Approaching Strategy as Practice: epistemological choices and research strategies

As we mentioned above, the set of papers included in this volume all rely heavily on qualitative data. Yet despite this common overall orientation, they adopt a range of different epistemological positions and research strategies. We shall draw on these illustrations to examine the main choices available and some of their implications. Eisenhardt’s (1989a) paper on ‘Making fast decisions in high velocity environments’ is an example of a study that takes what Guba and Lincoln (1994) would call a post-positivist position. It uses a comparative case study approach (Eisenhardt 1989b; Yin 2003) to develop middle-range theory relating 53 Doing research on doing strategy decision speed to a series of decision strategies used by top management teams.

Further more Information, Refer to What Is Strategy?

The assumptions are that a unique reality exists and that the data collected (mainly through in-depth interviews) provide access to this reality. The theory developed takes the form of a set of causal relationships whose generalizability can be tested in further studies. The author distils the ‘facts’ about each case from multiple interview accounts by verifying different sources against one another and classifying responses into carefully constructed categories. She is then able to examine the correlations and chains of causality among these categories. In contrast, the paper by Gioia and Chittipeddi (1991) adopts an interpretivist position and an ethnographic method to study strategy formation in a single organization over time. The authors were interested in how cognitive understandings of organizational participants evolved during the initiation of strategic change. Here, the assumption is that reality is socially constructed and the analysis focuses on the meanings attributed by organization members to ongoing events. The data collected are seen as representations rather than facts, and the researchers argue for the need for deep involvement in the cultural setting in order to reflect these representations fairly.

Beyond the classic epistemological distinction between positivism and interpretivism, the work of Gioia and Chittipeddi (1991) as well as that of Barley (1986) and Balogun and Johnson (2004) can be distinguished from that of Eisenhardt (1989a) on another important dimension: the attention given to tracing temporal dynamics or to understanding phenomena in process rather than variance terms (Mohr 1982; Langley 1999). This is another key choice for the researcher that has consequences for research design, data collection and theorizing. While variance theories explain phenomena such as strategic change in terms of causal variables, process theories explain them in terms of the activities, events and choices that constitute them over time. Both types of theories can contribute to improving understanding of Strategy as Practice. However, many of the theoretical resources introduced in the previous Issues tend to favour process theorizing.3 Another set of illustrative articles differ from those mentioned above in the way they pay particular attention to language. In these cases, the data collected 3 The reader should be aware that the term ‘process’ is used here to distinguish a specific type of theoretical form (involving sequences of events over time and temporal dynamics rather than relationships between variables (see Mohr 1982; van de Ven 1992).

This is different from the usage in earlier Articles where ‘strategy process’ is used to distinguish a specific research tradition interested in strategic management processes rather than ‘strategy content’. through interviews, recorded observations and documents are treated by the authors not as approximations to truth (as in Eisenhardt’s (1989a) study) nor even as pure interpretations of what is or was occurring (Gioia and Chittipeddi 1991), but rather as ‘discourse’ – talk or text that is consciously or unconsciously orientated towards an audience and serves a particular purpose. The task of the researcher in this case is to reveal the forces and purposes underlying this discourse and to examine its effects. For example, Samra-Fredericks’ (2003) study draws on conversation analysis embedded in an organizational ethnography to examine the rhetorical devices used by strategists to convince their colleagues. Oakes et al. (1998) also focus on the role of language, using documents and interviews to show how the introduction of the language of business planning subtly redefined who and what was important among the Alberta museum community (favouring a view of museums as serving customers in contrast to the traditional role of preserving artefacts). However, unlike most of the other papers presented in this volume, they explicitly adopt a critical paradigm in the sense that they attempt to unmask the latent power dynamics underlying what may on the surface be seen as normal, ‘rational’ and taken for granted. Finally, another orientation towards research on strategy practice is ‘action research’, in which researchers intervene (sometimes as consultants) and attempt to learn systematically from the results of their own interventions. Studies by Hodgkinson and Wright (2002), Greiner and Bhambri (1989) and Bürgi, Jacobs and Roos (2005) (the last included in this volume) are examples of this type of work. Here, the qualitative data are often a by-product of an intervention or a strategic consultancy process. This approach has the advantage of providing closer access to strategy practice than is normally possible. However, the credibility of the conclusions may be problematic, given the close proximity between the observer and what is observed. The various research orientations described above (post-positivist versus interpretivist versus critical; discourse analysis; action research; process versus variance theorizing) suggest a range of different ontological and epistemological choices that may drive research on Strategy as Practice.

Clearly, a first decision that a researcher will need to make concerns where to position him or herself among these possibilities. This choice might be based on natural orientation or on a deliberate selection. But in any case, this choice will be related to the types of research questions that are being asked and will in turn determine the research designs, data sources and analysis techniques that are appropriate. The meanings attributable to the data (fact, representation or discourse?) and the types of conclusions that will be drawn will also depend on these choices. The key here is to achieve consistency between assumptions and methods.

