PART 1: BACKGROUND TO THE STUDY
1 The Challenge of Infertility in Global Reproductive and Sexual Health
4.3 Research strategy: qualitative component
4.3.10 Qualitative data analysis
This section describes how qualitative data were transformed into results chapters. The aim was to conduct a thorough, systematic and transparent analysis, to avoid drawing anecdotal conclusions and focusing on colourful cases to the detriment of those which do not fit in with existing preferences. Two approaches to qualitative analysis were taken, according to the data source. In-depth life history interviews were subject to a case analysis, looking at the data from a narrative perspective (a ‘narrative case analysis’), in which women’s life history narratives were compared to pre-existing and emergent hypotheses concerning the effects of infertility on their lives. The remaining qualitative data were analysed thematically, allowing for emergent hypotheses.
Narrative case analysis meant that each woman’s life history was treated as a narrative, and examined as an individual case (as opposed to aggregating data from many women). Analysing the data case by case involved testing the ‘goodness of fit’ of each woman’s life history to hypotheses laid out in the conceptual framework. This aspect of analysis was largely concerned with the content of interviews: the social events and processes described. The aim was to help to explain observed quantitative associations, contribute to understanding of what factors might make women fit these hypotheses whilst others did not, and suggest new lines of enquiry. When cases did not fit hypotheses, this suggested that the hypotheses needed to be re-examined, refined or extended in some way (Eisenhardt 1989).
Yet interviews were not just the sum of their contents, and were treated as ‘narratives’, which involved examining the manner in which stories were recounted, as well as their content. Narrative analyses ‘work through examining the nature and sources of the ‘frames of explanation’ used by the interviewee’ (Silverman 2000; 157). These frames
of explanation can refer to local belief systems, gendered structures, or moral values. Several lines of enquiry in narrative analysis were particularly appropriate to this study’s research questions (Reissman 1993). They involve posing the following sorts of questions of the data:
- How do narratives persuade the interviewer of their authenticity and identities? - What linguistic and cultural resources (‘frames of explanation’) does the
interviewee draw upon?
- Why did the interviewee tell a story in a particular way?
- What dynamics are witnessed in the interaction between interviewer and interviewee, and what do they tell us?
By analysing these aspects of narratives, the social and moral structures that influenced women’s lives emerged. The way that narratives are constructed can indirectly provide insight into wider moral and political dimensions, which is useful for situating data in its broader structural context (Rice and Ezzy 1999).
The process of narrative case analysis took place as follows:
1. Analysis during fieldwork. Each interview was read and a personalised interview schedule developed for a follow-up interview, if necessary. Emergent themes, confusing findings and knowledge gaps were noted.
2. Within-case analysis. A detailed, descriptive case study was written for each woman, incorporating data from one or both interviews and field notes. This was central to the generation of insight, as it involved close and repeated examination of the data, and helped deal with the large volume of data. Illustrative sections of dialogue were incorporated for further analysis. Each summary was re-read and key themes identified, with the aim of becoming intimately familiar with each case. This allowed patterns to emerge from each case before holding cases up against hypotheses and pushing for generalisations. 3. Cross-case analysis. Each case was held up to see how well or poorly it stood in
relation to hypotheses. Cross-case comparisons were made. New hypotheses and constructs also emerged at this stage.
At the cross-case analysis stage, participants were retrospectively classified as having experienced infertility or not. As there was no firm local definition of the length of time without pregnancy or a live birth that constituted infertility, a working definition of infertility was developed to identify current or previously infertile women. If they had
failed to have a live birth after trying to get pregnant for over a year (following the end of post partum insusceptibility15 in the case of secondary infertility) they were classified as ‘infertile’ or ‘previously infertile’. This one-year period was an arbitrary but sensible cut off point that reflects the clinical definition of infertility, and which also roughly reflected local perceptions16. Comparisons were then made between the experiences of fertile and infertile or previously infertile women.
Each life history was given equal weight during analysis so that extreme examples were not given undue attention at the expense of more everyday cases. The main difficulty with this approach was representing less forthcoming women’s stories. Several interviews were largely comprised of short exchanges between interviewer and participant, which gradually built up a life story, and were not particularly amenable to quotation. Their stories were summarised into case studies just as other women’s were, which meant that they were not unfairly neglected just because their accounts were not as rich.
Focus group and other interview transcripts were summarised, and quotes and the principal investigator’s interpretations were organised into key analytical themes. The aim was to build constructs that could be used to describe the local social context and situate life history data. A ‘construct’ in this sense is what a researcher creates when they group aspects of a phenomenon together and attempt to write a coherent, useful description which helps to make sense of the data. Examples of constructs built up through this process included ‘becoming tired of treatment-seeking’ and ‘when women have had enough of marriage’.
The process of writing up life history and other qualitative data was the final, and one of the most important, stages of analysis. Structured by constructs identified during initial analysis, long and detailed chapters were written, which were full of ‘raw’ interview data. Generalisations were then drawn about the range and content of life history and other qualitative data. The next step was to build theory: generalisations that best explained the observed data. This included theorising about structural aspects of social life: the kinship system, gender norms, power relations, and the significance of infertility within this broader context.
15 The infecund period following pregnancy caused by amenorrhea, breast feeding or post-partum abstinence.