In defence of the case study methodology for research into strategy practice

Vaara and Whittington (2012) trace the origins of the wider practice perspective back to the works of Wittgenstein (1951) and Heidegger (1962). Wittgenstein is viewed as having the most significant influence upon the placement of the practical dimension of philosophy front and centre of philosophical debate in modern times (Deslandes, 2011). The foundations established by Wittgenstein’s work were built upon by Heidegger, with the former’s assertions being untangled and given structure by the latter (Chia, 2006). Subsequent years saw a proliferation of practice-based theories across the social sciences (Vaara and Whittington, 2012), with its emergence in the strategy field occurring in just the past couple of decades. The suggestion was that strategy research could be advanced through the adoption of social theories of practice (Hendry, 2000; Hendry and Seidl, 2003), most prominently those of Giddens (1976), MacIntyre (1985), Foucault (1980), Bourdieu (1990), and De Certeau (1984), and the influence of this work is evident in the present-day literature (Chia, 2006; Deslandes, 2011; Splitter and Seidl, 2011). Research into SAP has drawn upon a diverse foundation of practice theories emanating from within the social sciences. However, from a review of the extant SAP literature, it would appear that activity theory (Jarratt and Stiles, 2010; Jarzabkowski, 2003; Johnson et al., 2003) and structuration theory (Feldman and Orlikowski, 2011; Jarzabkowski, 2008) have gained the most traction amongst the pre-eminent authors.

SAP is ‘concerned with the doing of strategy; who does it, what they do, how they do it, what they use and what implications this has for shaping strategy’ (Jarzabkowski and Spee, 2009: 69). It is heavily focussed on the social dimension of the strategy-making endeavour by ‘bringing human actors and their actions and interactions to the centre stage of strategy research’ (Jarzabkowski and Spee, 2009: 70). In targeting the everyday minutiae of what strategy practitioners actually do, strategy may be understood as something people do rather than something that firms in their markets have (Jarzabkowski and Seidl, 2008: 1391). The key tenets of the practice perspective are its ‘3Ps’ of practice, praxis, and practitioners (Whittington, 2006). The level at which praxis transpires is categorised into micro, meso, and macro levels. Strategic episodes and the wider social environment both form additional signposts for the collection and analysis of data during a SAP-based study. The following sections provide an overview of each of these integral constructs.

Strategy practices are viewed by Jarzabkowski and Whittington (2008: 101) as involving the ‘various routines, discourses, concepts and technologies through which this strategy labour is made possible – not just obvious ones such as strategy reviews and off-sites, but also those embedded in academic and consulting tools (Porterian analysis, hypothesis testing, and so on) and in more material technologies and artefacts (PowerPoints, flip-charts, and so on)’. In a sense, they provide the boundaries within which strategic activity occurs, imbuing behavioural, cognitive, procedural, discursive, and physical resources upon strategists (Jarzabkowski et al., 2007). However, they do not enforce rigid boundaries but rather allow for iteration and adaptation (Vaara and Whittington, 2012). Thus, there exists an entwinement between strategy practices and strategy praxis, with practitioners playing somewhat of a mediating role between both concepts and the extent to which each shapes the other.

At its broadest level, praxis refers to the ‘vast social enterprise of day-to-day activity’ (Campbell-Hunt, 2007: 794). It comprises the ‘interconnection between the actions of different, dispersed individuals and groups and those socially, politically, and economically embedded institutions within which individuals act and to which they contribute’ (Jarzabkowski et al., 2007: 9). Such a definition is highlighted as being pertinent, as it captures both how practice can be operationalised at multiple levels, i.e. from macro to micro, as well as its inherent adaptability as it shifts through the interactions between these levels.

