Case Study Research Design
2.2 A Rationale for Using Case Studies as the Research Method Research Method
The case study research method was selected over alternatives for three key reasons60.
First, the research is a continuation of another case study and the
decision for utilising the Commission’s report was taken at an early stage in the research. Utilising this kind of qualitative evidence was determined to be the best route for accomplishing the research objectives: the investigation is bound by a consistent theme and provides an extremely rich and detailed historical narrative of the marketing practices of the various national manufacturers in natural settings (Vella and Foxall 2011). In addition, the report was generated and compiled by an unrelated third party and for purposes different to those that govern this research. Therefore, the evidence is not biased by the researcher’s own theoretical perspectives, expectations, assumptions, and otherwise in any way. A robust method was required to handle this kind of qualitative
evidence61.
Second, the case study is an ideal method for specifying and delineating key dimensions of the subject of theoretical or empirical interest as a bounded system and intensively studying a single or a relatively small sample of cases of that phenomenon and its dimensions (Stake 1995; Lee 1999; Gerring 2004;
George and Bennett 2005; Bryman and Bell 2007; Gerring 2007; Flyvbjerg
60 Within this research, a case study is defined as a method, i.e., a formal and distinctive approach to empirical inquiry with its own unique designs (Stake 1978; Eisenhardt 1989; Stake 1995; Lee 1999; Hammersley and Gomm 2000; Berg 2001; Rowley 2002; George and Bennett 2005; Eisenhardt and Graebner 2007; Yin 2014). Each research method is best suited to accomplish different tasks to the extent that methodological choices should be dictated by research objectives and the degree to which each method is suited best to accomplish those goals and provide suitable answers to research questions (Eisenhardt 1989; Crotty 1998; Berg 2001; George and Bennett 2005; Bryman and Bell 2007; Creswell 2009; Yin 2009).
61 Appendix A2.1 provides a brief rationale why the case study method was selected in favour of alternative methods on its capacity to robustly handle historical qualitative evidence.
2011; Yin 2014)62. The method locates cases within a particular context (Stake 1995; Lee 1999; Gerring 2004; Bryman and Bell 2007; Gerring 2007; Flyvbjerg 2011; Yin 2014) signalling the dynamics, complexities, and ambiguities of the real world contexts of behaviour (Campbell 1984b; Eisenhardt 1989; Dyer Jr and Wilkins 1991; Stake 1995; Lee 1999; Gerring 2004; George and Bennett 2005; Gerring 2007; Yin 2014)63. Cases are also considered across time thereby providing a temporal or historical essence to case studies (Stake 1995;
Lee 1999; Bryman and Bell 2007; Gerring 2007; Flyvbjerg 2011). There is an emphasis on the unfolding nature of interrelated events surrounding the case providing a more holistic perspective (Flyvbjerg 2011; Yin 2014): “Because the contextual conditions may interact in subtle ways with the case, a good case study should therefore lead to an insightful understanding of a case and its internal as well as its external complexity” (Yin 2014, p. 209).
Third, the historical emphasis of case studies makes the method very appropriate for both the research objectives and the retrospective nature of operant and evolutionary interpretations. The method possesses a
characteristic propensity to induce a very detailed investigation from the
perspective of a case sampled from a population and located within specific real world contexts as events gradually or dramatically unfold over time (e.g.,
George and Bennett 2005; Yin 2014). As discussed hereunder, historical evidence is imperative in demonstrating operant conditioning and in framing the research in evolutionary terms.
62 The extent to which each unit may be studied intensively depends on the number of cases being studied – the lesser the number of cases considered, the deeper the case study will be (Dyer Jr and Wilkins 1991; Hammersley and Gomm 2000; Gerring 2007). As shall be seen in Section 2.3.1C, the present case is a single case study. One associated trade-off is that single case studies may not allow for any meaningful comparisons across different units (e.g., Dyer Jr and Wilkins 1991). This may either weaken the theory that is being developed since more cases provide cumulative evidence in its support, rejection, or refinement (Eisenhardt 1989) or simply render the theoretical conclusions as tentative (Dyer Jr and Wilkins 1991, p. 615) and,
therefore, as the basis for future research (George and Bennett 2005).
