Author: Robin Ray

Analysis processes

The validity of the analysis process and the reliability of the findings were contingent on discussion with research colleagues and other people working with motor neurone disease. In this section, the use of the pronoun "we" is essential for effective, trustworthy data analysis. Prof Annette Street was an integral part of the "we" in this project. Her insight and ability to interpret the data were invaluable.

The interview data and field notes were loaded into NVivo, a soft ware package designed to manage qualitative data and enable line-by-line analysis. Ecomaps were linked to their corresponding transcripts using the databite function so that relationship detail in the transcript could be connected with the relevant ecomap. Transcripts were read several times to get a sense of their meaning, before codes were assigned to the relevant passages. Writing an explanation of each code was a laborious task and it was tempting to just go on with the analysis thinking that we would remember what we meant by each code. However, this step was invaluable because as the analysis continued we found that our understandings of descriptors changed. At times we needed to go back and revisit an explanation and look at where we had previously applied that code. This process lead to one of two outcomes: either the name of the code had to be modified and previous work reanalysed, or a new code was created for the new material. We applied the words and phrases used by the participants to label each code. As the list of codes grew, it became increasingly difficult to keep track of each one, so we began grouping them to themes, this time using concepts to label each theme.

Our next problem was how to get the ecomap data into a form so that comparisons could be made and the patterns of social change identified. We counted the number of relationship lines in each ecomap, then drawing on interview data, we made statements about the strength and usefulness of relationships as they evolved in specific cases across the data collection period. These statements also contributed to theme development.

This is an extract from our themes list that relates specifically to ecomap data. Note the code occurring after the hyphen.

Network - considered our needs
Network - died or relocated
Network - drifted away
Network - emotional support
Network - fits in
Network - good intentions
Extended family - emotional support
Extended family - invited in
Extended family - not seen
Extended family - practical support

Tabulating the data from the ecomaps seemed like a good idea. Yet, trying to capture the strength of relationships across three time periods and eighteen case studies resulted in large unwieldy tables that had to be printed out on several sheets of paper and glued together. We went through several versions of tables and consulted a statistics expert before settling on a useable format. We found that dividing up the support networks (as depicted by the ecomaps) into health care workers, family, and community, enabled us to achieve tables that were easier to read while still capturing the relationship detail.

It became obvious that social networks declined over time, but we wanted to see if there was a correlation between network decline and stage of disease. As motor neurone disease is a degenerative illness we wondered if social networks declined as the person became more disabled and needed more care. We went back to the literature to search the trajectories of the various forms of the disease and used this information together with caregivers' stories about the type of care they were providing, to develop a care need indicator. This was added to the table to provide a point of comparison, recognising that it was not a tested tool.

Tables gave us information about the strength of relationships and numbers of persons in a network, but reduced our ability to actually visualise the changes in networks over time. Graphs enabled us to represent the stability or decline of a social network and to see which elements of the network were disengaging or were lost [Insert link to 08-Robin table and graphs.doc].

Data analysis requires many hours of reading, thinking and discussing to go beyond description and discover what the data is really saying. We began to notice that some themes occurred repeatedly across cases. NVivo software enabled us to gather the thematic data across cases and see what each caregiver was actually saying in relation to that theme.

It was clear that further tools were needed for this analysis and we decided that social theory would be helpful. Thus, themes that had already been identified in the analysis were examined using social theory lenses. Applying theories of social capital (Coleman, 1999; Cox & Caldwell, 2000; Putnam, 2000; Putnam, Leonardi, & Nanetti, 1993) enabled us to clarify the connectedness in family, community and health care relationships. Anthony Giddens (1984; Giddens, 1990; Giddens & Pierson, 1998) theories concerning changing social structure and interactions with expert systems provided insight into the society of the caregiver and the social dynamics that impact on support networks.

Constant comparative analysis of data through descriptive coding and conceptual ordering resulted in 21 major themes. These themes were conceptually ordered as the caregiver, their interactions with expert systems, their social relationships and caregiving. This process was helpful in identifying the central concepts arising from the study as presented in the following table.

Caregiver	Expert System	Social Network	Caregiving
Caregiver Relationships Control	Getting a diagnosis Information Formal care Government Financial issues	Network Extended family Community Reciprocity Asking for help	Person living with MND Care role Symptoms Quality of life Future

Conceptual theorising about the data increased the validity of our findings and situated them in the social context being examined by this study.

Back to Project Home Page