‘Strategy’s practitioners are defined widely, to include both those directly involved in making strategy – most prominently managers and consultants - and those with indirect influence – the policy-makers, the media, the gurus and the business schools who shape legitimate praxis and practices.’ (Jarzabkowski and Whittington, 2008: 101-102)

Practitioners are interrelated with both practices and praxis as it is the prevalent organisational practices that engender strategy practitioners with agency in regard to how they act and behave (Jarzabkowski et al., 2007). Practitioners represent a suitable unit of analysis for a SAP-based study, as it is they who shape the strategy that will ultimately determine the future direction of the organisation.

SAP-based studies may employ a multi-level perspective upon praxis, with three such levels identified within the SAP literature (Jarzabkowski and Spee, 2009). Micro refers to studies concerning an individual’s or a group’s experience of a particular episode, i.e. a decision or a meeting. Meso refers to studies located at an organisational or sub-organisational level, i.e. a change programme or a strategy process. Macro studies explore praxis at an institutional level, i.e. attempting to explain praxis within a specific industry.

Within the SAP literature resides the recurrent theme of the strategic episode. Incidents of strategic episodes are an explicit component of Whittington’s (2006) framework. An episode can be defined as a ‘sequence of communications structured in terms of its beginning and ending’ (Hendry and Seidl, 2003: 176), the beginning and ending of which is not only observable by the external observer but also by the communication within the sequence. Indeed, conversational and discourse analysis have proven valuable research topics within the SAP domain due to their apparent linkages to the wider social component of strategising (Balogun et al., 2014; Hardy and Thomas, 2014).

A SAP lens allows researchers to more closely examine and describe exactly what is involved in organisational phenomena as they unfold (Sandberg and Dall’Alba, 2010). However, it also encapsulates the grounding of these micro activities within their wider social context (Bürgi et al., 2005; Palmer and O’Kane, 2007), i.e. ‘actors in their micro-situations are not acting in isolation but are drawing upon the regular, socially defined modes of acting that arise from the plural social institutions to which they belong’ (Jarzabkowski et al., 2007: 6), thus linking the micro to the macro context (Oliver and Bürgi, 2005; Regner, 2008; Seidl and Whittington, 2014; Whittington et al., 2011). There exist explicit calls for the adoption of a SAP perspective within the SISP domain (Teubner, 2013), as it has the potential to result in the design of better strategising processes and practices (Bryson et al., 2010).

When referring to SISP, it is the planning or formulation stage of IS strategising that is paramount. The remit of SISP research is elucidated by Segars et al. (1998; quoted in Brown, 2004: 20), who define its focus as being ‘on the strategic to tactical level’ and that ‘in terms of scope, it is organisational; in perspective of top management, in terms of level of abstraction, more conceptual than physical, and in time frame, medium to long’. Ward and Peppard (2002: 118) proffer a similar definition, viewing it as ‘thinking strategically and planning for the effective long-term management and optimal impact of information in all its forms: information systems (IS) and information technology (IT)...’ SISP is more complex than an exercise, during which the technical merits of an array of alternatives are evaluated. Rather, in congruence with SAP studies, the social dimension of the process (Avgerou and McGrath, 2007) is of particular importance due to the disparate organisational backgrounds of those involved. This lends itself to examination under the ‘sociological eye’ (Whittington, 2007) of a practice-based perspective, and thus highlights the inherent compatibility evident between both research domains.

While there exists evident entanglement across the various streams of literature as it pertains to IS strategy at large, a delineation is provided by Chen et al. (2010) through their categorisation of three closely related avenues of research. They include SISP (Brown, 2004; Merali et al., 2012; Teubner, 2013), the alignment of business strategies and IS strategies (Chan and Reich, 2007; Henderson and Venkatraman, 1993; Kearns and Lederer, 2000), and the use of IS for competitive advantage (Eng and Luff, 2011; Okpattah et al., 2014; Piccoli and Ives, 2005).