63 In contrast to experimental methods, which create a controlled, simplified, and contrived setting and wherein researchers have direct control over variables, case studies capture real world contexts and, typically, do not involve any control over variables (Hammersley and Gomm 2000; Gerring 2007; Yin 2014). The choice of a case setting, however, should not be mistaken with the selection of a manipulated setting (cf. Gerring 2012). Experimental investigators have direct control over the architecture of the setting. Case study researchers do not. Case studies, however, may be designed to gain some of the advantages of experiments (see Section
2.3.1D).
From an operant perspective, the reason why an individual acts the way she does within a given context – that is the meaning of behaviour – lies in providing an account that describes the intersection of her learning history of reinforcement and punishment with the present context of behaviour (Foxall 1995c, a, 1997b, 1998b). In other words, the meaning of an emission is expressed in terms of the antecedents of the individual’s behaviour (learning history which summarises the characteristic effects or consequences or function of particular emissions in the past), the signalling function that the stimulus events within the current environment perform because of that history, and the consequences of behaviour as signalled by these antecedents (Foxall 2010b).
Thus, the essential components in constructing behaviourist explanations from the perspective of the individual lies in examining both her learning history (as a personal variable) and the behaviour setting (as an extra personal variable) (Foxall 1995c, a, 1997b, 1998a, b, 1999c, 2005, 2010b)64. The explanatory core of the BPM is a construct termed the situation, which allows the deepest possible examination of situated behaviour (Foxall 1996b, 1997b, 1998b, 2005, 2010b) – the learning history of the individual denotes the temporal dimension whereas the behaviour setting marks a spatial dimension (Foxall 1990,
1997b)65. These constructs provide a thorough analysis of what is taking place at a specific point in time.
A central requirement for producing an interpretation of real world behaviour within an overarching evolutionary understanding relates to
accessing data that features dynamics, processes, socio-economic interaction within relationships, and cumulative outcomes and change over a relatively
64 In this sense, once a learning history is constructed and located within a particular immediate behaviour setting, the meaning of behaviour (the reason why the individual behaves the way she does) is “subjective,” i.e., from the perspective of the individual participant. Behaviourists hold and deploy a different understanding of and approach to what constitutes the perspective of participants within a specific context. An analysis of behaviour relies only on observables and subjective meaning is constructed in term of those observables only. The reason why an individual acts the way she does within the current context – that is the meaning of behaviour – lies in providing an account that describes the intersection of her history of learning with the current setting wherein behaviour occurs (Foxall 1995c, a, 1997b, 1998b). In contrast, cognitive approaches define subjective meaning in terms of unobservables focusing on individual
perceptions, “intentions, choices, objectives, values, perspectives, needs, desires, and agency”
(Miles et al. 2013, p. 222). Thus, “qualitative data are not so much about behaviour as they are about actions, which carry with them intentions and meanings and lead to consequences” (Miles et al. 2013, p. 11).
65 The construct is examined in greater detail in the explanation of the BPM as it relates to consumer and marketer behaviour in Chapter 4, Section 4.2 and Section 4.3.
substantial history of inter-related events (e.g. Nelson and Winter 1982; Barnett and Burgelman 1996; Marengo and Willinger 1997; Nelson and Winter 2002;
Dosi and Marengo 2007; Morlacchi and Nelson 2011; Bairstow and Young 2012; Lavie and Singh 2012; Dosi 2013). Ideally, this history should be continuous. Evolutionary research incorporates a time scale that is long
enough to demonstrate related processes and outcomes over generations and emphasises a complete coverage of the historical events experienced by a given population (Aldrich and Ruef 2006).