For example, those interested in developing theories that relate particular strategic practices to outcomes in a positivist tradition (like Eisenhardt (1989a)) will tend to prefer a comparative case study approach and a method of summarizing data that allows the development of well-defined constructs that can be tabulated and compared. They will see a need for a reasonable sample size of cases (from four to ten is ideal according to Eisenhardt (1989b)) and will therefore tend to favour breadth over depth in qualitative data collection and analysis. In contrast, those interested in participants’ interpretations, like Gioia and Chittipeddi (1991), will tend to prefer ethnography or in-depth interviewing as a research strategy and will be looking for depth, detail and nuance rather than convergence on well-defined constructs (Dyer and Wilkins 1991).

In turn, researchers interested in the development of process theories that capture temporal dynamics will tend to see longitudinal data collection as essential to capture the evolution of events over time. They may be more suspicious of retrospective reports than variance theorists (because of the potential distortion of time sequences) and will favour ethnography, historical analysis or real-time interviewing. However, depending on the focus of their theorizing (longer-term change processes or micro-processes that are more temporally bounded), they may be more or less concerned with fine-grained detail. Those interested in the role of language in strategic practice will have different needs again. They will generally require access to naturally occurring talk and text. Here, the precise words people use will be particularly important. Thus, if documents are the object of analysis, there may be a need to collect not only final versions of reports but also their preceding drafts and other related documents. If people’s utterances are the focus, electronic recording devices become essential. Because most discourse analysis methods are extremely time consuming, a high degree of selectivity is generally required. The trick is to pick precisely the pieces of data that offer the most useful insights. Another key issue that researchers need to consider before proceeding to develop more detailed research designs is the degree of theoretical and conceptual framing needed before entering the field. Interestingly, many if not most of the illustrative papers in this volume appear to have a large inductive component. None is a hypothesis-testing piece in the traditional positivist sense. This does not mean that they are atheoretical (most have a strong theoretical orientation) or that there were not some concepts and ideas driving initial data collection. Some authors (e.g. Eisenhardt 1989a; Bürgi et al. 2005) clearly discovered their theoretical angles and particular research questions only after data collection was underway (for example see statement by Eisenhardt (1989b)). Others had clear research questions from the beginning (Langley 1989; Gioia and Chittipeddi 1991) but these were open-ended and could have led to a variety of responses and theoretical contributions.

Finally, others began with sets of ‘sensitizing concepts’ derived from a general broad meta-theoretical scheme (for example symbolic interactionism for Barley (1986, 1990), and Bourdieu’s (1990) theory of fields of production for Oakes et al. (1998)). To some extent, this rather inductive orientation reflects the state of knowledge in the field. It also perhaps reflects our own bias for work that generates new insights in a field where quantitative survey research seems limited in its capacity to tap into the detailed micro-processes we are describing. It is true also that, whatever the starting point for a study, most qualitative research generates unexpected discoveries. Qualitative data have a multivocal character: they may speak to several different issues and may even take on different meanings when different epistemological assumptions are applied (see Alvesson and Sveningsson 2003). The creative researcher will be open to exploiting unexpected opportunities for learning as they occur. That said, we recommend all researchers to enter the field with clear research questions and some conceptual handles with which to bound the focus of their study and to structure their data collection and analysis. (If a researcher is short of ideas for good research questions, strategy practitioners can undoubtedly contribute useful suggestions.) Some initial exploratory work may be desirable to get a feel for the feasibility of a full-scale project, but good qualitative studies have a logical design at their end, if not at their beginning. Most of us cannot afford to spend too much time in random exploration before converging towards that design. We address more explicitly the issue of design in the next section.

Bounding Strategy as Practice: sampling and research design

One of the thorny issues that immediately confronts any researcher interested in strategy practice is the identification of appropriate units of analysis, the definition of their boundaries, and the nature of the corresponding ‘sample’ to be studied. We consider these issues in this section.

Choosing units of analysis

The unit of analysis refers to the precise object of the research – the entity about which one is trying to draw conclusions (Patton 2002; Yin 2003). To illustrate the dilemma associated with this choice, consider what is and is not included in the idea of strategizing. How would one draw boundaries around an incident of strategizing? How would one construct a sample of such incidents? The concept itself is open-ended, and the more so if we adopt a perspective in which not only top managers but also other organizational members are seen to be contributing to the strategizing effort – a contention that has attracted several researchers to this approach (Westley 1990; Balogun and Johnson 2004; Rouleau 2005). To the extent that one adopts a definition of strategizing as including any activity that might contribute to the orientation of the organization (and not just the formal and explicit strategic planning process), the concept further spreads itself out over space and time (Pettigrew 1990) to the point where almost any organizational activity can be considered part of strategizing. The resource-based view of the firm indeed locates competitive advantage in organizational routines and dynamic capabilities that are both deeply embedded and widely distributed across the firm.

