3 Research Methodology
3.5 Trustworthiness
A major criticism of case study research has been related to its perceived lack of rigour (Stuart et al. 2002). This concern that has been addressed extensively by Yin (1989) who outlines four logical tests for judging case research regarding its construct validity, internal validity, external validity and reliability. This research aims to address all of these areas, as they are also a concern in logistics research (Mentzer and Flint 1997, Aastrup and Halldórsson 2008). However, this thesis follows the work of Lincoln and Guba and uses their criteria for the evaluation of qualitative research, namely trustworthiness and authenticity (Guba 1981, Lincoln and Guba 1985, Lincoln and Guba 1986). This section outlines how trustworthiness is achieved here. Trustworthiness consists of four criteria, namely credibility, transferability, dependability, and confirmability.
Credibility is of particular importance in critical realist research since it stresses
the existence of multiple accounts of an aspect of social reality thus making it important that the findings are established to be credible (Bryman 2012). This has been achieved through respondent validation, seeking corroboration of the findings by gaining feedback from the participants in the study. Credibility parallels internal validity in the evaluation of quantitative research (Guba 1981, Lincoln and Guba 1985, Lincoln and Guba 1986). As indicated by Yin (1989), key informants have been asked to review the draft of the subsequent chapters that detail the findings and their discussion to ensure the correct understanding and interpretation. Comments from these informants have then been integrated into subsequent versions of those chapters. Internal validity is concerned with making proper inferences from the data (Ellram 1996) and the identification of causal relationships (Yin 1989, Aastrup and Halldórsson 2008), and thus of particular importance in case study research under a critical realist paradigm (Sayer 2000). It is suggested to conduct “pattern matching” (Yin 1989), demonstrating that actual data patterns match predicted patterns, further strengthening the confirmation by replicating these findings across similar cases (literal replication) or showing that these patterns do not exist in dissimilar cases (theoretical replication) (Stuart et al.
2002). This thesis presents the findings from several cases, thus allowing for theoretical replication. However, due to the messy nature particularly of the social activities within the open systems under investigation (Sayer 2000), prediction of patterns is not seen as expedient (Aastrup and Halldórsson 2008). While internal validity is thus not a primary concern, due to the abductive research approach, there is a constant iteration between proposed and identified patterns, between theory and empirical data, and in RRREI the Elimination phase deals with competing generative mechanisms to make sense of the data, thus adding to validity (Spring and Santos 2015). However, the key focus here is on credibility, which has been achieved through respondent validation.
Transferability parallels external validity (Guba 1981, Lincoln and Guba 1985,
Lincoln and Guba 1986), which is concerned with “the domain to which a study’s
findings or presumed causal relationships may be generalized” (Stuart et al. 2002, p.
430), a major aspect of the overall criticism aimed at case study research because of the generally small sample size (Yin 1989, Gomm et al. 2000, Schofield 2000). Unlike survey research, case study research does not rely on statistical generalization, but on analytical generalization similar to experiments (Yin 1989). Case selection is not based on representative random samples (Eisenhardt 1989). Generalization does not occur from sample to population as in the case of statistical generalization, but from each case to a broader theory, which can—in the extreme—be possible from just one case as it uncovers reality (Stuart et al. 2002, Easton 2010). External validity is an issue to be addressed in the design phase of each study; here it is addressed by conducting multiple case studies and verifying patterns in these (Ellram 1996). However, from a critical realist perspective generalization is not a primary concern, and if external validity matters at all, it is for reasons of transferability and contextualization rather than generalization (Tsoukas 1989, Aastrup and Halldórsson 2008). Following Lincoln and Guba, the approach taken here is to focus on the contextual uniqueness of the cases studied, but aiming to produce thick description, a detailed account of the cases that can enable researchers to make an informed judgment about the possible transferability of findings (Guba 1981, Lincoln and Guba 1985, Lincoln and Guba 1986). This is achieved here through an extensive and detailed findings chapter that expresses the richness of the primary data and a detailed understanding of the cases.