As alluded to by Chen et al. (2010), the boundaries between each literature set are blurred. While a large body of the reviewed SISP literature focusses upon SISP tools, the practitioners involved in SISP, and the organisational traits conducive to effective SISP, so too does a significant portion of literature pertain to planning for IS strategic alignment or planning for the use of IS to obtain competitive advantage. This is not surprising as each of these dimensions can be viewed as a core SISP component and the complete disaggregation of each of these three streams of literature would prove an unrealistic goal due to the inherent commonality between each of them (Chen et al., 2010). Such a finding is supported by Earl (1989, 1993), who proposes that SISP targets the alignment of investment in IS with business goals and the exploitation of IS for competitive advantage. Lederer and Sethi’s (1996: 35) definition of SISP as being ‘the process of identifying a portfolio of computer-based applications that will assist an organisation in executing its business plans and realising its business goals’ also corroborates this finding.

Remaining true to the very essence of a practice perspective to research, i.e. proclivities towards the actual ‘doing’ of strategy work on the part of practitioners, the dominant research methodologies are conducive to an in-depth analysis of the focus of investigation (Table 1). Of the 48 empirical papers reviewed featuring some form of primary data collection, either a single-case study (21) or a small number of case studies (17) were the dominant approaches employed. Ethnographic studies accounted for a further nine studies, again highlighting both a requirement and a desire on the part of researchers to get as close as possible to the data.

Table 1

Analysis of the sap literature

The prescriptive literature often points to the suitability of an ethnographic approach to practice-based research (Chia and MacKay, 2007; Fenton and Langley, 2011; Jarzabkowski et al., 2007; Miettinen et al., 2010; Rasche and Chia, 2009; Whittington, 2004). However, the prevalence of ethnographic studies does not quite match the sentiment of goodwill expressed towards its applicability to practice-based research, an anomaly most likely manifested out of the inherent difficulties associated with conducting an ethnographic study. Such difficulties include gaining the requisite access to candidate organisations, the immense personal commitment required of the researchers during prolonged periods of immersion within the field, the validity of the research approach itself, and the practical fear of having to justify oneself to reviewers and editors for departing from the more conventional research approaches of survey or interview-based research (Myers, 1999; Watson, 2011).

Finally, just a single survey-based study was conducted by Gunn and Williams (2007), which again highlights the inherent bias towards a qualitative approach within practice-based research.

The substantially larger body of empirically grounded SISP literature in comparison to its SAP counterpart (126 versus 48, respectively), combined with its more expansive time frame, allowed for a more substantial macro-level analysis to be conducted. Examination of the featured 126 empirical studies from a methodological perspective yields some interesting findings (Table 2).

Table 2

Methodological analysis of the SISP literature

While a number of reviews of IS research at large (Avison et al., 2008; Chen and Hirschheim, 2004; Hirschheim and Klein, 2012; Orlikowski and Baroudi, 1991; Palvia et al., 2007; Vessey et al., 2002; Williams et al., 2009) have produced varying results due to the specific literature sets under review, i.e. journal specific (Palvia et al., 2007; Avison et al., 2008) or region specific (Chen and Hirschheim, 2004), the current review of research focussed upon SISP produced encouraging results. It would appear that earlier calls for a more varied approach to IS research (Galliers, 1993; Mingers, 2001; Orlikowski and Baroudi, 1991) have been somewhat answered, with the dominance of the positivist, survey-based approach being somewhat eschewed within the SISP domain. This is perhaps understandable due to the nature of the focus of investigation and the inherent difficulty associated with quantifying the phenomenon. Thus, it lends itself to exploratory research of a qualitative nature, and hence the prevalence of case-based research is logical. The findings are in tandem with the work of Hirschheim and Klein (2012), which also found a growing acceptance within the IS research community of interpretive research approaches using a variety of qualitative methods.

Within the reviewed literature, the case study approach is the most prevalent, accounting for 50 studies. A closer inspection reveals that single-case studies (27) slightly out-matched a multiple-case study design (23). The survey approach – with 49 studies – was narrowly eclipsed into second position. Mixed methods (e.g. Applegate and Elam, 1992) proved to be the third most frequently employed methodological approach, accounting for 12 studies. Finally, research employing either some form of qualitative approach (8), such as focus groups or interviewing (Lederer and Mendelow, 1989), or quantitative methods (7), such as statistical analysis (Barua et al., 1991), completed the literature set.

