2 Theoretical Framework
3.6 Analysis Strategies
The volume of data generated by processual research in particular, led Pettigrew (1990:281) to describe the obstacle faced by processual researchers as “…death by data asphyxiation”. The qualitative data software NVivio 9 was used by the researcher to facilitate data organisation and analysis. The researcher in conducting a thematic analysis drew on Ryan and Bernard’s (2003) in depth review of techniques to aid the identification and treatment of themes from the rich interview data. The study used the constructs identified within Dawson’s (2003a) framework to
construct the narrative case study centred on three main categories associated with the context, substance and politics of change. Each of these categories contained sub-categories and interview data expressing an example of the same was sorted into these. Ryan and Bernard (2003) assert that social scientists use different terms in relation to themes which relate to discrete concepts but that categories, codes and labels are interchangeable. Following construction of the case narrative based on the processual/contextualist approach, a further analysis was conducted to identify expressions of policy as a process as identified within the theoretical framework under the following themes, incoherence in policy implementation, unanticipated consequences to purposive actions, contextually contingent outcomes and fluid implementation structures (Trowler, 2002 , Trowler, 2008 , Trowler and Knight, 2002).
The volume of data may constitute one challenge but a number of other issues faced the researcher in the analysis and manipulation of such data. Langley (1999:692) describes these concerns as the complexity of data. The fluid nature of data and the effort to situate data within context and usually multiple contexts presents a further complication (Pettigrew, 1992). The temporality associated with events may be oblique to a researcher attempting to document a sequence of events (Langley, 1999:692-693). Within this study participants identified contexts which spanned many years before AFI implementation one such was identified which ranged over a twenty year period prior to the researcher’s employment with the university. The researcher had to engage with participants to uncover their constructions of these contexts, and to locate documentation from this period whilst attempting to develop her own construction. Collecting documents from this period involved contacting former employees of the university and the accessing of many personal archives.
Langley (1999) uses Weick’s (1979 , 1995) principle of sensemaking to develop seven strategies to analyse, manipulate and present process data. Sensemaking for Langley (1999:694) implies that a “...variety of “senses” or theoretical understandings may legitimately emerge from the same data” – different theoretical interpretations may have varying strengths and weaknesses in interpreting the data, an issue that Langley considers important. Langley (ibid) also asserts that sensemaking allows the researcher to consider theory and data from varying perspectives either deductively or inductively, with perhaps numerous iterations between both. Within this study deductive categories were used to originally construct the case narrative and further themes were identified from the literature following this original a priori categorisation (Ryan and Bernard, 2003). Coffey and Atkinson (1996:158) support such an approach when they advise researchers that:
Theories are not added on only as a final gloss or justification…They are drawn on repeatedly as ideas are formulated, tried out, modified, rejected or polished.
Langley (1999) uses the categories of accuracy, generality and simplicity, based on the work of Thorngate (1976) and Weick (1979), to describe the theoretical content formed whilst implementing a range of analysis strategies, outlined in Table 11. Accuracy pertains to an analysis strategy that remains close to the data (Weick, 1979). The level of abstraction is thus limited. The consequence of an analysis with a high degree of accuracy is a lower degree of the second category generality i.e. that the abstraction can be applied to other cases or situations (Langley, 1999). The third category, simplicity refers to simply understood and constructed theories with good explanatory powers. These simple theories contrast with complex theories that may provide some modest degree of further explanation, but are neither readily understood nor simple in construction (ibid). Langley (1999) describes data in terms of depth and breadth in Table 11, with depth referring to the level of detail needed of a process and breadth linked to the number of cases needed.
Table 11 – Seven Strategies for Sense Making*
Strategy Key Anchor
Points
Exemplars Fit with Process Data Complexity Specific Data Needs Good Theory Dimensions Weick
(1979)
Form of Sensemaking
Narrative Time Pettigrew
(1985a)
Fits with ambiguous boundaries, variable temporal embeddedness and eclecticism
One or few very rich cases. Can be helped by comparison
High on accuracy. Lower on simplicity and generality Stories, meanings, mechanisms Quantification Events, Outcomes Van de Ven and Polley (1992)
Focuses on “events” and their characteristics. Eschews ambiguity
Needs many similar events for statistical analysis one or few cases is best
High simplicity, potentially high generality, modest accuracy, (abstraction from original data)
Patterns, mechanisms
Alternative Templates
Theories Collis (1991) Adaptable to various kinds of complexity. Different templates capture different elements
One case is enough, degrees of freedom come from multiple templates
Each theory can be simple and general. Together, they offer accuracy but simplicity and generality disappear with theory integration
Mechanisms
Grounded Theory Incidents (Units of text) Categories Gioia, Thomas, Clark & Chittipeddi (1994)
Adapts well to eclectic data and ambiguity. May miss broad high-level patterns
Needs detail on many similar incidences. Could be different processes or individual level analysis of one case
High on accuracy, moderate simplicity. May be difficult to go from substantive theory to more general level
Meanings, patterns
Visual Mapping Events, orderings
Langley & Truax (1994)
Deals well with time relationships, etc. less good for emotions and
interpretations
Needs several cases in moderate level of detail to begin generating patterns (5-10 or more)
Moderate levels of accuracy, simplicity and generality. Not necessarily good for detecting mechanisms
Patterns
Temporal Bracketing Strategy
Phases Doz (1996) Can deal with eclectic data, but needs clear temporal breakpoints to define phases
One or two detailed cases, if processes have several phases used for replication
Accuracy depends on adequacy of temporal decomposition. Moderate simplicity and generality
Mechanisms Synthetic Processes (e.g. decisions, change efforts) Eisenhardt (1989b)
Needs clear process boundaries to create measures. Compresses events into typical sequences
Needs enough cases (5+) to generate convincing relationship. Moderate level of detail needed for internal validity
Modest accuracy (but much better than questionnaire research). Can produce simple and moderately general theories
Prediction
* Note that the entries from this table are indicative only. There is obviously considerable variation amongst the research following each strategy
3.6.1 Analysis Strategy Selection
Based on Langley’s (ibid) categorisation two sense making strategies were employed as part of the construction of the case; narrative and visual mapping. A further thematic analysis was conducted by the researcher. Weick (1979) suggests that all research strategies contend with trade-offs associated with accuracy, generality and simplicity. Langley (1999) positions the process analysis strategies in relation to each category, Table 12.
