he should be told. There was also the potential agenda that admitting any misjudgements in any stage of the ERP process would be viewed as critical of certain personnel as well as the danger of revealing confidential information. The researcher also had inherent bias, owing to his knowledge and experience in the sector and hence his values and beliefs (Walsham, 1995).
The principle of an ethical approach to research was that participants should not be negatively affected, as a consequence of their involvement in it. Hence the participants in this study were ensured confidentiality regarding their input and that no individual could be recognised from any comment made in this thesis. The researcher did not name the companies, which agreed to take part, nor did he reveal details that would make them recognisable.
The researcher obtained permission from the companies who agreed to participate in the study and he made no contact with potential participants prior to that permission being granted. The researcher also made the senior personnel of the companies aware of the university’s ethical guidelines that he was obliged to follow. The locations for the interviews were chosen to ensure privacy.
All participants were provided with written details of the purpose of the study and the confidentiality of the information that would give to the researcher. Participants were informed that they could withdraw from the study at any time and refuse to answer any question, which they did not feel comfortable to discuss. Participants were requested to sign a written declaration concerning these points. All notes and documents related to the thesis were retained locked cabinets and the associated electronic data was securely, password protected (Ritchie and Lewis, 2010; Saunders et al. 2009).
4.6 Summary
This chapter aimed to describe and justify the research methods chosen during the course of this study, and to establish the aims and objectives of the investigation. The primary objective of this study is to establish a holistic framework for ERP implementation and to enable ERP decision makers to enhance the management of key stakeholders. As a result, the purpose of this research is exploratory, as it considers ERP implementation from a stakeholder perspective. As described above, the relative paucity of available data and previous studies in the field of ERP implementation demanded a particular research philosophy and approach to data collection. The adoption of both interpretivist and positivist approaches enabled a focus on the diverse opinions and beliefs of the individuals under study to be
the necessary flexibility demanded by the research problem and permitted the researcher to adapt quickly during the course of data collection, as the need arose. Furthermore, the use of case studies permitted the researcher to conduct an in-depth analysis in a genuine ‘real life’ context, drawing on multiple data sources across each organisation. By focusing on four specific cases that exhibited significant common factors in addition to profound differences, the data gleaned from interviews was maximised in scope and permitted a greater degree of generalisation necessary for the formulation of a potential model. The data collected was primarily subjective, which, as acknowledged above, is problematic in the formation of general models. However, it is expected that by setting this research in the context of other studies, sufficient validation may be established to draw effective conclusions.
In conclusion, it is suggested that the primary demands of this study required a holistic, flexible approach, which justifies the methods adopted. It should be noted, however, that a more expansive study accommodating more cases studies is desirable to validate the conclusions drawn from this investigation. In addition, this study must be situated in the context of other available data regarding ERP implementation, which may provide further validation of the conclusions drawn in subsequent chapters.