• No se han encontrado resultados

This research followed Braun and Clarke’s (2006) guidelines for carrying out thematic analysis. Merriam (1998) stressed the importance of undertaking analysis concurrent with the data collection process. After revising and refining transcripts, the author

started organising field notes and preparing visual materials (i.e., images, maps, and tabular and statistical data). Having done that, the researcher became more familiar with the range of material types at hand (Braun and Clarke, 2006). However, the research analysis in this study started in earnest after the author returned from Jordan in September 2015. It began with the reviewing, reading and rereading of the interview and focus group transcripts (Cope, 2003, cited in Greenaway, 2011, p.136; Platt, 1976, cited in Percy-Smith, 1999, p.205).

Richards (2009) considered that disaggregating the mass of qualitative data would be meaningful for systematically and rigorously reading and analysing data. Therefore, data organisation followed the adopted theoretical model. Data organisation helped the researcher to form preconceived ideas at the outset of the fieldwork rather than being confused about the large volume of data gathered.

Throughout the analysis process, memos were taken continuously on different-coloured stick-it notes (Roth and Bradbury, 2008, p.356; Malhotra and Birks, 2007, p.391). Every colour stood for a specific concept or theoretical theme by which to organise the data collection process. The use of coloured stick-it notes and markers is deemed useful for future pattern matching and then construction of concepts, particularly for those who conduct qualitative analysis manually. As a result, a plethora of views and concepts (sub-themes) emerged and were grouped together into the six conceptual dimensions of systemic institutional design provided by the theoretical model. These are, on one hand, social capital, intellectual capital and political capital in the global theme of soft infrastructure. On the other hand, in the global theme of hard infrastructure, are the legal system, the administrative system and the political system. It is worth mentioning that the structure of interview questions in the case study protocol in relation to the semi-structured interviews was organised and arranged in advance according to the theoretical model (see appendix 9-2).

Theoretical realm: The institutionalist and collaborative planning approach

Raw data sphere:

Systemic institutional design

Soft infrastructure Hard infrastructure

Cultural capital

Social

capital Intellectual capital Political capital system Legal Political system Administrative system Observing embedded theoretical pattern

Pattern matching Secondary data Documents: GAM publications, Ai publications, academic work, etc. Semi-structured interviews and focus group transcripts

Data organisation Legend Deductive Inductive Coding Searching/generating patterns 98

Figure 3-2: Structure of research analysis, source: the author

The key work of the researcher started with generating and applying initial codes to the data set. The initial data organisation resulted in large segments of data-rich information. Then these data segments were carefully studied by inspecting and interpreting them through a coding process (Richards, 2009, p.95). Coding is the process of categorising data under an idea, theme or category (Lewins and Silver, 2007, p.81, cited in Greenaway, 2011, p.136; Neuman, 2002, p.480), whereby subjective data generated from a qualitative approach become meaningful and comprehensible for readers (Richards, 2009, p.93).

Corbin and Strauss (2008) identified three types of theoretical coding: open coding, axial coding and selective coding. Open coding refers to describing phenomena in the form of concepts, which “allows the researcher to identify patterns within and between sources” (Corbin and Strauss, 2008, p.219) and to extract themes, topics or issues in a systematic manner.

At this stage, codes were preliminary and constructed from the literature review and raw data and the researcher’s understandings and memos. After the initial coding, tentative concepts were formed by merging and grouping those that were similar in meaning into larger units referred to as concepts. The naming of codes was decided on the basis of the review of the literature on the institutionalist and collaborative planning approach and other literature on development planning, governance, management and democracy.

Axial coding aims to refine theoretical concepts that emerge from raw data by finding connections and interrelations among different codes. It involves the move from codes to concepts through the selection of the most promising categories related to the research issue and the author’s perceptions. In this research, following the coding process, categories were further developed by searching for relationships, interconnections and generalities among codes. This step has been referred to as pattern matching (see figure 3-2, above). Pattern matching is the core procedure of testing theory to see how far the theoretical model (here, institutionalist analysis) is applicable or matches the context (here, Jordan). Pattern matching involves comparing the two patterns of each concept in both the theoretical realm and the emerging concepts from the raw data, then trying to find similarities or disparities. This transitional step has been interrogated and reviewed as a vital step in addressing research objectives and

related research questions. The last coding procedure is selective coding. It is done after the preliminary interrelating of codes for further refining of concepts in an iterative process. An example of the coding mechanism is presented in appendix 11.

The analysis in this research has been a process of conceptually stepping back and forth to review and retrieve data for further scrutiny. The qualitative research analysis process is not a distinct phase. Instead, it is an iterative process where the researcher moves back and forth between research design, primary data and analysis (Bryman and Burgess, 1994, p.217, cited in Percy-Smith, 1999, p.206).

Braun and Clarke (2006) noted how thematic analysis enables the researcher to identify the consistency, variety and divergence of raw data. Theoretical freedom enables the researcher to generate profound insights and meanings from respondents’ opinions. On this basis, thematic grouping (fragmenting and merging of collected data) helps to identify themes pertinent to research questions (Ryan and Bernard, 2003). Despite the many advantages it offers, thematic analysis has been criticised for following unclear ways of identifying themes and patterns (Braun and Clarke, 2006). A person-centred analysis approach would yield biased outcomes that diverge from the context of respondents’ accounts. To address these deficiencies, a full description of the data collection measures has been taken to ensure research reliability, and validity has been ensured through pattern matching in accordance with Yin (2009), as discussed in subsection 3.7.4, above, on internal validity.50

The research analysis in this study faced challenges in the form of many twists, dead ends and false starts. It is believed that not using qualitative analysis software means the process takes longer, particularly in correlating focus group data with semi- structured interview data, and with other secondary data in the pattern-matching stage. In contrast, a conventional method of coloured pens and stick-it notes was used for assigning codes and forming categories. The research community is sharply polarised in terms of whether or not to employ computer-assisted software in qualitative analysis (Crowley et al., 2002). In this regard, the author eschewed the use of computer-assisted analysis software for methodological reasons. The interpretivist approach requires the researcher to be closer to the data collected. Computer-assisted software (e.g. NVivo) would create distance between the researcher and the raw data collected. Regardless of

the extra time and effort spent, this gave the researcher the opportunity to be more engaged with and delve deeper into the raw data, and thus better understand the coding process and thus the concepts and emerging themes. Also, computer-assisted software is known for the large amount of time and high specification of computer properties required to master it (Zamawe, 2015; Welsh, 2002).