ANEXO I: CONDICIONES GENERALES DE LAS
13. FUERZA MAYOR; OTROS CAMBIOS EN LAS CONDICIONES REINANTES:
The strategy for presenting the analysis and discussion in Chapter VI is founded on the Harding (2013, p. 4) framework. Harding emphasised four approaches relevant for the analysis of qualitative data: thematic analysis, comparative analysis, content analysis, and discourse analysis. The latter option, discourse analysis, is not utilised for this study, however, the remaining three approaches are adopted and can be defined as follows:
Thematic analysis requires the researcher to identify themes emerging from the literature Comparative analysis is used to compare findings across different industry or customer
participants
Content analysis requires a more systematic approach to sift through the data by creating codes In the context of this study, the Harding framework is employed to create a funnelling effect and narrowing down of the case findings. Thematic analysis is applied to concepts such as SCM, SCO and strategic orientation. This is followed by comparative analysis, comparing case company and other interview participants’ main findings. Finally, content analysis is deployed to examine and interpret
the interview data in relation to the literature findings and gaps identified in the literature (presented in Chapter II).
In terms of analysis tools, a Google Scholar search was conducted in 2011 to identify the extent of author interest in qualitative analysis tools specific to SCO research. The majority of SCO studies were in the logistics, supply chain and marketing research fields. Of the 124 SCO studies identified and examined, the majority (74%) utilised quantitative methodologies and there was little evidence of qualitative research, with even less evidence of computer assisted qualitative data analysis software (CAQDAS) packages such as NVivo or Atlas ti. being utilised. The search findings are presented in Table 4.13.
RESEARCH DESIGN NUMBER OF STUDIES
(2000-2010)
Case Study 117
AHP (Analytical Hierarchy Process) 9
NVivo or qualitative software 1
Table 4.13: Dissemination of research methods for studies in supply chain orientation (SCO)
The PhD Tools considered for analysis in this study included Atlas ti. V6, NVivo, Analytic Hierarchy Process (AHP) and Fuzzy Analytic Hierarchy Process (AHP). Each were explored by reviewing existing debates in the literature, consulting with the supervising team, reading weblogs, visiting QAS websites, interviewing both fellow doctoral research students and experienced researchers. In addition, two training courses were taken: a one day introductory course in July 2009 at Cardiff University, School of Social Sciences and an advanced three day training web seminar in May, 2012 (by Ricardo Contreras).
All computer assisted qualitative data analysis (CAQDAS) packages seemed to allow the researcher to store and manipulate text documents and codes. Early (1990s) software packages had been criticised as mere organising tools (Smith and Hesse-Biber, 1996). More current CAQDAS packages, such as Atlas had been recommended as time-saving devices, with the proviso that the coding process was carefully planned (Contreras, 2012). A new text by Friese (2012) became a major source of information and guidance.
Atlas ti. V6 was chosen for this research study. Its strengths were that it was a systematic approach to collating, sorting, coding and analysing data. The updated version of Atlas (7) was emphasised as radically different and time consuming to make the transition (Contreras, 2012).
Bazeley (2008) argued that to increase rigor in quantitative research, qualitative software packages should be considered to assist in improving the research design and the flexibility of the research. One area where rigour is required is in the coding approach, which Bazeley, (2008) argued could fundamentally strengthen (or weaken) the arguments being presented. The manual coding option has been subjected to criticism (Friese, 2011), being referred to as being a less robust methodology. Therefore, in support of a more effective qualitative analysis, the systematic coding of data and ordering of the codes was undertaken for the analysis. A systematic coding approach has been argued by authors as important for promoting research validity (Bazeley, 2008; Friese, 2011). In agreement with Yin (2014), this part of the process was felt the most time consuming, but it did help in developing an understanding of the importance of the ordering and placement of the variables in the research model. This coding process was then matched to the findings from the literature review in Chapter II to verify the order of precedence between variables, such as leadership, trust and communication.
Yin (2014) raised caution that that software packages are mere tools for analysis and rely heavily on the researcher to design the formulae and Friese (2012) stressed that if the researcher tries to structure the data too soon this will result in problem later into the research. Friese argued that Atlas software operates on two levels:
Textual level – promoting the coding, searching and memo writing to start the analysis; Conceptual level – linking the codes to become semantic networks in relation to key concepts. Appendix 9 details the process of using Atlas ti. V6. One of the main benefits of using Atlas ti. V6 was that it helped to visualise the development model as the codes were being sorted and data analysed (Smit, 2002). These are visually and clearly presented on the screen. Other advantages included the ability to utilise numerous documents which could be bundled together under the Hermeneutic Unit (HU) and stored as a single unit. This simplified the selection process using a single file. Secondly, personal documents (PDs) such as PDF files and text documents could have quotations filed without corrupting the original documents. This was useful if errors were made, as changes could be easily overridden. Furthermore, any section of the PDF document (e.g. journal paper) could be coded: text or graphics. The compatible format of PDF allowed several types of documents to be included, such as, PowerPoint, Excel and Word and Open Office. Atlas ti.offered a central portal for the interview documentation gathered during the data collection period (2009-2011). This was useful because notes could be stored to underline or link key concepts under examination and by working with the data on the computer screen primary documents (research reports) could be switched for easier comparisons (Mayring, 2000) did though present problems. For example, when using data files as external sources for the PDs, source files had to remain at the referenced location and could not be renamed or moved,
otherwise the HU could not find it. Friese (2012) stressed it is important not to alter any documents even outside of the HU incase the files become corrupted or invalid.
To start the analysis process, Friese (2012) proposed three stages of the NCT model: notice, collecting, thinking (Figure 4.11). Each of these three stages is explained in Appendix 9, which also details the process undertaken of using Atlas ti. V6 in this study, and applied in Chapter VI for the data analysis and discussion.
Figure 4.11: NCT Model of data analysis (developed from Friese, 2012, p.57)
Appendix 9 lists the process and tools for Atlas software. Case study findings for Manufacturer A and Manufacturer B, Supplier, Supplier B, Retailer A and Retailer B were written up into reports and these were downloaded onto the hermeneutic unit (HU). The term, hermeneutic derives from “the art of fortune-telling and text interpretation” (Friese, 2012, p.9). The sentence content within the reports was broken down to manageable pieces of text in the process of identifying the meaningful comparisons of data. Content analysis has been supported by authors such as Weber (1990, p. 5) for classifying “…large sections of text and organising it into relevant and manageable bits of data”.
A screen shot of the Atlas desktop can be seen in Figure 4.12. Smit (2002) and Friese, (2012) underlined the importance of appropriate coding as essential for theory building. The codes were initially created using the open coding system, with codes being derived from the literature and in-vivo codes, which are in turn derived from the textual content. In-vivo codes are necessary when quotations fall easily into headings: for example, market orientation.
N Noottiicciinngg interesting things from
such as industry reports, flip charts, excel spreadsheets
C Coolllleeccttiinngg these documents and filing them to the HU T
Thhiinnkkiinngg about how these relate
to the research model by searching and linking key concepts
Figure 4.12: Screen Shot of the Hermeneutic Unit (HE)
Knowing how many codes to create is more of a challenge. Friese (2012, p. 93) stressed that “some researchers develop 40 codes, others a few hundred or even a few thousand”. Friese suggested the coding process should be completed in two cycles. However, running through the coding process for this research identified the need for a third cycle. This ensured the codes were reduced to a manageable number for enabling easier linking to the literature. This research process utilised a combination of descriptive, topic and an open coded set of procedures.