Artículo VIII.- Los jueces no pueden dejar de administrar justicia por defecto o deficiencia de la ley. En tales casos, deben aplicar los principios
2.3. Análisis Jurisprudencial
This research collected both quantitative and qualitative data, which were analysed separately. SPSS Version 13.0 for Windows was used to analyse the quantitative data.
SPSS is a comprehensive statistical programme and one of the most widely-used in the social sciences. It can be used to score and analyse quantitative data very quickly and in many different ways. It can also be used to carry out complicated statistical analysis (Bryman and Cramer, 2005).
Computer-assisted qualitative data analysis software (CAQDAS), NVivo 7 was used to assist analysis of the qualitative data. NVivo 7 was developed out of an earlier version of QSR NUDIST (Bryman, 2004). I used it in my qualitative data analysis for several reasons: a) it is the most-used computer software in qualitative data analysis (Richards, 2005), b) it made the coding and retrieving fast and efficient, c) it made qualitative data analysis more transparent (Bryman, 2004), d) it was relatively simple to use, e) it can import documents directly from a word processing
programme for coding on screen, and e) you can write memos within the software and link the memos with relevant data (Welsh, 2002).
Analysing quantitative data
The data from the 100 questionnaires were entered and analysed using SPSS to provide descriptive information about service users and their level of satisfaction. I
carried out univariate analyses first to display each variable’s frequency and central tendency and to produce frequency tables, bar charts, pie charts and histograms (Bryman, 2004). I also carried out bivariate analyses to display some of the non-causal relationships such as the correlations between age and using multiple services;
gender and using multiple services; living alone or with someone and using multiple services; Locality Teams and satisfaction levels; Locality Teams and waiting times;
professional groups and waiting times and professional groups and satisfaction levels.
Because the method of data collection for the survey was face-to-face interviews and interviews through the phone, test was done to see whether there was any difference in these two ways of data collection. The hypothesis that face-to-face interviewees had higher satisfaction scores than telephone interviewees was tested and result of the test showed that there was no difference in these two ways of data collection.
Analysing qualitative data
The qualitative data in this study included policy documents (n29), minutes (n76), informal interviews (n15), the answers to the open questions on the users’ survey (n69), staff interviews (n8) and user interviews (n27). There are many different approaches to analysing qualitative data (Welsh, 2002). Thematic analysis is widely used ‘for identifying , analysing and reporting patterns (themes) within data’ (Braun and Clarke, 2006:79). I took the thematic analysis approach because it can
‘potentially provide a rich and detailed, yet complex, account of data’ (Braun and Clarke, 2006:78). It was a useful foundational method for qualitative analysis. It was flexible, being independent of theory and epistemology, and could be freely applied to arrange theoretical and epistemological approaches. It could be fitted into the theory-led programmatic approach and constructionist position. From a
constructionist standpoint, the views and experiences of service users are socially constructed and reconstructed rather than inhering within individuals (Burr, 1995).
Thematic analysis within a constructionist framework tries to:
theorize the sociocultural contexts, and structural conditions, that enable the individual accounts that are provided.(Braun and Clarke, 2006:85)
In qualitative data analysis the emphasis is on how interviewees’ understandings or interpretations are understood or interpreted.
Qualitative data analysis using NVivo comprised eight steps (see Table 3.6, Steps of qualitative data analysis). There are some concerns about analysing qualitative data using CAQDAS. Too much emphasis on its coding and retrieving process can cause a ‘fragmentation of the textual materials’ leading to loss of the ‘narrative flow’
(Bryman, 2004:419) or decontextualisation of the data (Buston, 1997; Fielding and Lee, 1998). To avoid decontextualising the data or losing the narrative flow during the analysis, I related my uncoded and coded text to relevant socio-demographic and personal information, referred to as ‘face-sheet variables’ (Mangaberia 1995, cited in Bryman, 2004:420), such as staff job titles and service users’ completed
questionnaires. Because the sample for the Phase Three semi-structured interview was derived from the Phase Two users’ survey, each of the interviewees in Phase Three had already been surveyed and had completed a questionnaire. I attached these relevant socio-demographic and personal information and research notes to their interview transcriptions. I read this information first so the context of the data would be in my mind during the analysis. This not only prevented me from
decontextualising data or losing the narrative flow during the coding, but it also helped me to understand and interpret data in a broader context.
