• Nenhum resultado encontrado

The data analysis was done concurrently with the data collection. The analy- sis consisted of within- and cross-case analyses, literature reflection, and conclusion drafting. Like the data collection, the data analysis followed the tenets of the Gioia methodology for creating a grounded theory (Gioia et al., 2013). The goal of the analysis was to give as much voice as possible to the leaders’ understanding of strategic renewal. The purpose was to become in- timately familiar with each case data and allow each one’s unique patterns to emerge before generalising these patterns across cases (Eisenhardt, 1989).

Thus, the leaders’ voice was the primary source of understanding for the 1st and 2nd order concepts emerging from the data. Relevant theory was only in- vestigated after these initial concepts were drawn from the data, to allow as much room as possible for novel findings to emerge. Theory was specifically used to give more insight into the initial definitions of the 1st order concepts and to elevate the 2nd order concepts into theoretically relevant 2nd order themes. Finally, these 2nd order themes were grouped together and further compared to existing theory to create the aggregate dimensions. These three

46

layers of 1st order concepts, 2nd order themes, and aggregate dimensions cre- ated a theoretically salient data structure. Further analysis of the connections between these three layers emerging from the cases helped transform the data structure into a process model portraying the interconnected enabling factors of successful strategic renewal.

The within-case analysis began by meticulously quoting and coding line-by- line each of the transcribed interviews using the Atlas.ti software. The goal here was to create the initial 1st order concepts with as much voice as possible given to the leaders’ thoughts. The guideline was that the leaders’ unique thoughts needed to be present even in the final data structure (see chapter 4 and Figure 5 for the data structure). This approach to coding resulted, as is typical in the grounded theory research (Gioia et al., 2013), in a dizzying number of initial 1st order concepts. These initial concepts were formulated into the final 1st order concepts with continuous iteration: First individually within-case, later together with the cross-case analysis, and finally by reflect- ing with the relevant theory. The approach of meticulously coding everything and then continuously iterating the concepts into the 1st order concepts helped to handle the high volume of data acquired from the interviews. Trust- ing the process was a necessity for this phase to reach completion.

Especially two analysis methods were used to create the 1st order concepts.

The first method was to plainly understand what the interviewees spoke of.

This method was used to create the initial and explicit concepts focusing on the different aspects of leadership philosophy, strategic renewal, and effects of education. The second method was to understand what the interviewees meant and how they spoke. This more interpretive and reconceptualising analysis (Alvesson, 2003) was used to understand how the interviewees im- plicitly understood the different focal aspects. This analysis shed light to top- ics that might have not been apparent at first glance. Thus, this analysis was complementary to the first method, as it was reasonable to interpret the in- terviewees only after their explicit meaning had been discovered.

47

The in-depth understanding of each individual case worked as the baseline for the cross-case comparison. In the cross-case analysis the aim was to dis- cover commonalities and relevant differences between individual codes to further refine the 1st order concepts and to generate the 2nd order themes. The data began to make sense as the research progressed, with evident similari- ties and differences starting to form initial categorisations. These emerging categories, known as the 2nd order concepts, were first given interviewee-cen- tric labels and then used to feed the interviews with questions aimed to more deeply understand the emerging concepts.

Three methods were utilised in the cross-case analysis. First method was to simply discover similar codes between the cases. These codes were grouped to form the initial 2nd order concepts, the basis for the eventual 2nd order themes. The second method was to identify differences between the cases.

This was done by finding within-group similarities together with intergroup differences from the cases (Eisenhardt, 1989). These differences were pri- marily grounded in the theoretical criterion of having both women and men, and engineers and non-engineers, present in the data sample. Thus, two cat- egory pairs were used to find the similarities and differences: women and men, and engineers and non-engineers. The third method was to discover possible similarities between the previously established category pairs. The reason for trying find similarities between different categories was to dis- cover novel findings and to counteract the typical tendency of reaching prem- ature or false conclusions due to biases in information processing (Eisen- hardt, 1989).