Dependability parallels reliability in quantitative research (Bryman 2012), and is
suggested to entail an auditing approach with high importance placed on keeping complete records of the research process (Guba 1981, Lincoln and Guba 1985, Lincoln and Guba 1986). This strongly resembles reliability, which is concerned with the
repeatability of the study and the ability to achieve the same results (Kidder and Judd 1986, Ellram 1996). Dependability is achieved here by maintaining a case study database, through the electronic storage of all data and notes that clearly document the research process and thus would allow future researchers to repeat the analytical procedures (Stuart et al. 2002). Furthermore, a case study protocol has been developed including the interview guide, as well as the procedures to be followed in using it (Ellram 1996). Despite the accurate record keeping, there has been no auditing process in place as only one researcher was involved in this research. This is in keeping with standards in qualitative research, where auditing is very rare because of its demanding nature (Bryman 2012).
Confirmability is concerned with ensuring that researchers have acted in good
faith and not overtly allowed personal bias to sway the conduct of the research and reporting of the findings deriving from it (Guba 1981, Lincoln and Guba 1985, Lincoln and Guba 1986). This parallels reliability, which is concerned with the repeatability of the study and the ability to achieve the same results (Kidder and Judd 1986, Ellram 1996). As recommended (Stuart et al. 2002), this chapter has provided a detailed description of the data collection process. Furthermore, a clear chain of evidence has been established and the data analysis is extensively documented (Ellram 1996). This opens the work up to close examination and increases transparency (Ketokivi and Choi 2014, Spring and Santos 2015). Confirmability is achieved here by maintaining a case study database, through the electronic storage of all data and notes that clearly document the research process and thus would allow future researchers to repeat the analytical procedures (Stuart et al. 2002). Furthermore, a case study protocol has been developed including the interview guide, as well as the procedures to be followed in using it (Ellram 1996). The review of drafts by key informants is also seen to have contributed to a reduction of bias and to confirm that research was conducted in good faith.
3.6 Summary
This chapter has addressed matters of research philosophy and methodology. It provides the background to the primary research conducted as part of this thesis. In particular, it addresses questions of rigour and quality in the research process. Table 11 summarises key aspects of research philosophy and methodology in this thesis. A discussion of different research paradigms and the underlying epistemologies and ontologies has been provided at the beginning of this chapter, resulting in the
justification of the choice of the critical realist paradigm and the interpretivist epistemology and realist ontology it incorporates. In line with the critical realist paradigm, an abductive research approach was taken. Primary data was collected in 44 semi-structured interviews, and a case study approach was utilized, with an initial round of interviews and then three case studies of humanitarian responses that emerged from the previously gathered data in a casing approach. Interviews were conducted face-to- face, over the telephone or online; all interviews were recorded and transcribed to then be entered into NVivo for data analysis. Analysis took place through several rounds of coding, and occurred on a case basis, as well as across cases. Informed consent was given by all participants and anonymity was guaranteed throughout the study. A selection of key informants commented on drafts of the findings and discussion that is presented in the subsequent chapters.
Table 11: Summary of key aspects of research philosophy and methodology in this thesis Research Paradigm Epistemology Ontology Critical Realism Interpretivist Realist Research Approach Research Methodology Research Method Abductive
Multiple case studies Semi-structured interviews Unit of Analysis Development of Cases Timeframe Data Analysis CAQDAS Humanitarian response Casing Cross-sectional
Thematic analysis, cross-case analysis NVivo
Based on the process outlined here, primary research was conducted; data was gathered and analysed. Chapter 4 presents the findings from this data, split into the initial round of interviews and the three case studies on different humanitarian responses. Each humanitarian response is described in greater detail at the beginning of the relevant section of Chapter 4. RRREI forms the basis of this analysis. Chapter 4 is organised foremost by case studies and then by the first level codes employed in the data analysis, which correspond to the five elements of the conceptual framework for MSCs previously established. At first, each case is analysed individually, with cross- case analysis occurring in Chapter 5, thus answering a call for more cross-organisational
research in the context of HL (Kunz and Reiner 2012), as well as the more general call for more empirical research (Kovács and Spens 2011).