The scope of a case study is defined by Yin (1994: 13) as follows: ‘A case study is an empirical inquiry that investigates a contemporary phenomenon within its real-life context, especially when the boundaries between phenomenon and context are not clearly evident’. Gerring (2004: 342) believes a case study is best defined as ‘an intensive study of a single unit for the purpose of understanding a larger class of (similar) units’. Eisenhardt (1989: 534) simply defines the case study as ‘a research strategy which focuses on understanding the dynamics present within single settings’.

Case studies are a commonly employed empirical strategy in IS research (Lee and Hubona, 2009; Orlikowski and Baroudi, 1991) and have become increasingly popular within the field (Doolin, 1996; Steventon et al., 2012). They can be used to achieve various research aims, inclusive of providing rich descriptions of phenomena in context and developing or testing theory (Darke et al., 1998; Eisenhardt, 1989). Case studies are acknowledged to possess an advantage when engaging in research that is exploratory in nature (Gerring, 2004). It has become an increasingly attractive approach for researchers looking to investigate the development, implementation, and use of IS within organisations (Darke et al., 1998). Theory building from case-based studies is growing in acceptance (Hoon, 2013), with a large number of influential works emanating from the approach (Eisenhardt and Graebner, 2007). As an exploratory study, the research also benefits from the potential richness of the accrued data, which is a major strength of the case-based approach (Siggelkow, 2007). Kuhn (1987) opines that a discipline without a substantial body of thoroughly executed case studies is a discipline without a systematic production of exemplars, and that such a discipline is an ineffective one.

The value of the research is dependent on more than just the number of cases chosen. The quality of the cases selected and the researchers’ analytical and investigative abilities are of equal, if not greater, importance (Perry, 1998). The choice of case sites should involve discretion and judgement (Amaratunga and Baldry, 2001). Cases should be selected for their ability to contribute to the overall investigation rather than a logic predicated on convenience (Poulis et al., 2013; Stake, 1995). The identification of a population of possible case study sites is a good starting point and is crucial to the case selection process. An appropriate population ‘helps to define the limits for generalising the findings’ (Eisenhardt, 1989: 537). From the finalised population, theoretical sampling may be used for case selection, i.e. cases are chosen for theoretical, not statistical reasons (Glaser and Strauss, 1967), with the goal being to extend the emergent theory (Eisenhardt, 1989). Given that each case would emanate from the population of Irish public service organisations, the study’s case selection criteria were as follows:

A key consideration in the design of a research strategy is whether to employ a single- or a multiple-case approach (Yin, 2009). A single case can allow for the collection of extremely rich data due to its explicit focus on just one research site (Franklin et al., 2014), being particularly appropriate for extreme or exemplary situations to which the researcher has gained access (Zivkovic, 2012). However, such an approach is not without its potential drawbacks. A number of such concerns are raised by Lee (1989) and Gable (1994), with a single case study’s lack of generalisability (Tsang, 2014) identified as the principal concern with such a design (Van de Ven and Huber, 1990). Conversely, a multiple-case approach may diminish the potential richness of the data collected with each additional case added, but equally increases – in theory – generalisability as more cases are incorporated into the study. Multiple cases typically provide a stronger base for building theory (Benbasat et al., 1987; Yin, 2009), tending to result in ‘more robust, generalisable, and testable theory than single-case research’ (Eisenhardt and Graebner, 2007: 27). A multiple-case study design also allows for cross-case comparison (Darke et al., 1998; Tsang, 2013), which further contributes to the building of rich theory (Benbasat et al., 1987). Perry (1998) depicts the many differing views in regard to the ideal number of cases, ranging from two at a minimum up to 15 at the maximum end of the scale.