Eventually the 2nd order themes started to emerge from the cross-case anal- ysis. The focus was especially on the emerging concepts that could help de- scribe the phenomena targeted by this research, with special attention put towards novel concepts that seemingly had not been well represented in the existing literature (Gioia et al., 2013). With continuous iterations between theory and within- and cross-case analyses, the initial 2nd order concepts were morphed into a theoretically relevant set of 2nd order themes. These 2nd order themes were then investigated to look for commonalities between

48

them, with similar 2nd order themes being further grouped to form the aggre- gate dimensions. With the now generated 1st order concepts, 2nd order themes, and aggregate dimensions, the basis for building a comprehensive data structure was established (Gioia et al., 2013). Eventually this data would show the results of the analysis in an easily understandable form.

The next part was shaping the hypotheses based on the data analysis (Eisen- hardt, 1989). The goal was to compare the emergent themes with each case to ensure that the found themes truly fit the case data. Shaping the eventual hypotheses consisted of two parts. The first part was to continuously sharpen the themes by comparing them with the data. This part was especially im- portant due to the theory-building nature of the study because in theory- building research the constructs and their definitions emerge from the anal- ysis rather than being defined prior to the study (Eisenhardt, 1989). The sec- ond part of hypothesis formulation was to verify that the relationships be- tween the themes meticulously captured the true interconnections present in each case. This part ensured that there truly was causality between the gen- erated themes.

Existing theory was only majorly considered after the initial hypotheses were drawn, as premature intimate knowledge of the literature might have af- fected the perception of the data and led to hypothesis bias (Gioia et al., 2013). The hypotheses were compared with the current literature to find pos- sible gaps, similarities, or contradictions. The goal was to see whether the findings had been discovered previously or if the study had discovered novel concepts. A broad range of literature was considered to generate as theoreti- cally valid and generalisable constructs as possible. Supporting literature was accepted and conflicting studies embraced to provide the deepest possible insight into the topic. This acquired insight was then iteratively used to add new relevant cases to support the research with additional and complemen- tary perspectives. Importantly, this interplay between data and theory was used to discover the dynamic relationships existing in the data and to con- struct a dynamic grounded theory from the previously static data structure.

The goal was to make transparent the relational dynamics between the

49

concepts, themes, and dimensions existing in the data structure. This contin- uous interaction between existing theory and the novel findings eventually resulted in a novel grounded theory. After multiple iterations, this novel the- ory was deeply tied with both the research findings and the existing theory to provide an internally valid and generalisable theoretical process model on leading strategic renewal.

Reaching research closure was a result of achieving theoretical saturation.

Theoretical saturation is typically reached when the research has a sufficient amount of cases and when further iterations between data and theory provide only minimal incremental improvement (Eisenhardt, 1989). For this re- search a total of 13 cases were collected. Although the number of cases was approximately planned beforehand due to pragmatic consideration of time, the number proved to provide ample theoretical saturation.

The next chapter will reveal the findings from the iterative process of data collection and analysis.

50

4 Findings

This chapter presents the findings from the analysis. Interestingly, there were no major differences in the findings between engineers and leaders with a non-engineering background. However, few clear differences were discov- ered between women and men. Women shared all the same focuses in lead- ership philosophy and strategic renewal as men. However, women addition- ally emphasised the importance of diversity of thinking in their leadership philosophies and as an enabler of strategic renewal. Finally, trust as an ena- bler of renewal was only emphasised by women.

Figure 5 presents the findings of the study in a data structure model. The data structure shows all the relevant information from the data analysis: the in- formant-centric 1st order concepts, their groupings under the theoretically relevant 2nd order themes, as well as the aggregate dimensions tying the the- oretical structure together.

The findings presented in the data structure are divided into the following four chapters. The first chapter presents the typical leadership philosophy of a leader experienced in successful strategic renewal. The second chapter fo- cuses on the context of the strategic renewal. Here, the renewal philosophy, and the impulses, goals, and effects of successful strategic renewal are pre- sented. The third chapter moves the focus to the enablers and barriers of re- newal. In this chapter, the enabling and hindering factors are meticulously discussed. Importantly, this chapter presents an analysis-based process model of leading a successful strategic renewal. Finally, the fourth chapter discusses how the leaders perceived their education had prepared them for renewal. Here the focus is on understanding how education benefited the leaders’ way of renewing, and how future education could prepare students better for the challenges of strategic renewal.

51 Figure 5: Data structure

52