Table 12 – Sense making Strategies and Accuracy, Simplicity and Generality
Strategy Accuracy Simplicity Generality
High Low Low
Narrative Grounded Theory Temporal Bracketing Visual Mapping Synthetic Strategy Quantification Computer Simulation
Low High High
The orderings in this table are approximate; there are variations among specific applications. In particular, while accuracy and simplicity are almost always in opposition to one another, the generality of emerging theories will depend on other factors, such as the degree and scope of replication and the source of conceptual ideas.
Source: Langley (1999:706) Langley’s (1999) positioning of strategies is an effort to demonstrate that process research can be undertaken using a variety or combination of these strategies depending on the research aims of the study; the data available; epistemological approach and the insight, creativity and imagination of the researcher to complete the research process (Weick, 1989). To aid with the selection and combination of research strategies, Langley (1999) provides the following sequentially related groupings of the sense making strategies, Table 13. Langley advises that these groupings provide a touchstone for the researcher engaging with process research and should not be viewed as a single solution for all research contexts (ibid).
Table 13 – Grouping of Sense making Strategies
Grouping Strategies Rationale
Grounding Grounded theory Alternative templates
Both strategies provide sources for concepts to be used in the context of other strategies – from data-driven categories of grounded theory to the theory driven constructs of alternative templates. The strategies represent the purist forms of inductive and deductive reasoning. They provide the basis to feed into other strategies such as narrative and quantification. Organizing Narrative
Visual Mapping
Both strategies provide a description of tools to systematically organise process data, usually the first step taken in sensemaking. They also can feed into other strategies and can form the basis for hypothesis and propositions.
Replicating Temporal Bracketing Quantification Synthesis
These strategies are considered replicating strategies as they allow for the decomposing or reduction of data for the replication of a
theoretical proposition. Most other strategies feed into them.
Source: Adapted from Langley (1999:707) These strategies and their use within the research are elaborated on in the following sections.
3.6.2 Narrative Strategy
Langley (1999) maintains that all studies make use of this strategy during process research at some point. The purpose of the narrative, however, varies depending on the researcher’s objectives. For some researchers, it serves as a first step in organising data in preparation for analysis (Eisenhardt, 1989a). This step was taken in this study to prepare a case narrative as the narrative strategy holds a deeper analytical context in implementation research (Pettigrew, 1990). For those pursuing a naturalistic or constructivist perspective, it is the richness, detail and contextualisation of the narrative, i.e. the rich presentation of different viewpoints within the narrative that supports the reader in judging the transferability of the analysis to other “situations” (Langley, 1999:695). An analysis centred on this strategy is expected to be high in accuracy (Weick, 1979). The approach does not necessarily provide simple or general theory. Langley (1999:697), however,
recognises the benefit of the approach for “…communicating the richness of the context to readers, most of us expect research to offer more explicit theoretical interpretations”. Langley (ibid) considers that a narrative strategy can provide more than just an authentic account and it can aid the reader to apply what he/she has learnt to other situations (Golden-Biddle and Locke, 1993). Langley (1999:697) advises the researcher to combine the narrative strategy with another/or other analytic strategies to lessen the difficulty of “…telling the story and selling the plot…”. The subsequent thematic analysis was implemented by the researcher in an effort to avoid this pitfall associated with the narrative strategy.
3.6.3 Visual Mapping Strategy
Narrative and grounded theory strategies typically provide analysis in an “extended text” format (Miles and Huberman, 1994:91). Miles and Huberman (ibid:91) argue for the integration of illustrative and matrix representations of data and analysis beyond pure text:
By display we mean a visual format that presents information systematically, so the user can draw valid conclusions and take needed action.
The displaying of data using illustrative or matrix forms allows for large quantities of data to be represented in reduced formats, provides a visual template for the development and refinement of theoretical constructs and represents data using various dimensions and contexts, e.g. temporality, precedents or antecedents and parallel processes (Langley, 1999:700). The use of such visual mapping strategies is described as an “intermediary step” (ibid). Miles and Huberman (1994) describe the process of developing and using visual displays as a natural step in analytic progression, as described in Carney’s (1990) ladder of analytical abstraction. This strategy increases flexibility in data reduction and synthesis, providing for moderate accuracy (Langley, 1999). Moderate generality and simplicity are features of this
strategy and the strategy can be a useful building block in the construction of theory (ibid). This strategy was used particularly in the development of the case and was a useful strategy to depict the development of issues and events associated with the research.