NVivo calls codes ‘nodes’. Nodes are coded related materials that represent themes,
ideas, people or places and are created by collecting and assigning the related
materials during the NVivo data analysis process. They can be refined or removed at any point. Nodes consist of distinguished free nodes and tree nodes. Free nodes are not part of a hierarchy; they are either an initial coding or a node that has no logical connection with other nodes. Tree nodes are organised into a hierarchical structure with a parent node containing multiple child nodes. The tree nodes helped in locating nodes quickly and did not necessarily have to represent axial coding or other
qualitative methods. The eight steps of my qualitative data analysis process are described below.
Table 3.7 Steps of qualitative data analysis
Steps Description of the process
Transcribe, read and re-read the data with relevant notes. Note down initial ideas.
3. Open coding Go through the entire document. Code interesting features of the data by creating nodes and apply nodes to relevant segments of text.
They were free nodes.
4. Search for themes
Collate nodes into the five themes of contextual conditions, mechanism, implementation process, intended gaols and achieved outcomes. Try to gather all data relevant to each of the three themes.
5. Review themes Check all the extracts coded at each theme, read through them.
Check the entire data in relation to themes.
6. Categorise themes
Analyse the nodes under each theme. Categorise and then subcategorise nodes under the main themes.
7. Generate theoretical ideas
Link coded extracts with memos and with the interrelationship between nodes within each theme, and between themes. Generate theoretical ideas across the data.
8. Check theoretical ideas
Check through the data and whether coded extracts are relevant to identified theoretical ideas.
Step one: Setting up NVivo project
After setting up my NVivo project I inputted all the data into the project and numbered each document, minute and interview.
Step two: Familiarising myself with the data
I collected all the data and transcribed half the verbal data myself. The other half was transcribed by others, but I checked their transcriptions against the original audio recordings and corrected them. The process of transcription and checking is ‘a key phase of data analysis within interpretative qualitative methodology’ (Bird,
2005:227). It provided me with some prior knowledge of the data. I started the process of data analysis by transcribing it and reading and re-reading the transcriptions. I read the transcription with relevant ‘face-sheet variables’
(Mangaberia 1995 cited in Bryman, 2004:420) and research notes in order to
familiarise myself with the data and gain an understanding of both data and context. I also noted down initial ideas.
Step Three: Open coding
I went through the entire document I was analysing, coding interesting features of the data by creating nodes and applying each new note or existing note to relevant
segments of text. These were free nodes. For example, I highlighted the following text: ‘They are in such a hurry. They want to get done and out.’ (U2), created a node for ‘hurry’, and then applied the hurry node to the highlighted area.
Step four: Searching for themes
Searching for themes also involved building a nodes tree. After analysing the five longest user interviews, I went through the nodes and coded data and collated the nodes into five themes (contextual conditions, mechanism, implementation process, intended gaols and achieved outcomes). Then I tried to gather all the data relevant to each of the five themes.
Step five: Reviewing themes
Having completed the open coding of all data with 1496 tree nodes and free nodes, I went back to check each node by gathering all the extracts coded at each node and reading through them. I checked themes and every node in relation to the coded extracts. I then recoded or uncoded some of the extracts and renamed, merged, moved or deleted some nodes accordingly.
Step six: Theme categorising
I analysed the nodes under each theme and linked nodes with relevant information, research notes and memos to identify any interrelationships between nodes and how they linked together, categorising them under the themes into subthemes. Initially I categorised the nodes under the theme of contextual conditions into seven
subthemes, namely national; county; agency; user; political; population; and social and environmental contexts. I also identified another sub-theme of contexts, ‘barriers to the integration’. It turned out that these categories of contexts were messy and failed to explain anything. After analysing the nodes and all the extracts coded at each node under the contextual conditions theme to identify any interrelationships between them, I then recategorised the nodes under the contextual conditions theme into three new subthemes, namely, background of the integration, enabling factors and obstacles to the integration. I adopted the concepts of change outcome,
maintenance or prevention outcome, and service process outcome from Qureshi et al’s (1998) outcome study during data analysis to categorise all the nodes under the outcome theme because I found that all the nodes under the outcome theme could be fitted into Qureshi et al.’s three main groups of outcome, or the three subthemes.
Then I subcategorised the nodes under the subthemes. For example, six nodes were subcategorised under the subtheme ‘enabling factors’. These were shared value and vision; a good level of trust between stakeholders; previous positive experiences of
integration; the structure of the integration programme and the model of service;
pooled budget; and culture of integration programme. In NVivo language, this process was to create the child nodes and grand-child nodes and build a nodes tree.
Step seven: Theoretical idea generating
I linked coded texts with memos and with the interrelationship between nodes within each theme and between the themes, pulling out theoretical ideas across the data.
Step eight: Theoretical idea checking
I went through the data and checked whether coded extracts were relevant to the theoretical ideas identified to validate the research results (Welsh, 2002).