Analysis of findings was conducted throughout the data collection process, as emerging findings were reflected on for meaning and shared with participants for
their interpretation. These analyses were then used to prompt further data collection and/or changes in practice. In this sense, analysis was a joint activity between the participants, other researchers involved and myself.
After data collection had ended, a formal phase of analysis began which I mainly carried out, and it is this process that is described below. In this phase, I took regular opportunities to reflect on emerging findings with other researchers, some but not all of whom had been involved in data collection in this study. Also, written drafts of the findings were shared with the key participants, and their comments and queries were actively sought in workshops and meetings set up for this purpose. Under my direction, a second researcher assisted with the analysis of the patient profile data. This was a joint endeavour, but I had the main influence in the translation of raw data into findings. Because of this, the findings presented are therefore framed by my personal background, values and experiences.
4.8.1 Interview, focus group and field note data
Interview, focus group and field note data were analysed using the software package NUD*IST (Non-numerical Unstructured Data Indexing Searching and Theorizing) (V4.0). The process used within NUD*IST began with description and sorting of the data, and concluded with theorising. The stages were coding,
describing, summarising, interpreting, and writing. These stages are aimed at describing the broad process, although, in reality, the work was inter-linked and interdependent, and the stages are therefore not mutually exclusive. Each stage required judgement and decision-making on my part.
Each piece of data was read and reflected on, and assigned one or more codes. The codes were developed inductively from the research questions and the data but it is also likely that my own personal framework that had built up from my in-depth knowledge of the study setting influenced the emergence of the coding framework. The framework or ‘tree’ that emerged represented the relationship of the different codes to each other. Each code had an assigned ‘node’ in the tree. The final tree contained 129 nodes. In addition, 24 free nodes were assigned which did not directly link into the tree. The final list of nodes is shown in Appendix Four (p. 225). Appendix Four also includes an excerpt from the tree to illustrate how the nodes were linked to each other. The whole tree is too large to show. The links
between nodes can also be inferred by their numbering in the list shown in Appendix Four.
Where the label for the node was not sufficient description, nodes had a description attached to enable accurate coding. As new codes were developed, it was
necessary to review all the data previously categorised to see if any of it applied to the new codes. The emphasis in this stage was on descriptive codes, rather than interpretive ones.
Data within each node were then read and reflected on to ensure that each node contained data with a similar meaning. Some new nodes were developed at this stage.
Data within each node were then summarised so that the whole body of data became a more manageable size. Each piece of summarised data was given one or more ‘location codes’ so that it could be linked back to the original data source if this was required. During this summarising process, constants checks were made with the original data context to ensure meaning was not lost.
Summarised data in each node were then read, reflected on and re-categorised into wider themes. For example, a main theme was developed of ‘What IPCCs do’. This included the main activities of:
• Moving patients through the system quickly/relieving acute pressures • Helping/supporting interprofessional colleagues
• Knowing about patients and resources • Ensuring appropriate care
• Ensuring appropriate discharge • Being known
• Aiming towards discharge all the time
Each of these main activities had associated summarised data (with location codes) that provided in-depth detail of work in each of these activities. This activity of interpreting involved checking back to original data sources again to ensure that original meaning was preserved.
4.8.2 Documentary analysis
Relevant Trust policies and other documents were read and reflected on to illuminate the intentions behind setting up the IPCC role, and the existence of a formal structure to support IPCCs and wider interprofessional working.
4.8.3 Patient profile
Details were entered onto a File Maker Pro database (version 5.0). Microsoft Excel 2000 and Minitab (version 10) were used to perform a range of descriptive and comparative statistical calculations. Frequency and proportion were calculated for each variable. In addition, patient age and length of stay were described for each IPCC and for the IPCCs as a group using the following calculations: frequency, mean, median, mode, standard deviation and range. Pearson’s correlation was used to explore the relationship between age and length of stay. One-way analysis of variance was used to explore differences in length of stay between the IPCCs.
4.8.4 Crystallisation of data
As analysis progressed, similarities and inconsistencies in the data raised perspectives that could be alternatively viewed using other data sources. This process of identifying, analysing and reporting multiple perspectives is known as crystallisation, rather than triangulation. Triangulation assumes there is a fixed point or object that can be triangulated or, in other words, a ‘master reality’ to be discovered (Richardson 2000; Murphy et al. 1998, p. 11). It can discourage the researcher from analysing the data in context and uncovering the situated work (Murphy et al. 1998).
In contrast, crystallisation reflects the concept of multiple perspectives, the view of which is dependent on one’s angle of repose (Richardson 2000). Crystallisation provides us with a ‘deepened, complex, thoroughly partial, understanding of the topic’ (Richardson 2000, p. 934). It offers opportunities to compare contrasting findings such as differences between policy and practice. It does not deny that there is a real world out there, but recognises that attempts to measure and describe that world are influenced by a complex range of social and cultural factors. In crystallisation, the challenge is to reflect multiple perspectives, while recognising that one’s account can only ever be a partial understanding of the topic. Because of
this, I needed to be reflexive throughout the research process about my own values and intentions as a researcher.