Further more Information, Refer to competitive advantage

Yet, to paraphrase Wildavsky (1973), if strategizing is everything, then maybe it’s nothing. Most researchers will want to avoid this extreme ambiguity. The emphasis on activities suggests a need to define units of analysis in micro terms. But how micro is micro enough? The focal unit of analysis could be very narrow (e.g. individual strategy retreats or workshops, individual managers, meetings, conversations, etc.) or it could be broader (strategic decisions, strategic issues). Among the illustrative papers in this volume, we actually see a variety of strategies. For example, at the broadest level, Eisenhardt (1989a) focuses on a set of strategic decisions, Gioia and Chittipeddi (1991) focus on an episode of strategic change initiation, Oakes et al. (1998) focus on a broad planning philosophy among a set of organizations and its implementation amongst them. At the narrowest and most micro level, Samra-Fredericks (2003) focuses on small strips of conversation within the broader setting of meetings among strategists. Clearly, there are myriad possible choices, related to the research questions being tackled.

Drawing on Luhmann’s (1995) social systems theory, Hendry and Seidl (2003) introduce the notion of a ‘strategic episode’ as a particularly attractive unit of analysis for research on Strategy as Practice. An ‘episode’ in Luhmann’s theory is a ‘sequence of events structured in terms of a beginning and an 58 Strategy as Practice ending’ between which ‘the normal constraints of communicative practice are suspended and alternative communicative practices are explored’ (Hendry and Seidl 2003: 180). The idea put forward by Hendry and Seidl is that most organizational activity is structured around operational routines that are reproduced in patterned ways. Over time, random variations may occur in this stream of activity as patterns are imperfectly reproduced. However, unless the structures of communication change, these shifts will not necessarily be ‘goal directed’ or have any strategic significance. The creation of ‘strategic episodes’, in the form of strategy reviews, retreats or other kinds of meetings, etc., can be seen as the setting aside of certain specific times and places within which different rules apply and in which the structure of communication may be changed in such a way as to permit reflexive thought about the operational organizational routines that lie outside the episode. From this viewpoint, it is within these episodes that ‘strategizing’ is really taking place, because it is here that there is potential for altering the strategic trajectory. Hendry and Seidl (2003) propose taking these strategic episodes as a unit of analysis. Luhmann’s (1995) theory provides a number of indications for the types of features of these episodes that need to be examined if their role in the practice of strategy is to be understood. For example, the way in which the episode is initiated so that it is decoupled from operational routines that lie outside it may be important, as will the way the findings, reflexions and practices developed during the episode will be recoupled to those routines afterwards. Finally, the degree to which space is allowed to develop a new and different communication structure within the episode may determine its usefulness. The notion of strategic episodes as developed by Hendry and Seidl (2003) is thus an intriguing theoretically grounded way of defining units of analysis for the study of Strategy as Practice. Clearly there is room for the development of other ways of thinking about units of analysis for strategy practice research.

Defining and bounding units of analysis

Even when a clear logic for the choice of units of analysis exists, operationally defining them in an empirical study may still be challenging. Langley’s (1986) doctoral thesis (from which the paper reproduced in this volume was developed) illustrates this problem and the tactics that may be used to handle it. Langley was interested in the role of formal analysis in strategic decision making and confronted an initial difficulty of defining the two main units of analysis: what was actually meant by ‘an analysis’ and by a ‘strategic decision’? How would these units be recognized?

As described in Langley (1989), formal analysis was eventually defined as ‘a written document reporting the results of a systematic study of an issue’. This sounds simple on first reading but, in tracking down her sample, the author was forced to develop some ways to deal with the ambiguity surrounding its meaning while capturing what was most important (Langley 1986). The first tactic involved a fairly long and explicit list of exclusion criteria. These were used to avoid consideration of descriptive reports, multiple drafts of the same document and so on. This might be called an ‘isolation tactic’. This can only work to the extent that it is truly possible to develop objective criteria that allow such radical separation. The second tactic was to admit variation in the nature of the sample chosen and to capture it explicitly within the research design itself. Thus, all the documents collected were coded according to a series of criteria that measured their degree of analytical content. This became a control variable in the study. Bounding what may be considered to be a ‘strategic decision’ was equally challenging. Decisions in real life are not so easy to pin down – they are made and unmade sometimes without any visible traces, they string themselves together and generate sub-decisions on similar and related issues (Langley et al. 1995).

Langley eventually decided to change the unit of analysis from ‘strategic decision’ to ‘strategic issue’ because it seemed that organizational agendas were structured much more strongly around issues than decisions. Even here, however, it appeared that many strategic issues were interrelated in the organizations studied. To get around this ambiguity, the second tactic mentioned above was reused: variation was recognized and an explicit portrait of the extent to which different issues were related was included in the report (Langley 1986).

In summary, even once the main unit of analysis has been chosen, bounding it sufficiently to enable systematic empirical research is not always simple. Tactics for dealing with this may vary from reductionism (isolation tactics characterized by a focus on narrowly defined objects that exclude ambiguities) to the deliberate maintenance of ambiguity. Those adopting a positivist perspective will tend to prefer the first tactic while those adopting an ethnographic or interpretive perspective may prefer the latter tactic believing that, as Van Maanen (1995: 139) indicated, ‘to be determinate, we must be indeterminate’: the research itself must reflect the ambiguity present in the empirical situation, even including the ambiguity in its object. A middle-range tactic, and we believe a useful one, is the one emphasized by Langley (1986): explicitly admitting variation and mobilizing it within the research design.

Resource: Strategy as Practice Research Directions and Resources in Sharif Strategy