Two cases were decided upon for the current study. The researchers believe two cases strike a reasonable balance between having the necessary breadth from which to build credible theory and also allowing for the depth to pursue rich data due to the exploratory nature of the research. From the researchers’ review of the SAP literature, it is apparent that studies of this nature have a strong tendency to feature either a single or a small number of case studies. The reasons for such were realised during data collection and analysis. More than two cases could have resulted in data management issues: the employed two cases proved challenging by themselves due to the large quantity of data accrued for each case site. The researchers’ viewpoint, having now completed the study and being in a position to stand back and view it from a more objective position, is that an unnecessary quest to increase the number of cases would have proved detrimental to the quality and depth of the research. The researchers were unwilling to compromise on this attribute of the study.

Multiple data collection methods are normally employed during case-based research (Benbasat et al., 1987; Venkatesh et al., 2013). Table 3 details the sources of evidence that can be utilised during casework and those used for the current study. This multi-method approach ameliorates the validity of the results due to the triangulation of findings (Eisenhardt, 1989). It can also prove to be highly synergistic by improving data accuracy and limiting the researchers’ biases (Miles and Huberman, 1994). Sources of data include organisational documentation, archival records, interviews, direct observation, and physical artefacts (Benbasat et al., 1987), with the principal goal being to collect rich data that accurately represent the phenomena under investigation.

Table 3

Data collection: sources of evidence (adapted from Yin, 2009)

Table 4 provides an overview of the secondary data sources collected for the current study. The multi-level data collection (Kozlowski et al., 2013) approach employed is commensurate with the adopted SAP lens, representing a further use of Jarzabkowski and Spee’s (2009) construct of levels.

Table 4

Categorisation of secondary data sources

Data sources were categorised based upon specified levels, i.e. the European context, the public service context, the healthcare context, and the context of both host organisations. This rudimentary yet effective SAP-derived classification scheme proved a valuable organising mechanism for the wealth of collected data. Returning to the language of practice, the sources of secondary data from each of the European context, the public service context, and the healthcare context can be classified as emanating from the macro-level. The secondary data specific to both case sites are located at the meso-level. Thus, this process serves as an example of the SAP lens’s pervasive role during the execution of the current study. The comprehensive, structured collection of secondary data provided a solid foundation from which to build upon during the course of subsequent fieldwork.

The study then employed semi-structured interviews (Mojtahed et al., 2014) that adhered to an agreed-upon protocol in relation to recording techniques, the descriptive process, whether or not feedback will be provided, and additional ethical concerns (Myers and Venable, 2014; Myers and Newman, 2007; Stahl et al., 2014). Guidelines for the conduct of the interview process were obtained from Kvale (1996), Myers and Newman (2007), and Hermanowicz (2002). Interview questions possessed characteristics of a SAP perspective in their focus upon the 3Ps of practice, praxis, and practitioner (Seidl and Whittington, 2014; Whittington, 2006), with each forming broad topics for discussion as they pertained to the practice of SISP within each host organisation.

In a review of case-based research in four leading IS journals from 1989 to 1993, Doolin (1996) discovered interviews to be the primary method of data collection. Interviews are also the most widely used qualitative data collection method, being extensively used in an array of disparate disciplines (Schultze and Avital, 2010). In their study of IS research articles that relied primarily on interviewing methods conducted from 2004 to 2008, Schultze and Avital (2010) found that there was a near-three-fold increase in the frequency in which the method was deployed over the time frame of the study, increasing from 12 papers in 2004 to 35 in 2008. As opined by Hermanowicz (2002: 480), interviews have an ability to ‘bring us arguably closer than many other methods to an intimate understanding of people in their social worlds’, making it particularly appropriate for generating rich, qualitative data.