From an evolutionary perspective, learning history (and the regularities or stable contingency relations it summarises, i.e., rules) is analogous to the
biological genotype (Foxall 1993b, 1997b, 2010b). Within evolutionary
economics the genotype is interpreted in terms of ‘quasi-stable properties’ with a regulatory dimension (e.g., Hodgson 2003; Hodgson 2008, 2009a, b;
Hodgson and Knudsen 2010). From a consumer’s perspective, learning history represent ‘habit’ or the potential for the continuity of behaviour within sufficiently similar behaviour settings. Learning history holds replicator function (Foxall 1993b, 1997b, 2010b). Similarly, in application to the firm, learning history is analogous to the genotype of the firm and maintains replicator function. The phenotype is analogous to the actual and observable behavioural emissions by the firm within the current environment given the genotype. In this research, environmental interaction is characterised by behavioural emissions within mutually reinforcing social relations among firms and stakeholders (termed as bilateral contingencies) against a backdrop of contingencies describing natural phenomena (e.g., seasonality of the ice cream trade due to the vagaries of the weather and climatic conditions in the UK)66. These constructs emphasise stability and change over time, the explanatory core of an evolutionary perspective.
Figure 5 provides a diagrammatic representation of the Commission’s report. The evidence presented therein allows a distinction between two separate situations that are bound historically and, from an evolutionary perspective, are assumed to depict empirical snapshots of individual
66 These points will be elaborated further in Chapter 3 (Section 3.3 and Section 3.4) and Chapter 4 (Section 4.3).
generations of lineages of marketing behaviours and environmental contingencies67. Demarcating generations is an integral dimension of evolutionary explanations to: (a) distinguish between the changes occurring during a single iteration of variation, selective retention and elimination, and inheritance replication (within-situation analysis), and (b) understand the extent to which change is carried forward from a single iteration to the next (cross-situation analysis) (Metcalfe 1998; Becker 2001; Pepper and Knudsen 2001;
Knudsen 2002; Metcalfe 2005).
Figure 5 – Distinct Generation-Situations for Analysis and Comparison
Within-situation analysis allows examination of the proximal or immediate dynamics of selection by environmental consequences in a given generation.
That is, generation of the phenotype, the interaction of the genotype within the specific environment. Cross-situation analysis allows the examination of the nature of distal or remote features, processes, and effects of selection by environmental consequences over generations. That is, the extent to which
67 Refer to footnote 6 in Chapter 1 (Section 1.1.1 on page 13) for definitions of lineages and notional generations. “The challenge is to connect changes in the two sets of … variables;
changes in the distribution of phenotypes must be linked with changes in the distribution of genotypes … [This] boils down to two essential and independent sub-processes: (1) differential survival because of phenotype-environment interaction, and (2) the perpetuation of phenotypes to the next generation through the replication of genotypes. What has been spelled out may be termed evolution (cumulative causation over generations) by means of natural selection, and it is the accumulated effect over numerous generations which makes the account evolutionary”
(Knudsen 2002, p. 448).
Analysis and Interpretation of Retail Marketing Practices across Space (Distinct Situations characterised by Social Bilateral Contingencies and Physical Contingencies) and Time (Learning History of Wall's and
Environmental Change across Generations) the Weather, Refrigeration Technology and so
on. the Weather, Refrigeration Technology and so
on.
some phenotypes survive and propagate over generations due to replication of the genotype.
Case studies are important methods useful in uncovering and explaining causal relations, mechanisms, and outcomes in situ (Miles and Huberman 1994; Lee 1999; George and Bennett 2005; Gerring 2007; Yin 2009; Miles et al.
2013; Yin 2014). If carefully planned and conducted in a valid and reliable way, case studies have the capacity to deliver in-depth investigations of how causal mechanisms actually operate in the real world (Stake 1995; Lee 1999;
Hammersley and Gomm 2000; George and Bennett 2005; Gerring 2007;
Gibbert et al. 2008; Yin 2009, 2014)68.