The coding strategy employed was that proposed by Maykut and Morehouse, (1994). Having spent time considering various coding techniques (Charmaz, 2006; Corbin and Strauss, 2008; Glaser and Strauss, 1967; Maykut and Morehouse, 1994), it became clear that all of them possessed an inherent similarity; phases predominantly involved an initial sorting of the data, consolidation and ordering of categories, and the search for linkages between and across higher-order categories from which the theory could emerge. The selected strategy was found to possess such traits and be highly rigourous in its approach.

Maykut and Morehouse’s technique is largely based on the coding principles of grounded theory (Glaser and Strauss, 1967). It involves nine discrete cycles of analysis. At each stage of the analytical strategy, the researchers retained the use of the adopted SAP lens and its inherent levels perspective.

The use of a SAP perspective in conjunction with the case study methodology represents a deliberate, reasoned decision. There exist explicit calls for the adoption of a SAP perspective from within the IS domain (Peppard et al., 2014; Teubner, 2013). The recognition and uptake of the approach is evidenced in the 2014 special issue of the Journal of Strategic Information Systems, with Whittington’s (2014) commentary in particular applauding the SISP and SAP joint agenda. Such publications elucidate the distinction bestowed upon the approach by the IS academic community. Further encouragement may be found from related disciplines to have adopted a practice approach, inclusive of leadership as practice (Raelin, 2011), accountancy as practice (Kornberger and Carter, 2010), business ethics as practice (Clegg et al., 2007), and marketing as practice (Browne et al., 2014).

Evidence exists within the extant SAP literature of the fortitude of the perspective. Johnson et al. (2007) proffer four such advantages. First, in its purest form, the practice perspective sheds light on the actual ‘doing’ of strategy, enriching our knowledge of the ways in which strategising actually takes place (Vaara and Whittington, 2012). Since this is what managers in a learning environment are concerned with mastering, it is highly beneficial to executive education. Second, it removes the level of abstraction associated with traditional strategy research, thus going deeper into the core issues within the field. Third, as demonstrated in Whittington’s (2002, 2006) framework, a practice perspective can act as an integrating mechanism for the field of strategy research, whereby the central components can be interconnected into a coherent whole. So not only are more micro-level processes explored and detailed, but their linkages to the bigger macro-level picture may be traced in addition to the role of the strategy practitioners therein (Vaara and Whittington, 2012). Finally, the fourth advantage is a culmination of the previous three, in that strategy-as-practice research offers a new and exciting direction for the field. It unshackles itself from some of the constraints inherent in some of the more traditional strategy research approaches, allowing strategy practitioners to assume a more central role on the research agenda.

Researchers are enamoured with a staunch belief that the choice of a SAP perspective as a tool to incorporate within the wider study’s case study methodology is of merit. This view is supported by the growing popularity of the practice-as-a-perspective mode, its inherent suitability of situating the practitioner in context and linking micro and macro-level activity, the existence of a pedagogically digestible strategy-themed theory to guide the study in the form of Whittington’s (2006) 3Ps framework, and its suitability to ‘address organisational phenomena that are posited to be relational, dynamic, and emergent’ (Orlikowski, 2010: 27).

The topic of quality in IS research has been a feature of the extant literature (Conboy et al., 2012; Galliers and Huang, 2012; Klein and Myers, 1999). It has also been a feature of the more generalised prescriptions on the conduct of casework and the methodology’s tools for establishing validity (Stake, 1995; Yin, 2009; Zivkovic, 2012). Empirical research conducted within the social sciences realm has a recognised set of parameters by which quality can be established (Yin, 2009). Each of these four tests has specified tactics by which researchers can be guided for case-based studies. The tests are as follows:

(Adapted from Yin, 2009)

For the current study, the three applicable tests are construct validity, external validity, and reliability. Internal validity is primarily a concern for the case study researcher who wishes to explain how and why x leads to y, i.e. a causal case, which is not a feature of the employed research approach. To view how the researchers established quality, Table 5 provides an overview of the measures taken to ensure that each test was sufficiently satisfied.

Table 5

Ensuring quality: measures taken to establish validity (adapted from Yin, 2009)